The lack of XCPM support for my new Broadwell E processors was a bit frustrating and thus I had to do something about it. And I guess that most of you know me by now so it was just a matter of time. As in. Consider it done!
Yup. Time to say good-bye to the good old AppleIntelCPUPowerManagement.kext and the NullCPUPowerManagement.kext that served so many people for so long. Adieu my friends. They won’t be missed. LOL
Right. I can boot macOS Sierra and I see IOPPF: XCPM mode along with XCPM registered. That is pretty cool. In fact. This is amazing. Let’s go check some stuff.
sysctl -n machdep.xcpm.mode returns 1 so that is fine. Also. Running kextstat shows me that both AppleIntelCPUPowerManagement.kext and NullCPUPowerManagement.kext are missing in the list with loaded kexts. Great. Now a Geekbench v3.4.1 top score – one of four runs – of the Intel i7-6850K.
I must admit. The longer I use this processor the more I get impressed by it. Who would have thought that? And let’s not forget. This is with XCPM so it is the more impressive. In fact. People with an old i7-5930K may keep it.
One thing though and that is that the RAM can’t run at full speed (runs like 2133MHz due to some bug) but anyway. Here is the output of AppleIntelInfo.kext
AppleIntelInfo.kext v1.5 Copyright © 2012-2016 Pike R. Alpha. All rights reserved
Settings:
------------------------------------
logMSRs............................: 1
logIGPU............................: 0
logCStates.........................: 1
logIPGStyle........................: 1
InitialTSC.........................: 0x150cbc299c8
MWAIT C-States.....................: 8480
Model Specific Registers
-----------------------------------
MSR_CORE_THREAD_COUNT......(0x35) : 0x6000C
MSR_PLATFORM_INFO..........(0xCE) : 0x20080C3BF3812400
MSR_PMG_CST_CONFIG_CONTROL.(0xE2) : 0x8402
MSR_PMG_IO_CAPTURE_BASE....(0xE4) : 0x10414
IA32_MPERF.................(0xE7) : 0x387EBCA78F
IA32_APERF.................(0xE8) : 0x3A56B9B06F
MSR_0x150..................(0x150) : 0x0
MSR_FLEX_RATIO.............(0x194) : 0xE0000
MSR_IA32_PERF_STATUS.......(0x198) : 0x23CF00002500
MSR_IA32_PERF_CONTROL......(0x199) : 0x2400
IA32_CLOCK_MODULATION......(0x19A) : 0x0
IA32_THERM_STATUS..........(0x19C) : 0x88490000
IA32_MISC_ENABLES..........(0x1A0) : 0x840089
MSR_MISC_PWR_MGMT..........(0x1AA) : 0x402000
MSR_TURBO_RATIO_LIMIT......(0x1AD) : 0x2525252525252828
IA32_ENERGY_PERF_BIAS......(0x1B0) : 0x1
MSR_POWER_CTL..............(0x1FC) : 0x2904005B
MSR_RAPL_POWER_UNIT........(0x606) : 0xA0E03
MSR_PKG_POWER_LIMIT........(0x610) : 0x7FFF80015FFF8
MSR_PKG_ENERGY_STATUS......(0x611) : 0x90C48C3
MSR_PKG_POWER_INFO.........(0x614) : 0x1700460
MSR_PP0_POWER_LIMIT........(0x638) : 0x0
MSR_PP0_ENERGY_STATUS......(0x639) : 0x0
MSR_PKGC6_IRTL.............(0x60b) : 0x0
MSR_PKG_C2_RESIDENCY.......(0x60d) : 0xA4268B844
MSR_PKG_C6_RESIDENCY.......(0x3f9) : 0x79FDD64350
IA32_TSC_DEADLINE..........(0x6E0) : 0x150CD144AC8
CPU Ratio Info:
------------------------------------
CPU Low Frequency Mode.............: 1200 MHz
CPU Maximum non-Turbo Frequency....: 3600 MHz
CPU Maximum Turbo Frequency........: 4000 MHz
CPU P-States [ 33 37 (40) ]
CPU C6-Cores [ 0 2 5 7 9 10 ]
CPU P-States [ 31 33 37 (40) ]
CPU C6-Cores [ 0 2 3 4 5 7 9 10 ]
CPU P-States [ (12) 31 33 37 40 ]
CPU C6-Cores [ 0 1 2 3 4 5 6 7 9 10 11 ]
CPU P-States [ (12) 25 31 33 37 40 ]
CPU P-States [ 12 25 29 31 33 37 (40) ]
CPU P-States [ 12 25 29 30 31 33 37 (40) ]
CPU P-States [ 12 25 28 29 30 31 33 37 (40) ]
CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 ]
CPU P-States [ (12) 22 25 28 29 30 31 33 37 40 ]
CPU P-States [ (12) 21 22 25 28 29 30 31 33 37 40 ]
CPU P-States [ (12) 21 22 23 25 28 29 30 31 33 37 40 ]
CPU P-States [ (12) 20 21 22 23 25 28 29 30 31 33 37 40 ]
CPU P-States [ 12 20 21 22 23 25 28 29 30 31 32 33 37 (40) ]
CPU P-States [ 12 20 21 22 23 25 28 29 30 31 32 33 37 39 (40) ]
CPU P-States [ 12 20 21 22 23 25 27 28 29 30 31 32 33 37 39 (40) ]
CPU P-States [ (12) 17 20 21 22 23 25 27 28 29 30 31 32 33 37 39 40 ]
CPU P-States [ (12) 16 17 20 21 22 23 25 27 28 29 30 31 32 33 37 39 40 ]
CPU P-States [ (12) 16 17 18 20 21 22 23 25 27 28 29 30 31 32 33 37 39 40 ]
CPU P-States [ 12 16 17 18 19 20 21 22 23 25 27 28 29 30 31 32 33 37 39 (40) ]
CPU P-States [ 12 16 17 18 19 20 21 22 23 25 27 28 29 30 31 32 33 37 38 39 (40) ]
CPU P-States [ 12 16 17 18 19 20 21 22 23 25 26 27 28 29 30 31 32 33 37 38 39 (40) ]
CPU P-States [ 12 16 17 18 19 20 21 22 23 25 26 27 28 29 30 31 32 33 36 37 38 39 (40) ]
CPU P-States [ (12) 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 36 37 38 39 40 ]
CPU P-States [ 12 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 36 37 38 39 (40) ]
CPU P-States [ 12 14 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 36 37 38 39 (40) ]
CPU P-States [ (12) 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 36 37 38 39 40 ]
CPU P-States [ 12 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 (40) ]
CPU P-States [ (12) 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 ]
Hmm. C3 is missing. Seems like I forgot to enable C3 for Broadwell E processors and thus I have to fix this and re-compile the kext. On a second thought. The redirection bits are not set in MSR 0xE2 so this is expected result.
But both MSR_PKG_C2_RESIDENCY and MSR_PKG_C6_RESIDENCY are non-zero. A sweet spot.
Update: I don’t have a lot of time, but I also do not want to let you wait so here you have it!
// // kernel location 0x1fb451 (0xE9 in DP2/DP3/DP4, 0xF1 in DP1) _cpuid_set_info // // Broadwell E (0x4F) -> Broadwell (0x47) => 0x4F-0x47 = 0x08 -> change 0xE9 into 0xE1 // Haswell E (0x3F) -> Haswell (0x3C) => 0x3F-0x3C = 0x03 -> change 0xE9 into 0xE6 // Ivy Bridge E (0x3E) -> Ivy Bridge (0x3A) => 0x3E-0x3A = 0x04 -> change 0xE9 into 0xE5 ../.. // // kernel location 0x22a422 (0xC4 in DP2/DP3/DP4, 0x?? in DP1) _xcpm_bootstrap // // Broadwell E (0x4F) -> Broadwell (0x47) => 0x4F-0x47 = 0x08 -> change 0xC4 into 0xBC // Haswell E (0x3F) -> Haswell (0x3C) => 0x3F-0x3C = 0x03 -> change 0xC4 into 0xC1 // Ivy Bridge E (0x3E) -> Haswell (0x3C) => 0x3E-0x3C = 0x02 -> change 0xC4 into 0xC2 // // Broadwell E (0x4F) -> Haswell (0x3C) => 0x4F-0x3C = 0x13 -> change 0xC4 into 0xB1 ../.. // // If MSR(0xE2) is locked (bit-15 is set) then also change // // kernel location 0x220b30 and 0x220b5f in _xcpm_idle from 0x0f30 (wrmsr) to 0x9090 (nop nop). // // Additionally (if booting results in a immediate reboot) // // kernel location 0x22a820 change 0x55 into 0xC3 (ret) to stop the KP. //
Update: DP4 is now available and the byte patterns are still the same, but the locations have changed to: 0x1f8d81, 0x227d32, 0x21e440, 0x21e46f and 0x228130 (same order as above).
Update-2: DP5 is now available and the byte patterns are still the same, but the locations have changed to: 0x1f8930, 0x228af0, 0x21eff0, 0x21f01f and 0x228f50 (same order as above).
Update-3: DP6 is now available and the byte patterns are still the same, but the locations have changed to: 0x1f7cc0, 0x227f30, 0x21e430, 0x21e45f and 0x428390 (same order as above).
Update-4: DP7 is now available and the byte patterns are still the same, but the locations have changed to: 0x1f8d71, 0x227fc2, 0x21e460, 0x21e48f and 0x2283c0 (same order as above).
The patches can be done in at least three different ways and I picked what I think is the easiest one to understand. So I hope. Also. I don’t use FakeCPUID – nor Clover for that matter – with my patches, but some people using Clover reported a hang. For them it won’t boot without FakeCPUID (thanks to giacomoleopardo for the update).
Update: Owners of Ivy Bridge E or Haswell E processors do not need to patch the switch table used by _cpuid_set_info, because Apple already took care of it. Anyway. Have a look at this table:
The offset for the Broadwell E processor is missing in this table. There is an address but that jumps to a location for unsupported processors, which is why we need a patch for the Broadwell E processor (the red arrow shows you what we do with the patch) but I think that Apple will fix this with a future update of macOS Sierra so that we also no longer need to patch the table for Broadwell E processors.
Please note that the table was converted from the original one in the kernel, which looks like this:
0xffffff80003fbc6c dd 0xfffff82c, 0xfffff831, 0xfffff831, 0xfffff7fb 0xffffff80003fbc7c dd 0xfffff831, 0xfffff831, 0xfffff831, 0xfffff7fb 0xffffff80003fbc8c dd 0xfffff7fb, 0xfffff831, 0xfffff831, 0xfffff831 0xffffff80003fbc9c dd 0xfffff831, 0xfffff831, 0xfffff809, 0xfffff831 0xffffff80003fbcac dd 0xfffff831, 0xfffff831, 0xfffff831, 0xfffff810 0xffffff80003fbcbc dd 0xfffff831, 0xfffff809, 0xfffff810, 0xfffff7fb 0xffffff80003fbccc dd 0xfffff809, 0xfffff831, 0xfffff831, 0xfffff831 0xffffff80003fbcdc dd 0xfffff831, 0xfffff831, 0xfffff831, 0xfffff831 0xffffff80003fbcec dd 0xfffff831, 0xfffff831, 0xfffff831, 0xfffff817 0xffffff80003fbcfc dd 0xfffff831, 0xfffff802, 0xfffff81e, 0xfffff817 0xffffff80003fbd0c dd 0xfffff802, 0xfffff831, 0xfffff831, 0xfffff831 0xffffff80003fbd1c dd 0xfffff831, 0xfffff831, 0xfffff802, 0xfffff802 0xffffff80003fbd2c dd 0xfffff81e, 0xfffff831, 0xfffff831, 0xfffff831 0xffffff80003fbd3c dd 0xfffff831, 0xfffff831, 0xfffff831, 0xfffff825 0xffffff80003fbd4c dd 0xfffff831, 0xfffff831, 0xfffff831, 0xfffff831 0xffffff80003fbd5c dd 0xfffff831, 0xfffff831, 0xfffff831, 0xfffff831 0xffffff80003fbd6c dd 0xfffff831, 0xfffff831, 0xfffff831, 0xfffff831 0xffffff80003fbd7c dd 0xfffff831, 0xfffff831, 0xfffff831, 0xfffff825
Update: DP4 is now available and while the table is still the same, the location of it has changed to: 0xffffff80003f959c
Update-2: DP5 is now available and while the table is still the same, the location of it has changed to: 0xffffff8000428cbc
Update-3: DP7 is now available and while the table is still the same, the location of it has changed to: 0xffffff80003f958c
And here is what I did with the values. Let’s take the last value (0xfffff825 ) as an example:
0xffffffff - (0xfffff825 - 0x01) = 0x7DB
I then used the value as a negative offset, and we do that because the table is located at a higher memory address, so here we go:
0xffffff80003fbc6c - 0x7DB = 0xffffff80003FB491
Great. Now look at this disassembled code snippet:
0xffffff80003fb43b movzwl 0x6fcc42(%rip), %eax 0xffffff80003fb442 xorl %ebx, %ebx // Zero out %ebx 0xffffff80003fb444 movzbl %al, %ecx 0xffffff80003fb447 cmpl $0x6, %ecx // Check CPU family 0xffffff80003fb44a jne 0xffffff80003fb49d 0xffffff80003fb44c movzbl %ah, %eax 0xffffff80003fb44f addl $-0x17, %eax // Lower model number (example: 0x5e -> 0x4f) 0xffffff80003fb452 cmpl $0x47, %eax // Check for unsupported model numbers 0xffffff80003fb455 ja 0xffffff80003fb49d // Jump if greater than 0x47 (unsupported model numbers) 0xffffff80003fb457 leaq 0x80e(%rip), %rcx // Store address of case table 0xffffff80003fbc6c into %rcx 0xffffff80003fb45e movslq (%rcx,%rax,4), %rax // Move and sign-extend 32-bit value from switch table in %rax 0xffffff80003fb462 addq %rcx, %rax // Example: 0xffffff80003fbc6c - 0x7db = 0xffffff80003fb491 (CPUFAMILY_INTEL_SKYLAKE) 0xffffff80003fb465 jmpq *%rax // Jump to target address 0xffffff80003fb467 movl $0x6b5a4cd2, %ebx // CPUFAMILY_INTEL_NEHALEM 0xffffff80003fb46c jmp 0xffffff80003fb49d 0xffffff80003fb46e movl $0x10b282dc, %ebx // CPUFAMILY_INTEL_HASWELL 0xffffff80003fb473 jmp 0xffffff80003fb49d 0xffffff80003fb475 movl $0x573b5eec, %ebx // CPUFAMILY_INTEL_WESTMERE 0xffffff80003fb47a jmp 0xffffff80003fb49d 0xffffff80003fb47c movl $0x5490b78c, %ebx // CPUFAMILY_INTEL_SANDYBRIDGE 0xffffff80003fb481 jmp 0xffffff80003fb49d 0xffffff80003fb483 movl $0x1f65e835, %ebx // CPUFAMILY_INTEL_IVYBRIDGE 0xffffff80003fb488 jmp 0xffffff80003fb49d 0xffffff80003fb48a movl $0x582ed09c, %ebx // CPUFAMILY_INTEL_BROADWELL 0xffffff80003fb48f jmp 0xffffff80003fb49d 0xffffff80003fb491 movl $0x37fc219f, %ebx // CPUFAMILY_INTEL_SKYLAKE 0xffffff80003fb496 jmp 0xffffff80003fb49d 0xffffff80003fb498 movl $0x78ea4fbc, %ebx // CPUFAMILY_INTEL_PENRYN 0xffffff80003fb49d movl %ebx, 0x6fcd35(%rip) 0xffffff80003fb4a3 cmpl $0x0, 0x6fcb6e(%rip) 0xffffff80003fb4aa je 0xffffff80003fb4c3
Our calculated address is spot on. Now let’s have a look at the switch table that is used by _xcpm_bootstrap in macOS Sierra DP3:
0xffffff800042a58c dd 0xfffffeac, 0xfffffec7, 0xfffffeb8, 0xfffffeb8 0xffffff800042a59c dd 0xfffffeb8, 0xfffffeb8, 0xfffffeb8, 0xfffffeb8 0xffffff800042a5ac dd 0xfffffeb8, 0xfffffed3, 0xfffffee9, 0xffffff0a 0xffffff800042a5bc dd 0xfffffeb8, 0xfffffeb8, 0xfffffeb8, 0xfffffeb8 0xffffff800042a5cc dd 0xfffffeb8, 0xfffffeb8, 0xffffff16, 0xfffffeb8 0xffffff800042a5dc dd 0xfffffeb8, 0xfffffeb8, 0xfffffeb8, 0xfffffeb8 0xffffff800042a5ec dd 0xfffffeb8, 0xfffffeb8, 0xfffffeb8, 0xfffffeb8 0xffffff800042a5fc dd 0xfffffeb8, 0xfffffeb8, 0xfffffeb8, 0xfffffeb8 0xffffff800042a60c dd 0xfffffeb8, 0xfffffeb8, 0xffffff2c
Update: DP4 is now available and while the table is still the same, the location of it has changed to: 0xffffff8000427e9c
Update: DP7 is now available and while the table is still the same, the location of it has changed to: 0xffffff800042812c
A shorter table with 34 instead of 72 values, but the idea here is the same. And with the same kind of table conversion we get this:
This time the table starts at 0x3c instead of 0x17 and there are no addresses for the Ivy Bridge E, Haswell E and Broadwell E processors, again, for the same reason; it jumps to a location for processors that do not support XCPM.
Speaking about that address. All locations in the switch table in the kernel with the value 0xfffffeb8 jump to the same location for unsupported processors and here is how to calculate the offset and target address for the jmp instruction:
0xffffffff - (0xfffffeb8-0x01) = 0x148
0xffffff800042a58c - 0x148 = 0xffffff800042a444
And looking at the disassembled code snipped of _xcpm_bootstrap that is where XCPM gets disabled:
0xffffff800042a420 addl $-0x3c, %ebx // Lower model number (example: 0x5e -> 0x22(34) 0xffffff800042a423 cmpl $0x22, %ebx // Check for unsupported models 0xffffff800042a426 ja 0xffffff800042a444 // Jump if greater than 0x22 (unsupported model numbers) 0xffffff800042a428 leaq 0x15d(%rip), %rax // Store address of case table 0xffffff800042a58c into %rcx 0xffffff800042a42f movslq (%rax,%rbx,4), %rcx // Move and sign-extend 32-bit value from switch table in %rax 0xffffff800042a433 addq %rax, %rcx // Example: 0xffffff800042a58c - 0xd4 = 0xffffff80000x42a4b8 (Skylake) 0xffffff800042a436 jmpq *%rcx // Jump to target address 0xffffff800042a438 movl $0x4, _xcpm_cpu_model(%rip) // case 0: Haswell (0x3c-0x3c=0) 0xffffff800042a442 jmp 0xffffff800042a47f 0xffffff800042a444 movl $0x0, _xcpm_mode(%rip) // case 33: xcpm_mode = 0 (XCPM disabled) 0xffffff800042a44e jmp 0xffffff800042a580 0xffffff800042a453 movl $0x80, _xcpm_cpu_model(%rip) // case 1: Broadwell (0x3d-0x3c=1) 0xffffff800042a45d jmp 0xffffff800042a4ac 0xffffff800042a45f movl $0x10, _xcpm_cpu_model(%rip) // case 9: Haswell-ULT (0x45-0x3c=9) 0xffffff800042a469 movl $0x1, 0x602981(%rip) 0xffffff800042a473 jmp 0xffffff800042a47f 0xffffff800042a475 movl $0x8, _xcpm_cpu_model(%rip) // case 10: Crystalwell (0x46-0x3c=10) 0xffffff800042a47f movl $0x1, 0x602973(%rip) 0xffffff800042a489 movq $0x0, 0x6029dc(%rip) 0xffffff800042a494 jmp 0xffffff800042a50c 0xffffff800042a496 movl $0x40, _xcpm_cpu_model(%rip) // case 11: Broadwell-H (0x47-0x3c=11) 0xffffff800042a4a0 jmp 0xffffff800042a50c 0xffffff800042a4a2 movl $0x200, _xcpm_cpu_model(%rip) // case 18: Skylake (0x4e-0x3c=18) 0xffffff800042a4ac movl $0x1, 0x60293e(%rip) 0xffffff800042a4b6 jmp 0xffffff800042a50c 0xffffff800042a4b8 movq 0x6cebf1(%rip), %rdi // case 34: Skylake (0x5e-0x3c=34) 0xffffff800042a4bf testq %rdi, %rdi 0xffffff800042a4c2 jne 0xffffff800042a4d9 0xffffff800042a4c4 movq 0x6999b5(%rip), %rax 0xffffff800042a4cb movq 0x488(%rax), %rdi 0xffffff800042a4d2 movq %rdi, 0x6cebd7(%rip) 0xffffff800042a4d9 addq $0x2, %rdi 0xffffff800042a4dd movl $0x2, %esi 0xffffff800042a4e2 callq 0xffffff8000402580 0xffffff800042a4e7 movzwl %ax, %eax 0xffffff800042a4ea cmpl $0x1910, %eax 0xffffff800042a4ef movl $0x2000, %eax 0xffffff800042a4f4 movl $0x1000, %ecx 0xffffff800042a4f9 cmovel %eax, %ecx 0xffffff800042a4fc movl %ecx, _xcpm_cpu_model(%rip) 0xffffff800042a502 movl $0x0, 0x6028e8(%rip)
This routine has no matching case number for the Ivy Bridge E, Haswell E and Broadwell E processors. That is why we lower the model number to make it match with a supported processors model. Like we are using a normal Ivy Bridge, Haswell or Broadwell processor. A simple but effective trick it seems.
I hope that my explanation about all this helps you to understand what we are doing.
Edit: Make sure that you either use a SMBIOS model/board-id with FrequencyVectors data in its plist, or patch it with help of freqVectorsEdit.sh v2.3.
You can verify that the FrequencyVector data is loaded with help of sysctl -n machdep.xcpm.vectors_loaded_count
Tips:
1.) The X86PlatformPlugin.kext will only load with the plugin-type property is set on the first logical CPU. This however is not enough to enable XCPM mode. No. You may still use AppleIntelCPUPowerManagement.kext Even when X86PlatformShim.kext is loaded.
2.) The FrequencyVectors data in the plist is used to configure power management, and is not the same for all models/board-ids. Please use one that works for your setup.
3.) If sysctl -n machdep.xcpm.vectors_loaded_count returns 0 then the FrequencyVectors data is not being used. Backup the plist for your board-id and replace it with a different plist.
4.) If you use ssdtPRGen.sh to generate your ssdt_pr.aml then make sure to use the -turbo [top-turbo-frequency] argument for overclocked setups.
5.) Check for XCPM related errors at boot time. Like this: X86PlatformShim::start – Failed to send stepper. You got to fix errors or things may not work properly.
TODOs:
1.) Fix LFM frequency (fixed to 800MHz).
2.) Figure out what MSRs trigger a reboot and only block them. Not all other supported MSRs as well.
Pike's Universum 
Hi Pike,
Thanks for all info in you’r blog that really help me for set up my system. After long setup the power management it’s working but still missing the C3 States. And my MSR it’s still in lock. Am I need to patch bios for make this MSR unlock? And how to enable the C3 states, I’m using Asus X99 Strix and i7-6900k.
And also I am still use FakeCpuID 0x040674 for make system running.
Please take a look on the output of AppleIntelInfo.kext below, are theres something wrong that need to fix?
Thanks
AppleIntelInfo.kext v1.8c Copyright © 2012-2016 Pike R. Alpha. All rights reserved
Settings:
——————————————
logMSRs…………………………….: 1
logIGPU…………………………….: 0
logCStates………………………….: 1
logIPGStyle…………………………: 1
InitialTSC………………………….: 0xea91a5a5a16 (503 MHz)
MWAIT C-States………………………: 8480
Processor Brandstring………………..: Intel(R) Core(TM) i7-6900K CPU @ 3.20GHz
Processor Signature………………… : 0x406F1
——————————————
– Family…………………………. : 6
– Stepping……………………….. : 1
– Model………………………….. : 0x4F (79)
Model Specific Registers (MSRs)
——————————————
MSR_CORE_THREAD_COUNT…………(0x35) : 0xFFFFFF809CD8C300
——————————————
– Core Count……………………… : 8
– Thread Count……………………. : 16
MSR_PLATFORM_INFO…………….(0xCE) : 0x20080C3BF3812000
——————————————
– Maximum Non-Turbo Ratio………….. : 0x20 (3200 MHz)
– Ratio Limit for Turbo Mode……….. : 1 (programmable)
– TDP Limit for Turbo Mode…………. : 1 (programmable)
– Low Power Mode Support…………… : 1 (LPM supported)
– Number of ConfigTDP Levels……….. : 1 (additional TDP level(s) available)
– Maximum Efficiency Ratio…………. : 12
– Minimum Operating Ratio………….. : 8
MSR_PMG_CST_CONFIG_CONTROL…….(0xE2) : 0x8400
——————————————
– I/O MWAIT Redirection Enable……… : 1 (enabled, IO read of MSR(0xE4) mapped to MWAIT)
– CFG Lock……………………….. : 1 (MSR locked until next reset)
– C3 State Auto Demotion…………… : 0 (disabled/unsupported)
– C1 State Auto Demotion…………… : 0 (disabled/unsupported)
– C3 State Undemotion……………… : 0 (disabled/unsupported)
– C1 State Undemotion……………… : 0 (disabled/unsupported)
– Package C-State Auto Demotion…….. : 0 (disabled/unsupported)
– Package C-State Undemotion……….. : 0 (disabled/unsupported)
MSR_PMG_IO_CAPTURE_BASE……….(0xE4) : 0x10414
——————————————
– LVL_2 Base Address………………. : 0x414
– C-state Range…………………… : 1 (C6 is the max C-State to include)
IA32_MPERF…………………..(0xE7) : 0x91D99D7C2
IA32_APERF…………………..(0xE8) : 0x7E54D3A60
MSR_0x150……………………(0x150) : 0x72525
MSR_FLEX_RATIO……………….(0x194) : 0xE0000
——————————————
MSR_IA32_PERF_STATUS………….(0x198) : 0x18C000000D00
——————————————
– Current Performance State Value…… : 0xD00 (1300 MHz)
MSR_IA32_PERF_CONTROL…………(0x199) : 0xA00
——————————————
– Target performance State Value……. : 0xA00 (1000 MHz)
– Intel Dynamic Acceleration……….. : 0 (IDA engaged)
IA32_CLOCK_MODULATION…………(0x19A) : 0x0
IA32_THERM_STATUS…………….(0x19C) : 0x88460000
IA32_MISC_ENABLES…………….(0x1A0) : 0x850089
——————————————
– Fast-Strings……………………. : 1 (enabled)
– Automatic Thermal Control Circuit…. : 1 (enabled)
– Performance Monitoring…………… : 1 (available)
– Enhanced Intel SpeedStep Technology.. : 1 (enabled)
MSR_TEMPERATURE_TARGET………..(0x1A2) : 0x640A00
——————————————
– Turbo Attenuation Units………….. : 0
– Temperature Target………………. : 100
– TCC Activation Offset……………. : 0
MSR_MISC_PWR_MGMT…………….(0x1AA) : 0x402000
——————————————
– EIST Hardware Coordination……….. : 0 (enabled)
MSR_TURBO_RATIO_LIMIT…………(0x1AD) : 0x2828282828282828
——————————————
IA32_ENERGY_PERF_BIAS…………(0x1B0) : 0x0
MSR_POWER_CTL………………..(0x1FC) : 0x29040059
——————————————
– C1E Enable……………………….: 0
MSR_RAPL_POWER_UNIT…………..(0x606) : 0xA0E03
——————————————
– Power Units…………………….. : 3 (1/8 Watt)
– Energy Status Units……………… : 14 (61 micro-Joules)
– Time Units …………………….. : 10 (976.6 micro-Seconds)
MSR_PKG_POWER_LIMIT…………..(0x610) : 0x7FFF80015FFF8
——————————————
– Package Power Limit #1…………… : 4095 Watt
– Enable Power Limit #1……………. : 1 (enabled)
– Package Clamping Limitation #1……. : 1 (allow going below OS-requested P/T state during Time Window for Power Limit #1)
– Time Window for Power Limit #1……. : 10 (2560 milli-Seconds)
– Package Power Limit #2…………… : 4095 Watt
– Enable Power Limit #2……………. : 1 (enabled)
– Package Clamping Limitation #2……. : 1 (allow going below OS-requested P/T state setting Time Window for Power Limit #2)
– Time Window for Power Limit #2……. : 3 (20 milli-Seconds)
– Lock…………………………… : 0 (MSR not locked)
MSR_PKG_ENERGY_STATUS…………(0x611) : 0x32DFD0A
——————————————
– Total Energy Consumed……………. : 3255 Joules (Watt = Joules / seconds)
MSR_PKG_POWER_INFO……………(0x614) : 0x1780460
——————————————
– Thermal Spec Power………………. : 140 Watt
– Minimum Power…………………… : 0
– Maximum Power…………………… : 0
– Maximum Time Window……………… : 0
MSR_PP0_POWER_LIMIT…………..(0x638) : 0x0
MSR_PP0_ENERGY_STATUS…………(0x639) : 0x0
MSR_TURBO_ACTIVATION_RATIO…….(0x64C) : 0x0
MSR_PKGC3_IRTL……………….(0x60a) : 0x0
MSR_PKGC6_IRTL……………….(0x60b) : 0x0
MSR_PKG_C2_RESIDENCY………….(0x60d) : 0x0
MSR_PKG_C3_RESIDENCY………….(0x3f8) : 0x0
MSR_PKG_C2_RESIDENCY………….(0x60d) : 0x0
MSR_PKG_C6_RESIDENCY………….(0x3f9) : 0x0
IA32_TSC_DEADLINE…………….(0x6E0) : 0xEA91DE6DEC9
CPU Ratio Info:
——————————————
Base Clock Frequency (BLCK)…………. : 100 MHz
Maximum Efficiency Ratio/Frequency…….: 12 (1200 MHz)
Maximum non-Turbo Ratio/Frequency……..: 32 (3200 MHz)
Maximum Turbo Ratio/Frequency…………: 40 (4000 MHz)
P-State ratio * 100 = Frequency in MHz
——————————————
CPU P-States [ (12) 22 ]
CPU C6-Cores [ 0 2 4 6 8 10 13 15 ]
CPU C6-Cores [ 0 2 4 6 8 10 13 14 15 ]
CPU C6-Cores [ 0 2 4 6 8 10 12 13 14 15 ]
CPU P-States [ (12) 22 27 ]
CPU C6-Cores [ 0 2 4 5 6 8 10 12 13 14 15 ]
CPU P-States [ (12) 13 22 27 ]
CPU C6-Cores [ 0 1 2 3 4 5 6 8 10 12 13 14 15 ]
CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 12 13 14 15 ]
CPU P-States [ (12) 13 18 22 27 ]
CPU P-States [ 12 13 17 18 22 27 (37) ]
CPU P-States [ (12) 13 16 17 18 22 27 37 ]
CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 ]
CPU P-States [ (12) 13 16 17 18 22 26 27 37 ]
CPU P-States [ (12) 13 15 16 17 18 22 26 27 37 ]
CPU P-States [ (12) 13 15 16 17 18 22 26 27 28 37 ]
CPU P-States [ (12) 13 14 15 16 17 18 22 26 27 28 37 ]
CPU P-States [ (12) 13 14 15 16 17 18 20 22 26 27 28 37 ]
CPU P-States [ 12 13 14 15 16 17 18 20 22 24 26 27 28 (37) ]
CPU P-States [ 12 13 14 15 16 17 18 19 20 22 24 26 27 28 (37) ]
CPU P-States [ (12) 13 14 15 16 17 18 19 20 22 23 24 26 27 28 37 ]
CPU P-States [ 12 13 14 15 16 17 18 19 20 22 23 24 26 27 28 37 (40) ]
CPU P-States [ 12 13 14 15 16 17 18 19 20 21 22 23 24 26 27 28 (37) 40 ]
CPU P-States [ (12) 13 14 15 16 17 18 19 20 21 22 23 24 26 27 28 34 37 40 ]
CPU P-States [ (12) 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 34 37 40 ]
CPU P-States [ 12 (13) 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 30 34 37 40 ]
CPU P-States [ (12) 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 30 32 34 37 40 ]
CPU P-States [ (12) 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 32 34 37 40 ]
CPU P-States [ (12) 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 34 37 40 ]
CPU P-States [ 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 34 37 (38) 40 ]
CPU P-States [ 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 34 35 (37) 38 40 ]
CPU P-States [ 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 34 35 36 (37) 38 40 ]
CPU P-States [ (12) 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 40 ]
CPU P-States [ 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 (40) ]
Hi Pike,
I have continued my tests, and i can see the same results of a jumping CPU frequency with windows.
When i disable EIST (only use C-States in Bios) and boot into Windows, then the CPU jumps from min. to max. multiplier.
Read write everything shows me this msr settings
0x01A0: 840089
0x01AA: 400000
0xE2: 403
When i boot with EIST, then CPU stays at min. (12) multiplier.
0x01A0: 850089
0x01AA: 400000
0xE2: 403
The same MSR settings used with OS X give me bad CPU performance, my CPU don´t use max multiplier and can´t get full performance.
I also have noticed, xeon shows a value in msr 0x064C. Haswell-E/Broadwell-E shows there always 0x0?
I hope this help some specialist to investigate this problem.
I can get xcpm working with using FakeCPUID 0x040652, but only using this reboot fix:
reboot fix Sierra © Pike R. Alpha
Find: VUiJ5UFXQVZBVUFUU1BBidZBifdIiftFhf8PhA==
Replace: w5CJ5UFXQVZBVUFUU1BBidZBifdIiftFhf8PhA==
If I don’t have reboot fix, the boot is just stuck on:
OsxAptioFixDrv: Starting overrides for System\Library\CoreServices\boot.efi
Using reloc block: yes, hibernate wake: no
++++++++++++++++++++++++++++++++++++++++++++++
I have Asus X99-A patched using UEFIPatch.
AppleIntelInfo.kext v1.8c Copyright © 2012-2016 Pike R. Alpha. All rights reservedSettings:
------------------------------------------
logMSRs..................................: 1
logIGPU..................................: 0
logCStates...............................: 1
logIPGStyle..............................: 1
InitialTSC...............................: 0xf07f05a4c7 (31 MHz)
MWAIT C-States...........................: 8480
Processor Brandstring....................: Intel(R) Core(TM) i7-5820K CPU @ 3.30GHz
Processor Signature..................... : 0x306F2
------------------------------------------
- Family............................... : 6
- Stepping............................. : 2
- Model................................ : 0x3F (63)
Model Specific Registers (MSRs)
------------------------------------------
MSR_CORE_THREAD_COUNT............(0x35) : 0xFFFFFF803B2D9700
------------------------------------------
- Core Count........................... : 6
- Thread Count......................... : 12
MSR_PLATFORM_INFO................(0xCE) : 0x20080C3BF3812100
------------------------------------------
- Maximum Non-Turbo Ratio.............. : 0x21 (3300 MHz)
- Ratio Limit for Turbo Mode........... : 1 (programmable)
- TDP Limit for Turbo Mode............. : 1 (programmable)
- Low Power Mode Support............... : 1 (LPM supported)
- Number of ConfigTDP Levels........... : 1 (additional TDP level(s) available)
- Maximum Efficiency Ratio............. : 12
- Minimum Operating Ratio.............. : 8
MSR_PMG_CST_CONFIG_CONTROL.......(0xE2) : 0x400
------------------------------------------
- I/O MWAIT Redirection Enable......... : 1 (enabled, IO read of MSR(0xE4) mapped to MWAIT)
- CFG Lock............................. : 0 (MSR not locked)
- C3 State Auto Demotion............... : 0 (disabled/unsupported)
- C1 State Auto Demotion............... : 0 (disabled/unsupported)
- C3 State Undemotion.................. : 0 (disabled/unsupported)
- C1 State Undemotion.................. : 0 (disabled/unsupported)
- Package C-State Auto Demotion........ : 0 (disabled/unsupported)
- Package C-State Undemotion........... : 0 (disabled/unsupported)
MSR_PMG_IO_CAPTURE_BASE..........(0xE4) : 0x10414
------------------------------------------
- LVL_2 Base Address................... : 0x414
- C-state Range........................ : 1 (C6 is the max C-State to include)
IA32_MPERF.......................(0xE7) : 0x777659587
IA32_APERF.......................(0xE8) : 0x9D297D497
MSR_FLEX_RATIO...................(0x194) : 0xE0000
------------------------------------------
MSR_IA32_PERF_STATUS.............(0x198) : 0x295200002D00
------------------------------------------
- Current Performance State Value...... : 0x2D00 (4500 MHz)
MSR_IA32_PERF_CONTROL............(0x199) : 0xC00
------------------------------------------
- Target performance State Value....... : 0xC00 (1200 MHz)
- Intel Dynamic Acceleration........... : 0 (IDA engaged)
IA32_CLOCK_MODULATION............(0x19A) : 0x0
IA32_THERM_STATUS................(0x19C) : 0x88420000
IA32_MISC_ENABLES................(0x1A0) : 0x840089
------------------------------------------
- Fast-Strings......................... : 1 (enabled)
- Automatic Thermal Control Circuit.... : 1 (enabled)
- Performance Monitoring............... : 1 (available)
- Enhanced Intel SpeedStep Technology.. : 0 (disabled)
MSR_TEMPERATURE_TARGET...........(0x1A2) : 0x670A00
------------------------------------------
- Turbo Attenuation Units.............. : 0
- Temperature Target................... : 103
- TCC Activation Offset................ : 0
MSR_MISC_PWR_MGMT................(0x1AA) : 0x400000
------------------------------------------
- EIST Hardware Coordination........... : 0 (enabled)
MSR_TURBO_RATIO_LIMIT............(0x1AD) : 0x22222D2D2D2D2D2D
------------------------------------------
IA32_ENERGY_PERF_BIAS............(0x1B0) : 0x0
MSR_POWER_CTL....................(0x1FC) : 0x29040059
------------------------------------------
- C1E Enable............................: 0
MSR_RAPL_POWER_UNIT..............(0x606) : 0xA0E03
------------------------------------------
- Power Units.......................... : 3 (1/8 Watt)
- Energy Status Units.................. : 14 (61 micro-Joules)
- Time Units .......................... : 10 (976.6 micro-Seconds)
MSR_PKG_POWER_LIMIT..............(0x610) : 0x7FFF80015FFF8
------------------------------------------
- Package Power Limit #1............... : 4095 Watt
- Enable Power Limit #1................ : 1 (enabled)
- Package Clamping Limitation #1....... : 1 (allow going below OS-requested P/T state during Time Window for Power Limit #1)
- Time Window for Power Limit #1....... : 10 (2560 milli-Seconds)
- Package Power Limit #2............... : 4095 Watt
- Enable Power Limit #2................ : 1 (enabled)
- Package Clamping Limitation #2....... : 1 (allow going below OS-requested P/T state setting Time Window for Power Limit #2)
- Time Window for Power Limit #2....... : 3 (20 milli-Seconds)
- Lock................................. : 0 (MSR not locked)
MSR_PKG_ENERGY_STATUS............(0x611) : 0x1E8F6FC
------------------------------------------
- Total Energy Consumed................ : 1955 Joules (Watt = Joules / seconds)
MSR_PKG_POWER_INFO...............(0x614) : 0x1280460
------------------------------------------
- Thermal Spec Power................... : 140 Watt
- Minimum Power........................ : 0
- Maximum Power........................ : 0
- Maximum Time Window.................. : 0
MSR_PP0_POWER_LIMIT..............(0x638) : 0x0
MSR_PP0_ENERGY_STATUS............(0x639) : 0x0
MSR_TURBO_ACTIVATION_RATIO.......(0x64C) : 0x0
MSR_CONFIG_TDP_NOMINAL...........(0x648) : 0x21
MSR_CONFIG_TDP_LEVEL1............(0x649) : 0x940000001E0460
MSR_CONFIG_TDP_LEVEL2............(0x64a) : 0x94000000000000
MSR_CONFIG_TDP_CONTROL...........(0x64b) : 0x0
MSR_TURBO_ACTIVATION_RATIO.......(0x64c) : 0x0
MSR_PKGC6_IRTL...................(0x60b) : 0x0
MSR_PKG_C2_RESIDENCY.............(0x60d) : 0x0
MSR_PKG_C6_RESIDENCY.............(0x3f9) : 0x0
IA32_TSC_DEADLINE................(0x6E0) : 0xF0820FFEF1
CPU Ratio Info:
------------------------------------------
Base Clock Frequency (BLCK)............. : 100 MHz
Maximum Efficiency Ratio/Frequency.......: 12 (1200 MHz)
Maximum non-Turbo Ratio/Frequency........: 33 (3300 MHz)
Maximum Turbo Ratio/Frequency............: 45 (4500 MHz)
P-State ratio * 100 = Frequency in MHz
------------------------------------------
CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 ]
CPU P-States [ (12) 23 26 27 28 29 30 31 32 34 35 36 37 38 39 40 41 42 43 44 45 ]
Any recommendations? The system seems to be stable, just want to make sure everything is set up properly.
What I can tell you about the reboot fix is that you need it because your processor does not support certain MSR’s. You could experiment with specific patches for the unsupported MSR’s. That may give better results.
I see that the package limits #1 and #2 are wrong (4095 Watt) so that is something that you can fix by changing settings in the BIOS (see previous discussions about this). One other thing is that C1E is not enabled in the BIOS. You better enable it. And at all times.
Hi Pike,
What the package limits #1 and #2 should be? Can you let me know please?Thanks
TDP and TDP plus 20% or 25%.
Guys, I’m getting the same “Clover hw.busfrequency error detected”… what’s the exact syntax to fix the QPI in clover config.plist ? Would really appreciate it..
See zeta WiKi.
Thanks Piker, got it.. Any idea what QPI value would be ideal to put for my i7 4790k ? … Can’t seem to find any ressource on this, only the processor datasheet http://ark.intel.com/products/80807/Intel-Core-i7-4790K-Processor-8M-Cache-up-to-4_40-GHz
100 usually.
Worked like a charm. Thanks again Piker !! May the force be with you always
SystemParameters
NvidiaWeb
In clover config.plist SystemParameters add those two lines NvidiaWeb/true
You need to use codetags or it won’t show up properly. Post a new reply and I will fix it.
I seem to have the same bouncing frequency issue as others have reported. Tested in Windows, and the system idles nicely around 800Mhz and ramps up to 4Ghz as needed.
I noticed I also have the warning about the bus speed incorrect in Clover and applied the fix, my output is below. Interestingly, my C/P states are very much minimized with this QPI fix in place. Without it, I have a nice dump of states at the end of the kext output.
I posted earlier about this, but I also suffer from loss of throttling after a sleep.wake cycle, with the cpu peaking at a constant 4Ghz.
Any observations appreciated.
AppleIntelInfo.kext v1.8 Copyright © 2012-2016 Pike R. Alpha. All rights reservedSettings:
------------------------------------------
logMSRs..................................: 1
logIGPU..................................: 0
logCStates...............................: 1
logIPGStyle..............................: 1
InitialTSC...............................: 0x15985c3944 (2 MHz)
MWAIT C-States...........................: 8480
Processor Brandstring....................: Intel(R) Core(TM) i7-5820K CPU @ 3.30GHz
Processor Signature..................... : 0x306F2
------------------------------------------
- Family............................... : 6
- Stepping............................. : 2
- Model................................ : 0x3F (63)
Model Specific Registers (MSRs)
------------------------------------------
MSR_CORE_THREAD_COUNT............(0x35) : 0x6000C
------------------------------------------
- Core Count........................... : 6
- Thread Count......................... : 12
MSR_PLATFORM_INFO................(0xCE) : 0x80C3BF3812100
------------------------------------------
- Maximum Non-Turbo Ratio.............. : 0x21 (3300 MHz)
- Ratio Limit for Turbo Mode........... : 1 (programmable)
- TDP Limit for Turbo Mode............. : 1 (programmable)
- Low Power Mode Support............... : 1 (LPM supported)
- Number of ConfigTDP Levels........... : 1 (additional TDP level(s) available)
- Maximum Efficiency Ratio............. : 12
- Minimum Operating Ratio.............. : 8
MSR_PMG_CST_CONFIG_CONTROL.......(0xE2) : 0x403
------------------------------------------
- I/O MWAIT Redirection Enable......... : 1 (enabled, IO read of MSR(0xE4) mapped to MWAIT)
- CFG Lock............................. : 0 (MSR not locked)
- C3 State Auto Demotion............... : 0 (disabled/unsupported)
- C1 State Auto Demotion............... : 0 (disabled/unsupported)
- C3 State Undemotion.................. : 0 (disabled/unsupported)
- C1 State Undemotion.................. : 0 (disabled/unsupported)
- Package C-State Auto Demotion........ : 0 (disabled/unsupported)
- Package C-State Undemotion........... : 0 (disabled/unsupported)
MSR_PMG_IO_CAPTURE_BASE..........(0xE4) : 0x10414
------------------------------------------
- LVL_2 Base Address................... : 0x414
- C-state Range........................ : 1 (C6 is the max C-State to include)
IA32_MPERF.......................(0xE7) : 0x13CBACAEE
IA32_APERF.......................(0xE8) : 0x14E07E042
MSR_FLEX_RATIO...................(0x194) : 0xE0000
------------------------------------------
MSR_IA32_PERF_STATUS.............(0x198) : 0x266200002800
------------------------------------------
- Current Performance State Value...... : 0x2800 (4000 MHz)
MSR_IA32_PERF_CONTROL............(0x199) : 0xFF00
------------------------------------------
- Target performance State Value....... : 0xFF00 (25500 MHz)
- Intel Dynamic Acceleration........... : 0 (IDA engaged)
IA32_CLOCK_MODULATION............(0x19A) : 0x0
IA32_THERM_STATUS................(0x19C) : 0x883D0000
IA32_MISC_ENABLES................(0x1A0) : 0x850089
------------------------------------------
- Fast-Strings......................... : 1 (enabled)
- Automatic Thermal Control Circuit.... : 1 (enabled)
- Performance Monitoring............... : 1 (available)
- Enhanced Intel SpeedStep Technology.. : 1 (enabled)
MSR_TEMPERATURE_TARGET...........(0x1A2) : 0x5E0A00
------------------------------------------
- Turbo Attenuation Units.............. : 0
- Temperature Target................... : 94
- TCC Activation Offset................ : 0
MSR_MISC_PWR_MGMT................(0x1AA) : 0x400000
------------------------------------------
- EIST Hardware Coordination........... : 0 (enabled)
MSR_TURBO_RATIO_LIMIT............(0x1AD) : 0x2222282828282828
------------------------------------------
- Maximum Ratio Limit for C01.......... : 28 (4000 MHz)
- Maximum Ratio Limit for C02.......... : 28 (4000 MHz)
- Maximum Ratio Limit for C03.......... : 28 (4000 MHz)
- Maximum Ratio Limit for C04.......... : 28 (4000 MHz)
- Maximum Ratio Limit for C05.......... : 28 (4000 MHz)
- Maximum Ratio Limit for C06.......... : 28 (4000 MHz)
IA32_ENERGY_PERF_BIAS............(0x1B0) : 0x1
------------------------------------------
- Power Policy Preference...............: 1 (highest performance)
MSR_POWER_CTL....................(0x1FC) : 0x2104005B
------------------------------------------
- C1E Enable............................: 1
MSR_RAPL_POWER_UNIT..............(0x606) : 0xA0E03
------------------------------------------
- Power Units.......................... : 3 (1/8 Watt)
- Energy Status Units.................. : 14 (61 micro-Joules)
- Time Units .......................... : 10 (976.6 micro-Seconds)
MSR_PKG_POWER_LIMIT..............(0x610) : 0x7FFD00014EA82
------------------------------------------
- Package Power Limit #1............... : 3408 Watt
- Enable Power Limit #1................ : 1 (enabled)
- Package Clamping Limitation #1....... : 0 (disabled)
- Time Window for Power Limit #1....... : 10 (2560 milli-Seconds)
- Package Power Limit #2............... : 4090 Watt
- Enable Power Limit #2................ : 1 (enabled)
- Package Clamping Limitation #2....... : 1 (allow going below OS-requested P/T state setting Time Window for Power Limit #2)
- Time Window for Power Limit #2....... : 3 (20 milli-Seconds)
- Lock................................. : 0 (MSR not locked)
MSR_PKG_ENERGY_STATUS............(0x611) : 0x67D158
------------------------------------------
- Total Energy Consumed................ : 415 Joules (Watt = Joules / seconds)
MSR_PKG_POWER_INFO...............(0x614) : 0x1280460
------------------------------------------
- Thermal Spec Power................... : 140 Watt
- Minimum Power........................ : 0
- Maximum Power........................ : 0
- Maximum Time Window.................. : 0
MSR_PP0_POWER_LIMIT..............(0x638) : 0x14FFD0
------------------------------------------
- Power Limit.......................... : 4090 Watt
- Enable Power Limit................... : 1 (enabled)
- Clamping Limitation.................. : 0 (disabled)
- Time Window for Power Limit.......... : 10 (10240 milli-Seconds)
- Lock................................. : 0 (MSR not locked)
MSR_PP0_ENERGY_STATUS............(0x639) : 0x0
MSR_TURBO_ACTIVATION_RATIO.......(0x64C) : 0x0
MSR_CONFIG_TDP_NOMINAL...........(0x648) : 0x21
MSR_CONFIG_TDP_LEVEL1............(0x649) : 0x940000001E0460
MSR_CONFIG_TDP_LEVEL2............(0x64a) : 0x94000000000000
MSR_CONFIG_TDP_CONTROL...........(0x64b) : 0x0
MSR_TURBO_ACTIVATION_RATIO.......(0x64c) : 0x0
MSR_PKGC6_IRTL...................(0x60b) : 0x0
MSR_PKG_C2_RESIDENCY.............(0x60d) : 0x0
MSR_PKG_C6_RESIDENCY.............(0x3f9) : 0x0
IA32_TSC_DEADLINE................(0x6E0) : 0x159AABAB18
CPU Ratio Info:
------------------------------------------
Base Clock Frequency (BLCK)............. : 100 MHz
Maximum Efficiency Ratio/Frequency.......: 12 (1200 MHz)
Maximum non-Turbo Ratio/Frequency........: 33 (3300 MHz)
Maximum Turbo Ratio/Frequency............: 40 (4000 MHz)
P-State ratio * 100 = Frequency in MHz
------------------------------------------
CPU P-States [ 37 (40) ]
CPU C6-Cores [ 2 4 6 8 10 ]
CPU C6-Cores [ 2 4 6 8 9 10 11 ]
No idea yet, but I do see an awful lot of people with Low Power Mode Support enabled/supported with this problem.
Hi, Pike.
Sorry for my foolish question. And how to find out the data in a location? e.g. 0x1f8d71
Thanks…
Data? You mean the byte values? If that is what you are looking for, then see xxd or use a hex editor.
Yeah. Got it. But it seems that the offset changed under 16A323. It’s 0x1f8d51, 0x227fa2, 0x21e440, 0x21e46f, and 0x228380.
Sure. Likely, but I do not have the time to keep updating the address info. The message is out and you seem to got it so we’re done here. Right?
Of course yes.😜
And I have one more question to ask… Thai is, kernel location 0x228380 change 0x55 into 0xC3 (ret) to stop the KP not worked for me, and I had to use these patches (https://pikeralpha.wordpress.com/2016/07/26/xcpm-for-unsupported-processor/comment-page-1/#comment-6301), and confirmed, these byte values do not change. Still worked in 16A323. I would like to know why… I have no MSR lock, natively, not unlock this for my motherboard.
I think the link above contains too many data to patch, not convenient once the bytes changes.
Anyway, I’d like to use ret to prevent the immediate reboot…
Any idea? Thanks.
Hi Piker…i know my reply is typical by a newbie but please check my AppleIntelInfo log…something is not working correctly in my system (x99 deluxe U3.1 msr unlocked + 5960X) running sierra final
AppleIntelInfo.kext v1.8c Copyright © 2012-2016 Pike R. Alpha. All rights reservedSettings:
------------------------------------------
logMSRs..................................: 1
logIGPU..................................: 0
logCStates...............................: 1
logIPGStyle..............................: 1
InitialTSC...............................: 0x10f99e7de1a (38 MHz)
MWAIT C-States...........................: 8480
Processor Brandstring....................: Intel(R) Core(TM) i7-5960X CPU @ 3.00GHz
Processor Signature..................... : 0x306F2
------------------------------------------
- Family............................... : 6
- Stepping............................. : 2
- Model................................ : 0x3F (63)
Model Specific Registers (MSRs)
------------------------------------------
MSR_CORE_THREAD_COUNT............(0x35) : 0x0
------------------------------------------
- Core Count........................... : 8
- Thread Count......................... : 16
MSR_PLATFORM_INFO................(0xCE) : 0x20080C3BF3811E00
------------------------------------------
- Maximum Non-Turbo Ratio.............. : 0x1E (3000 MHz)
- Ratio Limit for Turbo Mode........... : 1 (programmable)
- TDP Limit for Turbo Mode............. : 1 (programmable)
- Low Power Mode Support............... : 1 (LPM supported)
- Number of ConfigTDP Levels........... : 1 (additional TDP level(s) available)
- Maximum Efficiency Ratio............. : 12
- Minimum Operating Ratio.............. : 8
MSR_PMG_CST_CONFIG_CONTROL.......(0xE2) : 0x400
------------------------------------------
- I/O MWAIT Redirection Enable......... : 1 (enabled, IO read of MSR(0xE4) mapped to MWAIT)
- CFG Lock............................. : 0 (MSR not locked)
- C3 State Auto Demotion............... : 0 (disabled/unsupported)
- C1 State Auto Demotion............... : 0 (disabled/unsupported)
- C3 State Undemotion.................. : 0 (disabled/unsupported)
- C1 State Undemotion.................. : 0 (disabled/unsupported)
- Package C-State Auto Demotion........ : 0 (disabled/unsupported)
- Package C-State Undemotion........... : 0 (disabled/unsupported)
MSR_PMG_IO_CAPTURE_BASE..........(0xE4) : 0x10414
------------------------------------------
- LVL_2 Base Address................... : 0x414
- C-state Range........................ : 1 (C6 is the max C-State to include)
IA32_MPERF.......................(0xE7) : 0x2A5A6D18C8
IA32_APERF.......................(0xE8) : 0x30BE11A19A
MSR_FLEX_RATIO...................(0x194) : 0xE0000
------------------------------------------
MSR_IA32_PERF_STATUS.............(0x198) : 0x271C00002800
------------------------------------------
- Current Performance State Value...... : 0x2800 (4000 MHz)
MSR_IA32_PERF_CONTROL............(0x199) : 0xFF00
------------------------------------------
- Target performance State Value....... : 0xFF00 (25500 MHz)
- Intel Dynamic Acceleration........... : 0 (IDA engaged)
IA32_CLOCK_MODULATION............(0x19A) : 0x0
IA32_THERM_STATUS................(0x19C) : 0x884E0000
IA32_MISC_ENABLES................(0x1A0) : 0x840089
------------------------------------------
- Fast-Strings......................... : 1 (enabled)
- Automatic Thermal Control Circuit.... : 1 (enabled)
- Performance Monitoring............... : 1 (available)
- Enhanced Intel SpeedStep Technology.. : 0 (disabled)
MSR_TEMPERATURE_TARGET...........(0x1A2) : 0x691200
------------------------------------------
- Turbo Attenuation Units.............. : 0
- Temperature Target................... : 105
- TCC Activation Offset................ : 0
MSR_MISC_PWR_MGMT................(0x1AA) : 0x400000
------------------------------------------
- EIST Hardware Coordination........... : 0 (enabled)
MSR_TURBO_RATIO_LIMIT............(0x1AD) : 0x2828282828282828
------------------------------------------
IA32_ENERGY_PERF_BIAS............(0x1B0) : 0x0
MSR_POWER_CTL....................(0x1FC) : 0x2904005B
------------------------------------------
- C1E Enable............................: 1
MSR_RAPL_POWER_UNIT..............(0x606) : 0xA0E03
------------------------------------------
- Power Units.......................... : 3 (1/8 Watt)
- Energy Status Units.................. : 14 (61 micro-Joules)
- Time Units .......................... : 10 (976.6 micro-Seconds)
MSR_PKG_POWER_LIMIT..............(0x610) : 0x7FFF80015FFF8
------------------------------------------
- Package Power Limit #1............... : 4095 Watt
- Enable Power Limit #1................ : 1 (enabled)
- Package Clamping Limitation #1....... : 1 (allow going below OS-requested P/T state during Time Window for Power Limit #1)
- Time Window for Power Limit #1....... : 10 (2560 milli-Seconds)
- Package Power Limit #2............... : 4095 Watt
- Enable Power Limit #2................ : 1 (enabled)
- Package Clamping Limitation #2....... : 1 (allow going below OS-requested P/T state setting Time Window for Power Limit #2)
- Time Window for Power Limit #2....... : 3 (20 milli-Seconds)
- Lock................................. : 0 (MSR not locked)
MSR_PKG_ENERGY_STATUS............(0x611) : 0x1F2EE3
------------------------------------------
- Total Energy Consumed................ : 124 Joules (Watt = Joules / seconds)
MSR_PKG_POWER_INFO...............(0x614) : 0x1280460
------------------------------------------
- Thermal Spec Power................... : 140 Watt
- Minimum Power........................ : 0
- Maximum Power........................ : 0
- Maximum Time Window.................. : 0
MSR_PP0_POWER_LIMIT..............(0x638) : 0x0
MSR_PP0_ENERGY_STATUS............(0x639) : 0x0
MSR_TURBO_ACTIVATION_RATIO.......(0x64C) : 0x0
MSR_CONFIG_TDP_NOMINAL...........(0x648) : 0x1E
MSR_CONFIG_TDP_LEVEL1............(0x649) : 0x940000001B0460
MSR_CONFIG_TDP_LEVEL2............(0x64a) : 0x94000000000000
MSR_CONFIG_TDP_CONTROL...........(0x64b) : 0x0
MSR_TURBO_ACTIVATION_RATIO.......(0x64c) : 0x0
MSR_PKGC6_IRTL...................(0x60b) : 0x0
MSR_PKG_C2_RESIDENCY.............(0x60d) : 0x0
MSR_PKG_C6_RESIDENCY.............(0x3f9) : 0x0
IA32_TSC_DEADLINE................(0x6E0) : 0x10FA83D6090
CPU Ratio Info:
------------------------------------------
Base Clock Frequency (BLCK)............. : 100 MHz
Maximum Efficiency Ratio/Frequency.......: 12 (1200 MHz)
Maximum non-Turbo Ratio/Frequency........: 30 (3000 MHz)
Maximum Turbo Ratio/Frequency............: 40 (4000 MHz)
P-State ratio * 100 = Frequency in MHz
------------------------------------------
CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 ]
CPU P-States [ 12 18 19 20 21 22 23 24 25 27 28 29 30 31 32 35 36 37 38 (40) ]
Please read the other comments and then you should be able to figure it out yourself (Package power limit #1 and #2 errors).
Of course i noticed that there is something wrong in package limit but i don’t know how to solve. My bios settings are in Auto, just a moderate overclock to 4000…
The “Auto” setting in the BIOS appears to be the problem. You got to set it to a real value.
Thanks for your answer. I will try to set a real value after reading more about different bios settings. I will let you know.
Hi pike
XCPM worked well Gigabyte X99 Gaming G1
Thank you.
Last login: Sun Oct 9 00:51:54 on ttys000
manoranjans-Pro:~ manoranjan$ sudo chown -R root:wheel /Users/manoranjan/Desktop/AppleIntelInfo.kext
Password:
manoranjans-Pro:~ manoranjan$ sudo chmod -R 755 /Users/manoranjan/Desktop/AppleIntelInfo.kext
manoranjans-Pro:~ manoranjan$ sudo kextload /Users/manoranjan/Desktop/AppleIntelInfo.kext
manoranjans-Pro:~ manoranjan$ sudo cat /tmp/AppleIntelInfo.dat
AppleIntelInfo.kext v1.8e Copyright © 2012-2016 Pike R. Alpha. All rights reserved
Settings:
How do I find the location of 0x01A0 in different kernels so that I can do this myself after updates instead of needing someone else to provide me with patch info?
Thanks!
Thanks. This is a good question. Blog article updated. I hope that you like it 😉
Where?
Here
Sorry, another question. This method works perfectly with any Haswell-EP CPUs with a CPUID of 0x0306f2.
http://www.insanelymac.com/forum/topic/315579-el-capitan-xcpm-for-x99-one-patch-solution/
However, Xeon Haswell-E’s have a CPUID of 0x0306f1. What would I need to change to allow me to use that same method of achieving XCPM?
I tried FakeCPUID in Clover, but that did not change the reported CPUID in AppleIntelInfo, nor did XCPM behavior change.
Thanks!
Are you talking about the ‘turbo-drop’ problem (won’t run flat out all the time) or the ‘frequency-bounching’ problem (drops down to the minimum frequency and then back up)?
If that is a yes, then I would start by comparing the output of AppleIntelInfo.kext of both processors because there might be something that we are missing.
p.s. The Haswell-E and the Xeon processor both report 0x3F as model so that should be fine.
Hi Pike, On insanelymac forum, thanks to your work, people open many thread with different solution to have xcpm working in a best way.
If it possible could you “the Finder” make some clearance?
Your reboot fix patch method for sierra linked to fakecpuid 40764 and Brumbaer 5960x patch allows user with Broadwell E/EP cpu to have access at a partial XCPM.Common problem is no lower steps in appleintleinfo output, no c3 residency, no c3 states
in my case I have a stable system (Xeon broadwell EP 2696 V4)
Now in the last day other xcpm solution based from your finding in this thread is on air!
setting this Patches with clover:
find
qgEAANwzAAAAAAAAAAAAAAAAAAAAAAAAAQ==
replace
qgEAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA==
find
oAEAANwzAAAAAAAAAAAAAAAAAABAAAAAAQAF
replace
oAEAANwzAAAAAAAAAAAAAAAAAABAAAAAAQAE
find
OgYAANwzAAAAAAAAAAAAAB8AAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
replace OgYAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA
find
QgYAANwzAAAAAAAAAAAAAB8AAAAAAAAAGA==
replace
QgYAAAAAAAAAAAAAAAAAAAAAAAAAAAAAAA==
find
4gAAAEwAAAAAAAAAAAAAAA8EAAAAAAAABQAAHg==
replace
4gAAAEwAAAAAAAAAAAAAAA8EAAAAAAAABwAAfg==
produce an output with C3, a programmed msr 0xE2 =7E000007
this:
MSR_PMG_CST_CONFIG_CONTROL.......(0xE2) : 0x7E000007------------------------------------------
- I/O MWAIT Redirection Enable......... : 0 (not enabled)
- CFG Lock............................. : 0 (MSR not locked)
- C3 State Auto Demotion............... : 1 (enabled)
- C1 State Auto Demotion............... : 1 (enabled)
- C3 State Undemotion.................. : 1 (enabled)
- C1 State Undemotion.................. : 1 (enabled)
- Package C-State Auto Demotion........ : 1 (enabled)
- Package C-State Undemotion........... : 1 (enabled)
and this:
MSR_CONFIG_TDP_CONTROL...........(0x64b) : 0x0
MSR_TURBO_ACTIVATION_RATIO.......(0x64c) : 0x2E
MSR_PKGC6_IRTL...................(0x60b) : 0x0
MSR_PKG_C2_RESIDENCY.............(0x60d) : 0xAF73BC5368F
MSR_PKG_C3_RESIDENCY.............(0x3f8) : 0x3CE141
MSR_PKG_C6_RESIDENCY.............(0x3f9) : 0xB21573FDB7A
IA32_TSC_DEADLINE................(0x6E0) : 0x30D4F0BC715A
In my case I have also C3 cores fro 0 to 43
P states Always without lower steps from 13 to 18
same temperature and so on
What do you think about this 5 Patches?
It seems to have the same behaviour of reboot fix patch, but Output of AppleIntelInfo seems to be more complete
Thank you for your time
Fabio,
I do not believe in a “partial XCPM” as you say. To me it (XCPM) either works or not. The fact that it may react differently, than what you expect, is another matter. It may as well be very simple, by wiping the Package Clamping Limitation #1/#2 i.e. bits 16 and 48 of MSR(0x610). You can do that by compiling AppleIntelInfo.kext yourself with a small addition on line 1250:
UInt64 msr_value = rdmsr64(MSR_PKG_POWER_LIMIT);// Setting bits.
// msr_value |= (1L << 16);
// msr_value |= (1L << 48);
// Clearing bits.
msr_value &= ~(1L << 16);
msr_value &= ~(1L << 48);
wrmsr64(MSR_PKG_POWER_LIMIT, msr_value);
I have also included ways to sets bits.
Hi Pike Thank you for your help
I have the same your idea for XCPM (works or not)
But maybe X99 Platform has some strange behaviour
With EIST disabled also in windows there are frequency jumping
Bur With EIST enabled windows have a correct behaviour and no loss of performance
follow my result with adding line you suggested:
AppleIntelInfo.kext v1.8e Copyright © 2012-2016 Pike R. Alpha. All rights reservedSettings:
------------------------------------------
logMSRs..................................: 1
logIGPU..................................: 0
logCStates...............................: 1
logIPGStyle..............................: 1
InitialTSC...............................: 0x3da2c2e3378 (192 MHz)
MWAIT C-States...........................: 8480
Processor Brandstring....................: Intel(R) Xeon(R) CPU E5-2696 v4 @ 2.20GHz
Processor Signature..................... : 0x406F1
------------------------------------------
- Family............................... : 6
- Stepping............................. : 1
- Model................................ : 0x4F (79)
Model Specific Registers (MSRs)
------------------------------------------
MSR_CORE_THREAD_COUNT............(0x35) : 0x0
------------------------------------------
- Core Count........................... : 22
- Thread Count......................... : 44
MSR_PLATFORM_INFO................(0xCE) : 0x20080C3BF2811600
------------------------------------------
- Maximum Non-Turbo Ratio.............. : 0x16 (2200 MHz)
- Ratio Limit for Turbo Mode........... : 1 (programmable)
- TDP Limit for Turbo Mode............. : 1 (programmable)
- Low Power Mode Support............... : 1 (LPM supported)
- Number of ConfigTDP Levels........... : 1 (additional TDP level(s) available)
- Maximum Efficiency Ratio............. : 12
- Minimum Operating Ratio.............. : 8
MSR_PMG_CST_CONFIG_CONTROL.......(0xE2) : 0x7E000007
------------------------------------------
- I/O MWAIT Redirection Enable......... : 0 (not enabled)
- CFG Lock............................. : 0 (MSR not locked)
- C3 State Auto Demotion............... : 1 (enabled)
- C1 State Auto Demotion............... : 1 (enabled)
- C3 State Undemotion.................. : 1 (enabled)
- C1 State Undemotion.................. : 1 (enabled)
- Package C-State Auto Demotion........ : 1 (enabled)
- Package C-State Undemotion........... : 1 (enabled)
MSR_PMG_IO_CAPTURE_BASE..........(0xE4) : 0x10414
------------------------------------------
- LVL_2 Base Address................... : 0x414
- C-state Range........................ : 1 (C-States not included, I/O MWAIT redirection not enabled)
IA32_MPERF.......................(0xE7) : 0x4A5D07FE00
IA32_APERF.......................(0xE8) : 0x5DC28AAA97
MSR_0x150........................(0x150) : 0x1F00000000
MSR_FLEX_RATIO...................(0x194) : 0x0
------------------------------------------
MSR_IA32_PERF_STATUS.............(0x198) : 0x18ED00000C00
------------------------------------------
- Current Performance State Value...... : 0xC00 (1200 MHz)
MSR_IA32_PERF_CONTROL............(0x199) : 0x2500
------------------------------------------
- Target performance State Value....... : 0x2500 (3700 MHz)
- Intel Dynamic Acceleration........... : 0 (IDA engaged)
IA32_CLOCK_MODULATION............(0x19A) : 0x0
IA32_THERM_STATUS................(0x19C) : 0x88500000
IA32_MISC_ENABLES................(0x1A0) : 0x840089
------------------------------------------
- Fast-Strings......................... : 1 (enabled)
- Automatic Thermal Control Circuit.... : 1 (enabled)
- Performance Monitoring............... : 1 (available)
- Processor Event Based Sampling....... : 0 (PEBS supported)
- Enhanced Intel SpeedStep Technology.. : 0 (disabled)
- MONITOR FSM.......................... : 1 (MONITOR/MWAIT supported)
- CFG Lock............................. : 0 (MSR not locked)
MSR_TEMPERATURE_TARGET...........(0x1A2) : 0x640200
------------------------------------------
- Turbo Attenuation Units.............. : 0
- Temperature Target................... : 100
- TCC Activation Offset................ : 0
MSR_MISC_PWR_MGMT................(0x1AA) : 0x402000
------------------------------------------
- EIST Hardware Coordination........... : 0 (hardware coordination enabled)
- Energy/Performance Bias support...... : 1
- Energy/Performance Bias.............. : 0 (disabled/MSR not visible to software)
- Thermal Interrupt Coordination Enable : 1 (thermal interrupt routed to all cores)
MSR_TURBO_RATIO_LIMIT............(0x1AD) : 0x1E1F202122232525
------------------------------------------
- Maximum Ratio Limit for C01.......... : 25 (3700 MHz)
- Maximum Ratio Limit for C02.......... : 25 (3700 MHz)
- Maximum Ratio Limit for C03.......... : 23 (3500 MHz)
- Maximum Ratio Limit for C04.......... : 22 (3400 MHz)
- Maximum Ratio Limit for C05.......... : 21 (3300 MHz)
- Maximum Ratio Limit for C06.......... : 20 (3200 MHz)
- Maximum Ratio Limit for C07.......... : 1F (3100 MHz)
- Maximum Ratio Limit for C08.......... : 1E (3000 MHz)
MSR_TURBO_RATIO_LIMIT1...........(0x1AE) : 0x1C1C1C1C1C1C1C1D
------------------------------------------
- Maximum Ratio Limit for C09.......... : 1D (2900 MHz)
- Maximum Ratio Limit for C10.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C11.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C12.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C13.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C14.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C15.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C16.......... : 1C (2800 MHz)
MSR_TURBO_RATIO_LIMIT2...........(0x1AF) : 0x1C1C1C1C1C1C1C1C
------------------------------------------
- Maximum Ratio Limit for C17.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C18.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C19.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C20.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C21.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C22.......... : 1C (2800 MHz)
IA32_ENERGY_PERF_BIAS............(0x1B0) : 0x1
------------------------------------------
- Power Policy Preference...............: 1 (highest performance)
MSR_POWER_CTL....................(0x1FC) : 0x2904005B
------------------------------------------
- C1E Enable............................: 1 (enabled)
MSR_RAPL_POWER_UNIT..............(0x606) : 0xA0E03
------------------------------------------
- Power Units.......................... : 3 (1/8 Watt)
- Energy Status Units.................. : 14 (61 micro-Joules)
- Time Units .......................... : 10 (976.6 micro-Seconds)
MSR_PKG_POWER_LIMIT..............(0x610) : 0x685A0001484B0
------------------------------------------
- Package Power Limit #1............... : 150 Watt
- Enable Power Limit #1................ : 1 (enabled)
- Package Clamping Limitation #1....... : 0 (disabled)
- Time Window for Power Limit #1....... : 10 (2560 milli-Seconds)
- Package Power Limit #2............... : 180 Watt
- Enable Power Limit #2................ : 1 (enabled)
- Package Clamping Limitation #2....... : 0 (disabled)
- Time Window for Power Limit #2....... : 3 (20 milli-Seconds)
- Lock................................. : 0 (MSR not locked)
MSR_PKG_ENERGY_STATUS............(0x611) : 0xC9D50B
------------------------------------------
- Total Energy Consumed................ : 807 Joules (Watt = Joules / seconds)
MSR_PKG_POWER_INFO...............(0x614) : 0x2F0A18028804B0
------------------------------------------
- Thermal Spec Power................... : 150 Watt
- Minimum Power........................ : 0
- Maximum Power........................ : 323
- Maximum Time Window.................. : 0
MSR_PP0_POWER_LIMIT..............(0x638) : 0x0
MSR_PP0_ENERGY_STATUS............(0x639) : 0x0
MSR_TURBO_ACTIVATION_RATIO.......(0x64C) : 0x0
MSR_PKGC3_IRTL...................(0x60a) : 0x0
MSR_PKGC6_IRTL...................(0x60b) : 0x0
MSR_PKG_C2_RESIDENCY.............(0x60d) : 0x2364231F60
MSR_PKG_C3_RESIDENCY.............(0x3f8) : 0x227406
MSR_PKG_C6_RESIDENCY.............(0x3f9) : 0xB3C9160754
IA32_TSC_DEADLINE................(0x6E0) : 0x3DA39A6118C
CPU Ratio Info:
------------------------------------------
Base Clock Frequency (BLCK)............. : 100 MHz
Maximum Efficiency Ratio/Frequency.......: 12 (1200 MHz)
Maximum non-Turbo Ratio/Frequency........: 22 (2200 MHz)
Maximum Turbo Ratio/Frequency............: 37 (3700 MHz)
P-State ratio * 100 = Frequency in MHz
------------------------------------------
CPU P-States [ 28 35 (37) ]
CPU C3-Cores [ 0 1 2 6 8 11 15 16 22 26 30 ]
CPU C6-Cores [ 0 1 3 4 6 9 11 12 14 17 18 20 23 24 27 31 32 33 34 35 36 37 38 39 40 41 42 43 ]
CPU P-States [ (12) 27 28 35 37 ]
CPU C3-Cores [ 0 1 2 3 5 6 7 8 10 11 15 16 22 26 30 ]
CPU C6-Cores [ 0 1 2 3 4 6 7 8 9 11 12 13 14 15 17 18 20 21 23 24 25 27 29 31 32 33 34 35 36 37 38 39 40 41 42 43 ]
CPU C3-Cores [ 0 1 2 3 4 5 6 7 8 10 11 14 15 16 22 26 30 ]
CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 11 12 13 14 15 16 17 18 20 21 22 23 24 25 27 29 31 32 33 34 35 36 37 38 39 40 41 42 43 ]
CPU P-States [ (12) 26 27 28 35 37 ]
CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 20 21 22 23 24 25 27 29 31 32 33 34 35 36 37 38 39 40 41 42 43 ]
CPU P-States [ (12) 26 27 28 29 35 37 ]
CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 20 21 22 23 24 25 27 28 29 31 32 33 34 35 36 37 38 39 40 41 42 43 ]
CPU P-States [ (12) 26 27 28 29 30 35 37 ]
CPU C3-Cores [ 0 1 2 3 4 5 6 7 8 10 11 14 15 16 22 26 30 31 32 33 34 35 36 37 38 39 40 41 42 43 ]
CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 20 21 22 23 24 25 26 27 28 29 31 32 33 34 35 36 37 38 39 40 41 42 43 ]
CPU P-States [ (12) 26 27 28 29 30 34 35 37 ]
CPU C3-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 14 15 16 17 22 26 27 30 31 32 33 34 35 36 37 38 39 40 41 42 43 ]
CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 ]
CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 ]
CPU C3-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 14 15 16 17 22 23 26 27 30 31 32 33 34 35 36 37 38 39 40 41 42 43 ]
CPU P-States [ (12) 26 27 28 29 30 32 34 35 37 ]
Fabio,
Now that you can set/clear bits it is just a matter of time before someone finds out what the problem is. That is, if the values of certain MSR’s are the problem.
Pike, surely Knowledge will save us! Or it destroy us (more pessimistic vision) 🙂
till that moment, is there a chance to disable EIST if it is enabled in bios?
It seems that there was this possibility but I loose the bit to patch..If I don’t remeber well you talk in your blog about this.
I think it’s a problem with the programming msr 0x199 and reading 0x198?
When EIST is enabled and msr 199 is active used, CPU use lower P-States, but won’t stay at max multiplier under load. Performance is not stable, sometimes CPU uses max. multiplier, then it drops immediatelly down to min. under full load.
Without active msr 199 (EIST) disabled, we have max. performance, but CPU don’t use lower states + umping frequency in idle.
There Must be a problem, and we don’t find/see. All CPUs with more than 4 cores have this problems when xcpm is active.
Disabling msr 0x199 with a Kernel patch (edited the msr to 00 00 in the sub routine) and enabling EIST shows the same result. Full performance, but no lower p states. Msr 0x199 is not programmed.
I have no idea, and i don’t understand assembler code, but there is a msr199.kext where someone tried to find a solution. But no success, problems are not solved.
So should we take a closer look to this subroutines?
If it drops down to a lower frequency, then that is probably for a good reason. I think that the first thing you and others should do is figure out what that condition is. AppleIntelInfo.kext may provide the information that you are looking for (see MSR_CORE_PERF_LIMIT_REASONS/0x64F).
Hi pike
Thanks for XCPM supported.
When you have time please check my report.
Thank you.
Last login: Mon Oct 31 03:07:47 on console
manoranjans-Pro:~ manoranjan$ sudo chown -R root:wheel /Users/manoranjan/Desktop/AppleIntelInfo.kext
Password:
manoranjans-Pro:~ manoranjan$ sudo chmod -R 755 /Users/manoranjan/Desktop/AppleIntelInfo.kext
manoranjans-Pro:~ manoranjan$ sudo kextload /Users/manoranjan/Desktop/AppleIntelInfo.kext
manoranjans-Pro:~ manoranjan$ sudo cat /tmp/AppleIntelInfo.dat
AppleIntelInfo.kext v1.8e Copyright © 2012-2016 Pike R. Alpha. All rights reserved
Settings:
There is nothing to check. All data is missing. When you add data, then do that in a code block!!!
Hi Pike, how are you?
Mind if I tap into your wisdom for a bit? From what I can see there is a hard limit set in xnu for the amount of cores OS can use (seems like its set for 64 now), is there a way to bump that limit or it’s more integrated then just in kernel?
I am fine. Thanks.
About your question. Never been that fortunate, to have so many cores, but from what I remember it checks for 0x3f somewhere in the kernel. In _ml_init_max_cpus? Sorry. Can’t remember it. You may want to change it to a higher/lower value and give it a go.
Thank you for you feedback.
I’ve wasted good part of the day trying to figure out how, what and where.
Any hints on how to do that and convert that string to a something that Clover understands?
One more favor, I’ve been struggling with some odd issues on the already installed system, the terminal crashes all the time, weirdly seems like I’m the only one with that problem.
Crashed Thread: 0 Dispatch queue: com.apple.main-thread
Exception Type: EXC_BAD_ACCESS (SIGSEGV)
Exception Codes: KERN_INVALID_ADDRESS at 0x0000000000000010
Exception Note: EXC_CORPSE_NOTIFY
Termination Signal: Segmentation fault: 11
Termination Reason: Namespace SIGNAL, Code 0xb
Terminating Process: exc handler [0]
VM Regions Near 0x10:
–>
__TEXT 0000000105cbf000-0000000105d86000 [ 796K] r-x/rwx SM=COW /Applications/Utilities/Terminal.app/Contents/MacOS/Terminal
Full report:
http://pastebin.com/raw/JHNNuEbr
The only patch I’m using is CPUID patch.
Thank a lot,
1.) otool -tVj /S*/L*/Kernels/kernel > kernel-disassembled.txt
2.) open kernel-disassembled.txt
3.) Search for: “_ml_init_max_cpus:”
4.) Look for: “cmp rcx, 0x3f”
Tip: You can convert the hex byte code with help of:
echo -n '488d4fff4883f93f7712'|xxd -r -p|base645.) Replace the 3f with anything that you want.
About the terminal.app issue. Try to open the Terminal.app from another volume. If that works then either re-install or replace the Terminal.app
Sorry Pike, I must sound like a broken records… I’m just learning 🙂
I found this:
ffffff8000403e16 48 83 f9 3f cmpq $0x3f, %rcx
Its a bit different, but I’m trying to figure out which any of those number are what and how do you change them?
You are looking for the “3f” there i.e. 63 decimal. You take three lines of hex byte code and convert it like in my example. The one you showed here with “48 83 f9 3f” would be the second line. Don’t forget to remove the spaces.
Pike,
Thank you very much, after playing with different values for the cores 8f did the trick.
Now back to the drawing board on the new kernel panic 🙂
Hi Pike
I have put a lower value to test on a single cpu rig (22cores/44 threads) to force OS to 18 cores/36 threads bu t system produces this kernel panic type
https://postimg.org/image/lmavn9l9f/
could be useful to understand?
You cannot limit the number of cores this way – the BIOS must be set to the same number of cores as well.
Pike regarding this subject, could be related also a line present in AppleACPIPlatform.kext (cpu init) to try to bypass cpu limit (seems 64 cores/threads in Sierra)
Do you think it is possible?
Thank you
Sorry. I don’t understand what you mean. You found a CPU limit in AppleACPIPlatform.kext?
Hi Pike,
I confirm there is the same instruction cmp rcx, 0x3f , just one time in all AppleACPIPlatform.kext, even its patched the same as kernel, still crashing at boot
Hi Pike,
thank you for helping us! This ist great.
Only for information, this is the location where i startet some changing of bits….
sub_ffffff800041f1c0:
this sub_ffffff800041f1c0 is called from _xcpm_dvfs_start, from _xcpm_dvfs_configure,
and also called from: sub_ffffff800041e850 ; _xcpm_cpu_control+1590
_xcpm_cpu_control+1590 is called from sub_ffffff800041fc60 ; XREF=_xcpm_timer+1253, _xcpm_urgency+4778, sub_ffffff8000422680+422
or from _xcpm_cpu_control
or from sub_ffffff8000428f30 ; XREF=_xcpm_cpu_control+1769
really confusing to me 🙂
So I disabled msr 199 here:
ffffff800041f252 mov eax, ebx ; XREF=sub_ffffff800041f1c0+37
ffffff800041f254 shl eax, 0x8
ffffff800041f257 mov ecx, 0x199
ffffff800041f25c xor edx, edx
i used this HEX kernel patch: Find: 89D8C1E008B99901 Replace 89D8C1E008B90000, or 89D8C1E008B9C300
So, msr 0x199 was disabled. xcpm don´t write to it.
And with EIST enabled (msr 0x1A0:850089), the cpu uses max. multiplier of 35.
To solve this, i write manually to msr 0x199, set it to E2, and cpu uses max. multi. x46.
Then the Performace is perfect, same as disabling EIST (msr 0x1A0: 840089) and our old patches. But without EIST, CPU jumps at idle. We know this.
Then i tested some stupid things 🙂
I patched bits at this location:
ffffff800041f254 shl eax, 0x8
For fun: find: 89D8C1E008B99901 replace:89D8C1E00AB99901
i got better performance, CPU reaches faster x46, but it does this not permanent, it also goes down to a lower multiplier when running Geekbench. But the time to reach a higher multiplier is now shorter.
The idle frequency is now not so „smooth“ (with EIST and without patches, i got a flat line at x12).
I know, stupid to change this, but i tested again with a value of 89D8C1E004B99901.
With this, performance is really bad 🙂 it only uses x12.
I think msr 0x199 is very important, for full performance and for ar flat line at idle.
It´s really frustrating, but there is nothing at apples open source
/xnu/xnu-3789.21.1/osfmk/x86_64/xcpm was never published.
Hi Pike,
i read out msr 0x199 / x0198 in idle, and when Geekbench is running.
Test with EIST enabled!
And msr 0x199 don´t change!
MSR 0x198 shows only 3 states! (Jumping frequency, we already know, is here again???)
RDMSR 199 returns value 0x2e00
RDMSR 199 returns value 0x2e00
RDMSR 199 returns value 0x2e00
RDMSR 199 returns value 0x2e00
RDMSR 199 returns value 0x2e00
RDMSR 199 returns value 0x2e00
RDMSR 198 returns value 0x2300
RDMSR 198 returns value 0x2300
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0x2300
RDMSR 199 returns value 0x2e00
RDMSR 199 returns value 0x2e00
RDMSR 199 returns value 0x2e00
RDMSR 199 returns value 0x2e00
RDMSR 199 returns value 0x2e00
RDMSR 199 returns value 0x2e00
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0xc00
RDMSR 198 returns value 0x2300
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0xc00
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0xc00
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0x2300
RDMSR 198 returns value 0xc00
RDMSR 198 returns value 0x2e00
RDMSR 199 returns value 0x2e00
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0x2e00
RDMSR 198 returns value 0xc00
RDMSR 198 returns value 0x2300
RDMSR 198 returns value 0x2e00
This is unexpected, the should chnage????
Running Windows with EIST, this 2 msrs permanently show different states.
I think, xcpm don´t work correct.
We got only P-States working, and some C-States, controlled by Bios/CPU?
Hi Piker – I hope I am not dragging this off-topic and if so then I apologise as I am having a mad theory session again.
If Apple is migrating away from AICPM and going purely for Kernel based power management, could it be that Apple could be thinking of adopting different processors such as the soon to be released AMD Zen family.
If that is the case and if I understand the Zen cpu’s ability to do data encryption directly on the chip would this spell the end of the hackintosh ?
I can’t speak for Apple, but I personally hope that they stick to Intel processors. XCPM is working mighty fine for me with all processors thus far, including the Broadwell-E’s, so I wouldn’t mind.
About the end of hackintosh. Well. Apple’s new filesystem will technically eliminate legacy boot loaders, and who know what the future will bring. We’ll have to wait and see…
@balamut
ffffff8000403e164883f93f kernel value (3f means 63
ffffff8000403e164883f957 patch 57means 83
should be
find
////gABAPhZIg/k/
repalace
////gABAPhZIg/lX
you have to verify by yourself if i did in the right way
Hi, I did it with patching in your way with 59 for 87 cores ( my rig, 2×22 cores). Was able to boot without hyperthreading, 44 cores, so the patch doesn’t break something in the kernel ( EC10.11.6), but with the hyperthreading activated, got the same KP as usual with acpiplatform kext.
I also tried with only 16 activated cores per CPU and HT activated ( 64 cores), but only 32 were recognized by the OS, the os does’nt care about hyperthreading. So even it’s possible to break cpu limit, that need other stuffs to be useful.
Hi Philippe
Which motherboard do you use?
For now I have a single 2696v4 cpu with a single cpu slot motherboard
I did for the sake of testing a patch lowering 64 value to 32 in mac Os sierra kernel ..in my head I would like to force osx to use only 32 for cores/threads instead my real 44
With this patch system does not start..
So maybe it is not the right place to search a solution in my opinion
Hi Fabio, using X10DAX slightly overcocked motherboard with ElCapitan 10.11.6. ( If I find a way to boot Maveriks instead with GTX980 suppurt, It will be 100 times better for me.
To limit cores, that’s a Bios setting. Why do you want to lower value by patch ? I tryed you patch, able to boot withe 44 physical cores activated, but it doesn’t change anything If I activate hyperthreading : still crashing at boot
Hi Philippe
I don’t know if we are off topic here maybe we can discuss on that forum link if you want
I see different strange data in your output
Our cpu has a maximal turbo for 2 cores of x 37 (3700) . I don’t see it from your output
Then, in full load you should have a frequency of x28 (2800) for all your cores/threads
i.e. when you use cinebench 15 cpu bench or geek bench during some LLVM tests
you can monitoring these parameters using IPG (Intel power gadget app)
If in your bios you have EIST parameter try to disable it
if I have it enabled I have to use solution 3 with oarksit modified patch to have max performances in higher turbo levels
Hi Fabio, I uses X10DAX, a slightly overclocked mobo and running on it 10.11.6 with 44 physical cores.
To limit number of cores, first disable hyperthreading in Bios. But why do you want to limit them ?
Hi Philippe
I have tried to see if it was the right place to “hack” without applying any change in bios parameters, but that bit does not override bios parameters.
I have not a dual CPU system
It is curious that you see exactly 44 cores an not 64 cores as that bit limit (3f) could suggest
In that condition (patch without hyperthreading) is your system stable?
How do you know if it uses 22 cores per cpu or all 44 cores/thread of one of it?
thank you
To anwser you question, it seem to be possible in ox to limit cpu number with boot args ; take a look at this : https://www.cnet.com/news/boot-argument-options-in-os-x/
That’s not curious ; the system just see what the bios send to him ; in my case, 44 physical cores, the bios is set to disable hyperthreading, so no more cores wil be detected by the os. As I wrote, I tryed to get this number of 64 cores enabled, 32 physical and 32 from Hyperthreading, setting in bios 32 cores activated and hyperthreading activated, I got il the system only 32 cores recognized, like cores provided by hyperthreading weren’t recognized ; but if I boot the computer with all cores and hyperthreading activated (88) I get Kernel panic.
So remaining 2 questions :
-why hyperthreading is recognised by acpiplatform kext ( this one that causing KP because of the number of core limit to enable is break
-what other thing we have to patch in Kernel to remove this limit of 64 core ?
May be Pike could have an idea ?
Yes, old cpus=xx parameters it is working and can bypass bios values
https://postimg.org/image/wotjy8tal/
have you tried to put it to 88?
Hi Fabio,
I didn’t try yet, I’m travelling for 2 days. May b be the patch with this option could unlock the trouble…
Maybe Philippe,
It is useful if you have a dual system with windows/linux and OS X.
In this way also if OS X is not working with all cores/threads set in bios you can leave bios with Hyperthreading enabled and with cpus=44 you can force OS X to start ..maybe! 🙂
Hi Fabio, I’m back.
Tryed with the -cpus=88 but stlil kp whith acpiplatform kext…. May something else hav to be patched in kernel…
Hi Philippe,
maybe it does not change anything but right flag is without “-“, so it should be cpus=88
Hi Fabio. I tried cpu=88, still KP. Enabling just 32 core with cpu=64, I got hyperthreading working. For the full 88 cores we have to find something else.
Hi Pikeralpha, what do you think about all this ?
Digging on kernel I see many cpu thread count parameters..maybe Pike could help to try the right one…but maybe it is only impossible to hack! 🙂
edit
EI Philippe In that 64 cores condition your system is stable and working?
thank you
Stable and working, but nvidia driver refuses to load ( I have gtx 980 in my system). with 44 cores, nvidia drives loads. I did’t succeed to have it booting without nullcpupowermanagement.kext. I tryed to generate ssdt table, but with custom ssdt, it crashes after 10 minutes and still needs nullcpupowermanagement.kext.
to be able to boot.
Hi Philippe
Our cpu can work well without nullcpupm in a single cpu environment, take a look here:
http://www.insanelymac.com/forum/topic/316516-macos-sierra-on-broadwell-eep-not-a-guide-but-maybe-a-working-in-progress/page-1
maybe some patches could apply also to your rig
Then, if Pikeralpha have a spar etime to dedicate to 64Vs88 cores problem maybe you could have a pretty perfect and faster Mac ever! 🙂
For you, what is the best one of these three methods ? Is it working with El-capitan ?
El capitan has a different kernel so you have to modify them
But with correct el capitan patches it is working
I prefer solution with pikeralpha reboot fix
But all are working well and rig is stable at all
Hi Fabio,
following your link and post, with some 10.11.6 Pikeralpha’s patches, custom ssdt generated byPikeralpha’s script, I got this :
AppleIntelInfo.kext v2.0 Copyright © 2012-2016 Pike R. Alpha. All rights reserved
enableHWP…………………………..: 0
Settings:
——————————————
logMSRs…………………………….: 1
logIGPU…………………………….: 0
logCStates………………………….: 1
logIPGStyle…………………………: 1
InitialTSC………………………….: 0xc49299927a (35 MHz)
MWAIT C-States………………………: 8480
Processor Brandstring………………..:
Processor Signature………………… : 0x406F1
——————————————
– Family…………………………. : 6
– Stepping……………………….. : 1
– Model………………………….. : 0x4F (79)
Model Specific Registers (MSRs)
——————————————
MSR_CORE_THREAD_COUNT…………(0x35) : 0x0
——————————————
– Core Count……………………… : 22
– Thread Count……………………. : 22
MSR_PLATFORM_INFO…………….(0xCE) : 0x20080C3BFB811800
——————————————
– Maximum Non-Turbo Ratio………….. : 0x18 (2400 MHz)
– Ratio Limit for Turbo Mode……….. : 1 (programmable)
– TDP Limit for Turbo Mode…………. : 1 (programmable)
– Low Power Mode Support…………… : 1 (LPM supported)
– Number of ConfigTDP Levels……….. : 1 (additional TDP level(s) available)
– Maximum Efficiency Ratio…………. : 12
– Minimum Operating Ratio………….. : 8
MSR_PMG_CST_CONFIG_CONTROL…….(0xE2) : 0x8403
——————————————
– I/O MWAIT Redirection Enable……… : 1 (enabled, IO read of MSR(0xE4) mapped to MWAIT)
– CFG Lock……………………….. : 1 (MSR locked until next reset)
– C3 State Auto Demotion…………… : 0 (disabled/unsupported)
– C1 State Auto Demotion…………… : 0 (disabled/unsupported)
– C3 State Undemotion……………… : 0 (disabled/unsupported)
– C1 State Undemotion……………… : 0 (disabled/unsupported)
– Package C-State Auto Demotion…….. : 0 (disabled/unsupported)
– Package C-State Undemotion……….. : 0 (disabled/unsupported)
MSR_PMG_IO_CAPTURE_BASE……….(0xE4) : 0x10414
——————————————
– LVL_2 Base Address………………. : 0x414
– C-state Range…………………… : 1 (C6 is the max C-State to include)
IA32_MPERF…………………..(0xE7) : 0x5F4678D840
IA32_APERF…………………..(0xE8) : 0x215D23E60A
MSR_0x150……………………(0x150) : 0x0
MSR_FLEX_RATIO……………….(0x194) : 0x0
——————————————
MSR_IA32_PERF_STATUS………….(0x198) : 0x209A00001E00
——————————————
– Current Performance State Value…… : 0x1E00 (3000 MHz)
MSR_IA32_PERF_CONTROL…………(0x199) : 0x2500
——————————————
– Target performance State Value……. : 0x2500 (3700 MHz)
– Intel Dynamic Acceleration……….. : 0 (IDA engaged)
IA32_CLOCK_MODULATION…………(0x19A) : 0x0
IA32_THERM_INTERRUPT………….(0x19B) : 0x0
IA32_THERM_STATUS…………….(0x19C) : 0x882C0000
——————————————
– Thermal Status………………….. : 0
– Thermal Log…………………….. : 0
– PROCHOT # or FORCEPR# event………. : 0
– PROCHOT # or FORCEPR# log………… : 0
– Critical Temperature Status………. : 0
– Critical Temperature log…………. : 0
– Thermal Threshold #1 Status………. : 0
– Thermal Threshold #1 log…………. : 0
– Thermal Threshold #2 Status………. : 0
– Thermal Threshold #2 log…………. : 0
– Power Limitation Status………….. : 0
– Power Limitation log…………….. : 0
– Current Limit Status…………….. : 0
– Current Limit log……………….. : 0
– Cross Domain Limit Status………… : 0
– Cross Domain Limit log…………… : 0
– Digital Readout…………………. : 44
– Resolution in Degrees Celsius…….. : 1
– Reading Valid…………………… : 1 (valid)
MSR_THERM2_CTL……………….(0x19D) : 0x0
IA32_MISC_ENABLES…………….(0x1A0) : 0x850089
——————————————
– Fast-Strings……………………. : 1 (enabled)
– FOPCODE compatibility mode Enable…. : 0
– Automatic Thermal Control Circuit…. : 1 (enabled)
– Split-lock Disable………………. : 0
– Performance Monitoring…………… : 1 (available)
– Bus Lock On Cache Line Splits Disable : 0
– Hardware prefetch Disable………… : 0
– Processor Event Based Sampling……. : 0 (PEBS supported)
– GV1/2 legacy Enable……………… : 0
– Enhanced Intel SpeedStep Technology.. : 1 (enabled)
– MONITOR FSM…………………….. : 1 (MONITOR/MWAIT supported)
– Adjacent sector prefetch Disable….. : 0
– CFG Lock……………………….. : 0 (MSR not locked)
– xTPR Message Disable…………….. : 1 (disabled)
MSR_TEMPERATURE_TARGET………..(0x1A2) : 0x4E0A00
——————————————
– Turbo Attenuation Units………….. : 0
– Temperature Target………………. : 78
– TCC Activation Offset……………. : 0
MSR_MISC_PWR_MGMT…………….(0x1AA) : 0x402000
——————————————
– EIST Hardware Coordination……….. : 0 (hardware coordination enabled)
– Energy/Performance Bias support…… : 1
– Energy/Performance Bias………….. : 0 (disabled/MSR not visible to software)
– Thermal Interrupt Coordination Enable : 1 (thermal interrupt routed to all cores)
MSR_TURBO_RATIO_LIMIT…………(0x1AD) : 0x1C1D1E1F20212323
——————————————
– Maximum Ratio Limit for C01………. : 23 (3500 MHz)
– Maximum Ratio Limit for C02………. : 23 (3500 MHz)
– Maximum Ratio Limit for C03………. : 21 (3300 MHz)
– Maximum Ratio Limit for C04………. : 20 (3200 MHz)
– Maximum Ratio Limit for C05………. : 1F (3100 MHz)
– Maximum Ratio Limit for C06………. : 1E (3000 MHz)
– Maximum Ratio Limit for C07………. : 1D (2900 MHz)
– Maximum Ratio Limit for C08………. : 1C (2800 MHz)
MSR_TURBO_RATIO_LIMIT1………..(0x1AE) : 0x1C1C1C1C1C1C1C1C
——————————————
– Maximum Ratio Limit for C09………. : 1C (2800 MHz)
– Maximum Ratio Limit for C10………. : 1C (2800 MHz)
– Maximum Ratio Limit for C11………. : 1C (2800 MHz)
– Maximum Ratio Limit for C12………. : 1C (2800 MHz)
– Maximum Ratio Limit for C13………. : 1C (2800 MHz)
– Maximum Ratio Limit for C14………. : 1C (2800 MHz)
– Maximum Ratio Limit for C15………. : 1C (2800 MHz)
– Maximum Ratio Limit for C16………. : 1C (2800 MHz)
MSR_TURBO_RATIO_LIMIT2………..(0x1AF) : 0x1C1C1C1C1C1C1C1C
——————————————
– Maximum Ratio Limit for C17………. : 1C (2800 MHz)
– Maximum Ratio Limit for C18………. : 1C (2800 MHz)
– Maximum Ratio Limit for C19………. : 1C (2800 MHz)
– Maximum Ratio Limit for C20………. : 1C (2800 MHz)
– Maximum Ratio Limit for C21………. : 1C (2800 MHz)
– Maximum Ratio Limit for C22………. : 1C (2800 MHz)
IA32_ENERGY_PERF_BIAS…………(0x1B0) : 0x0
MSR_POWER_CTL………………..(0x1FC) : 0x2904005B
——————————————
– Bi-Directional Processor Hot……….: 1 (enabled)
– C1E Enable……………………….: 1 (enabled)
MSR_RAPL_POWER_UNIT…………..(0x606) : 0xA0E03
——————————————
– Power Units…………………….. : 3 (1/8 Watt)
– Energy Status Units……………… : 14 (61 micro-Joules)
– Time Units …………………….. : 10 (976.6 micro-Seconds)
MSR_PKG_POWER_LIMIT…………..(0x610) : 0x785A0001584B0
——————————————
– Package Power Limit #1…………… : 150 Watt
– Enable Power Limit #1……………. : 1 (enabled)
– Package Clamping Limitation #1……. : 1 (allow going below OS-requested P/T state during Time Window for Power Limit #1)
– Time Window for Power Limit #1……. : 10 (2560 milli-Seconds)
– Package Power Limit #2…………… : 180 Watt
– Enable Power Limit #2……………. : 1 (enabled)
– Package Clamping Limitation #2……. : 1 (allow going below OS-requested P/T state setting Time Window for Power Limit #2)
– Time Window for Power Limit #2……. : 3 (20 milli-Seconds)
– Lock…………………………… : 0 (MSR not locked)
MSR_PKG_ENERGY_STATUS…………(0x611) : 0x118C91D0
——————————————
– Total Energy Consumed……………. : 17970 Joules (Watt = Joules / seconds)
MSR_PKG_POWER_INFO……………(0x614) : 0x2F0A18034804B0
——————————————
– Thermal Spec Power………………. : 150 Watt
– Minimum Power…………………… : 0
– Maximum Power…………………… : 323
– Maximum Time Window……………… : 0
MSR_PP0_POWER_LIMIT…………..(0x638) : 0x0
MSR_PP0_ENERGY_STATUS…………(0x639) : 0x0
MSR_TURBO_ACTIVATION_RATIO…….(0x64C) : 0x0
MSR_PKGC3_IRTL……………….(0x60a) : 0x0
MSR_PKGC6_IRTL……………….(0x60b) : 0x0
MSR_PKG_C2_RESIDENCY………….(0x60d) : 0x1AFD50F208
MSR_PKG_C3_RESIDENCY………….(0x3f8) : 0x0
MSR_PKG_C6_RESIDENCY………….(0x3f9) : 0x8184A7F58
IA32_TSC_DEADLINE…………….(0x6E0) : 0xC4A5B40B5E
CPU Ratio Info:
——————————————
Base Clock Frequency (BLCK)…………. : 100 MHz
Maximum Efficiency Ratio/Frequency…….: 12 (1200 MHz)
Maximum non-Turbo Ratio/Frequency……..: 24 (2400 MHz)
Maximum Turbo Ratio/Frequency…………: 35 (3500 MHz)
P-State ratio * 100 = Frequency in MHz
——————————————
CPU P-States [ 27 (28) 33 ]
CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 ]
CPU P-States [ 26 27 28 (32) 33 ]
CPU P-States [ 26 27 28 32 33 (35) ]
CPU P-States [ 26 27 28 32 33 (34) 35 ]
CPU P-States [ 26 27 28 (30) 32 33 34 35 ]
CPU P-States [ 26 27 28 (29) 30 32 33 34 35 ]
CPU P-States [ 26 27 28 29 30 31 (32) 33 34 35 ]
CPU P-States [ (12) 26 27 28 29 30 31 32 33 34 35 ]
CPU P-States [ 12 20 26 27 28 29 30 31 32 33 34 (35) ]
CPU P-States [ (12) 20 25 26 27 28 29 30 31 32 33 34 35 ]
CPU P-States [ 12 20 24 25 26 27 28 29 30 31 32 33 34 (35) ]
CPU P-States [ 12 20 23 24 25 26 27 28 29 30 31 32 33 34 (35) ]
CPU P-States [ (12) 20 21 23 24 25 26 27 28 29 30 31 32 33 34 35 ]
But dispute of the fact I use your patch to get max turbo at 2700 Mhz, I succeed just one time to get max turbo at 2700 Mhz instead of 2500 Mhz, and stil not able to reproduce one more time
edit (add my Appleintelinfo output:
fabios-Mac-Pro:desktop fabio$ sudo cat /tmp/AppleIntelInfo.dat
AppleIntelInfo.kext v2.0 Copyright © 2012-2016 Pike R. Alpha. All rights reserved
enableHWP................................: 0
Settings:
------------------------------------------
logMSRs..................................: 1
logIGPU..................................: 0
logCStates...............................: 1
logIPGStyle..............................: 1
InitialTSC...............................: 0x1edb64ca384 (96 MHz)
MWAIT C-States...........................: 8480
Processor Brandstring....................: Intel(R) Xeon(R) CPU E5-2696 v4 @ 2.20GHz
Processor Signature..................... : 0x406F1
------------------------------------------
- Family............................... : 6
- Stepping............................. : 1
- Model................................ : 0x4F (79)
Model Specific Registers (MSRs)
------------------------------------------
MSR_CORE_THREAD_COUNT............(0x35) : 0xFFFFFF8110015500
------------------------------------------
- Core Count........................... : 22
- Thread Count......................... : 44
MSR_PLATFORM_INFO................(0xCE) : 0x20080C3BF2811600
------------------------------------------
- Maximum Non-Turbo Ratio.............. : 0x16 (2200 MHz)
- Ratio Limit for Turbo Mode........... : 1 (programmable)
- TDP Limit for Turbo Mode............. : 1 (programmable)
- Low Power Mode Support............... : 1 (LPM supported)
- Number of ConfigTDP Levels........... : 1 (additional TDP level(s) available)
- Maximum Efficiency Ratio............. : 12
- Minimum Operating Ratio.............. : 8
MSR_PMG_CST_CONFIG_CONTROL.......(0xE2) : 0x402
------------------------------------------
- I/O MWAIT Redirection Enable......... : 1 (enabled, IO read of MSR(0xE4) mapped to MWAIT)
- CFG Lock............................. : 0 (MSR not locked)
- C3 State Auto Demotion............... : 0 (disabled/unsupported)
- C1 State Auto Demotion............... : 0 (disabled/unsupported)
- C3 State Undemotion.................. : 0 (disabled/unsupported)
- C1 State Undemotion.................. : 0 (disabled/unsupported)
- Package C-State Auto Demotion........ : 0 (disabled/unsupported)
- Package C-State Undemotion........... : 0 (disabled/unsupported)
MSR_PMG_IO_CAPTURE_BASE..........(0xE4) : 0x10414
------------------------------------------
- LVL_2 Base Address................... : 0x414
- C-state Range........................ : 1 (C6 is the max C-State to include)
IA32_MPERF.......................(0xE7) : 0x4A31EB8AAF
IA32_APERF.......................(0xE8) : 0x4B3351B581
MSR_0x150........................(0x150) : 0x1F00000000
MSR_FLEX_RATIO...................(0x194) : 0x0
------------------------------------------
MSR_IA32_PERF_STATUS.............(0x198) : 0x263B00002500
------------------------------------------
- Current Performance State Value...... : 0x2500 (3700 MHz)
MSR_IA32_PERF_CONTROL............(0x199) : 0x2500
------------------------------------------
- Target performance State Value....... : 0x2500 (3700 MHz)
- Intel Dynamic Acceleration........... : 0 (IDA engaged)
IA32_CLOCK_MODULATION............(0x19A) : 0x0
IA32_THERM_INTERRUPT.............(0x19B) : 0x0
IA32_THERM_STATUS................(0x19C) : 0x88500000
------------------------------------------
- Thermal Status....................... : 0
- Thermal Log.......................... : 0
- PROCHOT # or FORCEPR# event.......... : 0
- PROCHOT # or FORCEPR# log............ : 0
- Critical Temperature Status.......... : 0
- Critical Temperature log............. : 0
- Thermal Threshold #1 Status.......... : 0
- Thermal Threshold #1 log............. : 0
- Thermal Threshold #2 Status.......... : 0
- Thermal Threshold #2 log............. : 0
- Power Limitation Status.............. : 0
- Power Limitation log................. : 0
- Current Limit Status................. : 0
- Current Limit log.................... : 0
- Cross Domain Limit Status............ : 0
- Cross Domain Limit log............... : 0
- Digital Readout...................... : 80
- Resolution in Degrees Celsius........ : 1
- Reading Valid........................ : 1 (valid)
MSR_THERM2_CTL...................(0x19D) : 0x0
IA32_MISC_ENABLES................(0x1A0) : 0x850089
------------------------------------------
- Fast-Strings......................... : 1 (enabled)
- FOPCODE compatibility mode Enable.... : 0
- Automatic Thermal Control Circuit.... : 1 (enabled)
- Split-lock Disable................... : 0
- Performance Monitoring............... : 1 (available)
- Bus Lock On Cache Line Splits Disable : 0
- Hardware prefetch Disable............ : 0
- Processor Event Based Sampling....... : 0 (PEBS supported)
- GV1/2 legacy Enable.................. : 0
- Enhanced Intel SpeedStep Technology.. : 1 (enabled)
- MONITOR FSM.......................... : 1 (MONITOR/MWAIT supported)
- Adjacent sector prefetch Disable..... : 0
- CFG Lock............................. : 0 (MSR not locked)
- xTPR Message Disable................. : 1 (disabled)
MSR_TEMPERATURE_TARGET...........(0x1A2) : 0x640200
------------------------------------------
- Turbo Attenuation Units.............. : 0
- Temperature Target................... : 100
- TCC Activation Offset................ : 0
MSR_MISC_PWR_MGMT................(0x1AA) : 0x402000
------------------------------------------
- EIST Hardware Coordination........... : 0 (hardware coordination enabled)
- Energy/Performance Bias support...... : 1
- Energy/Performance Bias.............. : 0 (disabled/MSR not visible to software)
- Thermal Interrupt Coordination Enable : 1 (thermal interrupt routed to all cores)
MSR_TURBO_RATIO_LIMIT............(0x1AD) : 0x1E1F202122232525
------------------------------------------
- Maximum Ratio Limit for C01.......... : 25 (3700 MHz)
- Maximum Ratio Limit for C02.......... : 25 (3700 MHz)
- Maximum Ratio Limit for C03.......... : 23 (3500 MHz)
- Maximum Ratio Limit for C04.......... : 22 (3400 MHz)
- Maximum Ratio Limit for C05.......... : 21 (3300 MHz)
- Maximum Ratio Limit for C06.......... : 20 (3200 MHz)
- Maximum Ratio Limit for C07.......... : 1F (3100 MHz)
- Maximum Ratio Limit for C08.......... : 1E (3000 MHz)
MSR_TURBO_RATIO_LIMIT1...........(0x1AE) : 0x1C1C1C1C1C1C1C1D
------------------------------------------
- Maximum Ratio Limit for C09.......... : 1D (2900 MHz)
- Maximum Ratio Limit for C10.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C11.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C12.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C13.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C14.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C15.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C16.......... : 1C (2800 MHz)
MSR_TURBO_RATIO_LIMIT2...........(0x1AF) : 0x1C1C1C1C1C1C1C1C
------------------------------------------
- Maximum Ratio Limit for C17.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C18.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C19.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C20.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C21.......... : 1C (2800 MHz)
- Maximum Ratio Limit for C22.......... : 1C (2800 MHz)
IA32_ENERGY_PERF_BIAS............(0x1B0) : 0x0
MSR_POWER_CTL....................(0x1FC) : 0x2904005B
------------------------------------------
- Bi-Directional Processor Hot..........: 1 (enabled)
- C1E Enable............................: 1 (enabled)
MSR_RAPL_POWER_UNIT..............(0x606) : 0xA0E03
------------------------------------------
- Power Units.......................... : 3 (1/8 Watt)
- Energy Status Units.................. : 14 (61 micro-Joules)
- Time Units .......................... : 10 (976.6 micro-Seconds)
MSR_PKG_POWER_LIMIT..............(0x610) : 0x785A0001584B0
------------------------------------------
- Package Power Limit #1............... : 150 Watt
- Enable Power Limit #1................ : 1 (enabled)
- Package Clamping Limitation #1....... : 1 (allow going below OS-requested P/T state during Time Window for Power Limit #1)
- Time Window for Power Limit #1....... : 10 (2560 milli-Seconds)
- Package Power Limit #2............... : 180 Watt
- Enable Power Limit #2................ : 1 (enabled)
- Package Clamping Limitation #2....... : 1 (allow going below OS-requested P/T state setting Time Window for Power Limit #2)
- Time Window for Power Limit #2....... : 3 (20 milli-Seconds)
- Lock................................. : 0 (MSR not locked)
MSR_PKG_ENERGY_STATUS............(0x611) : 0x10D7F3D
------------------------------------------
- Total Energy Consumed................ : 1077 Joules (Watt = Joules / seconds)
MSR_PKG_POWER_INFO...............(0x614) : 0x2F0A18028804B0
------------------------------------------
- Thermal Spec Power................... : 150 Watt
- Minimum Power........................ : 0
- Maximum Power........................ : 323
- Maximum Time Window.................. : 0
MSR_PP0_POWER_LIMIT..............(0x638) : 0x0
MSR_PP0_ENERGY_STATUS............(0x639) : 0x0
MSR_TURBO_ACTIVATION_RATIO.......(0x64C) : 0x0
MSR_PKGC3_IRTL...................(0x60a) : 0x0
MSR_PKGC6_IRTL...................(0x60b) : 0x0
MSR_PKG_C2_RESIDENCY.............(0x60d) : 0x6CCADF4CFA
MSR_PKG_C3_RESIDENCY.............(0x3f8) : 0x0
MSR_PKG_C6_RESIDENCY.............(0x3f9) : 0x102441FF6AE
IA32_TSC_DEADLINE................(0x6E0) : 0x1EDC720BDBD
CPU Ratio Info:
------------------------------------------
Base Clock Frequency (BLCK)............. : 100 MHz
Maximum Efficiency Ratio/Frequency.......: 12 (1200 MHz)
Maximum non-Turbo Ratio/Frequency........: 22 (2200 MHz)
Maximum Turbo Ratio/Frequency............: 37 (3700 MHz)
P-State ratio * 100 = Frequency in MHz
------------------------------------------
CPU P-States [ 22 35 (37) ]
CPU C6-Cores [ 1 2 4 5 6 7 8 10 11 12 14 17 19 21 24 26 29 31 32 33 34 35 36 37 38 39 40 41 42 43 ]
CPU P-States [ 22 26 35 (37) ]
CPU C6-Cores [ 0 1 2 4 5 6 7 8 10 11 12 13 14 15 16 17 18 19 20 21 22 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 ]
CPU P-States [ 22 25 26 (35) 37 ]
CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 ]
CPU P-States [ 22 25 26 30 35 (37) ]
CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 ]
CPU P-States [ 22 25 26 29 30 (34) 35 37 ]
CPU P-States [ 22 25 26 28 29 30 34 35 (37) ]
CPU P-States [ (12) 22 25 26 28 29 30 32 34 35 37 ]
CPU P-States [ 12 22 24 25 26 28 29 30 32 34 35 (37) ]
CPU P-States [ 12 22 24 25 26 27 28 29 30 32 34 35 (37) ]
CPU P-States [ (12) 22 23 24 25 26 27 28 29 30 32 34 35 37 ]
CPU P-States [ 12 21 22 23 24 25 26 27 28 29 30 32 34 35 (37) ]
CPU P-States [ 12 21 22 23 24 25 26 27 28 29 30 32 33 34 (35) 37 ]
CPU P-States [ 12 19 21 (22) 23 24 25 26 27 28 29 30 32 33 34 35 37 ]
CPU P-States [ (12) 19 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 37 ]
CPU P-States [ (12) 19 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 ]
CPU P-States [ (12) 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 ]
fabios-Mac-Pro:desktop fabio$
you can see many cpu c6 e P states
Hi Fabio,
we can continue on the other forum. My name there is php75
Hi, Pike.
Instead of using these patches you mentioned at https://pikeralpha.wordpress.com/2016/07/26/xcpm-for-unsupported-processor/comment-page-1/#comment-6301 , you can also use ret(urn) at kernel location 0x2283a0 (Darwin 16.0.0, 10.12.0, 16A323). And confirmed, this worked for me, here is the diff:
@@ -409785,7 +409785,7 @@
ffffff8000428392 5d popq %rbp
ffffff8000428393 e9 08 00 00 00 jmp 0xffffff80004283a0
ffffff8000428398 0f 1f 84 00 00 00 00 00 nopl (%rax,%rax)
-ffffff80004283a0 55 pushq %rbp
+ffffff80004283a0 c3 retq
ffffff80004283a1 48 89 e5 movq %rsp, %rbp
ffffff80004283a4 41 57 pushq %r15
ffffff80004283a6 41 56 pushq %r14
Okay. This is made based on your comment, and I saw all verbs you NOP called ffffff80004283a0, so I used ret(urn) for this and works well 🙂
Nothing new. This is my crude reboot patch. If you read my reply then you see that I gave some instruction to locate the specific MSR(s) that trigger the reboot.
Ah I see. Sorry for my ignorance…
Hi Pike,
I’ve enabled XCPM successfully by following your idea. Thanks. But some questions I’d like to ask…
Details:
CPU: Intel Xeon E3 1230 V2
SMBios: MacPro5,1 + MacBookPro 11,5 FrequencyVectors (Use ssdtPRGen.sh to inject LFM = 1200MHz and set “plugin-type” = 1 to load X86*, use freqVectorsEdit.sh to inject MacBookPro 11,5 FrequencyVectors.)
But… I can’t get any P-States lower than x16, here is the output of AppleIntelInfo.kext:
AppleIntelInfo.kext v2.0 Copyright © 2012-2016 Pike R. Alpha. All rights reserved
enableHWP…………………………..: 0
Settings:
——————————————
logMSRs…………………………….: 1
logIGPU…………………………….: 0
logCStates………………………….: 1
logIPGStyle…………………………: 1
InitialTSC………………………….: 0x1923645735c9 (837 MHz)
MWAIT C-States………………………: 4384
Processor Brandstring………………..: Intel(R) Xeon(R) CPU E3-1230 V2 @ 3.30GHz
Processor Signature………………… : 0x306A9
——————————————
– Family…………………………. : 6
– Stepping……………………….. : 9
– Model………………………….. : 0x3A (58)
Model Specific Registers (MSRs)
——————————————
MSR_CORE_THREAD_COUNT…………(0x35) : 0x0
——————————————
– Core Count……………………… : 4
– Thread Count……………………. : 8
MSR_PLATFORM_INFO…………….(0xCE) : 0x81010E0012100
——————————————
– Maximum Non-Turbo Ratio………….. : 0x21 (3300 MHz)
– Ratio Limit for Turbo Mode……….. : 0 (not programmable)
– TDP Limit for Turbo Mode…………. : 1 (programmable)
– Low Power Mode Support…………… : 0 (LMP not supported)
– Number of ConfigTDP Levels……….. : 0 (only base TDP level available)
– Maximum Efficiency Ratio…………. : 16
– Minimum Operating Ratio………….. : 8
MSR_PMG_CST_CONFIG_CONTROL…….(0xE2) : 0x1E000005
——————————————
– I/O MWAIT Redirection Enable……… : 0 (not enabled)
– CFG Lock……………………….. : 0 (MSR not locked)
– C3 State Auto Demotion…………… : 1 (enabled)
– C1 State Auto Demotion…………… : 1 (enabled)
– C3 State Undemotion……………… : 1 (enabled)
– C1 State Undemotion……………… : 1 (enabled)
– Package C-State Auto Demotion…….. : 0 (disabled/unsupported)
– Package C-State Undemotion……….. : 0 (disabled/unsupported)
MSR_PMG_IO_CAPTURE_BASE……….(0xE4) : 0x10414
——————————————
– LVL_2 Base Address………………. : 0x414
– C-state Range…………………… : 1 (C-States not included, I/O MWAIT redirection not enabled)
IA32_MPERF…………………..(0xE7) : 0x3F6EB20777
IA32_APERF…………………..(0xE8) : 0x303AC8CBA5
MSR_FLEX_RATIO……………….(0x194) : 0x0
——————————————
MSR_IA32_PERF_STATUS………….(0x198) : 0x1E6C00001100
——————————————
– Current Performance State Value…… : 0x1100 (1700 MHz)
MSR_IA32_PERF_CONTROL…………(0x199) : 0x2500
——————————————
– Target performance State Value……. : 0x2500 (3700 MHz)
– Intel Dynamic Acceleration……….. : 0 (IDA engaged)
IA32_CLOCK_MODULATION…………(0x19A) : 0x0
IA32_THERM_INTERRUPT………….(0x19B) : 0x10
——————————————
– High-Temperature Interrupt Enable…. : 0 (disabled)
– Low-Temperature Interrupt Enable….. : 0 (disabled)
– PROCHOT# Interrupt Enable………… : 0 (disabled)
– FORCEPR# Interrupt Enable………… : 0 (disabled)
– Critical Temperature Interrupt Enable : 1 (enabled)
– Threshold #1 Value………………. : 0
– Threshold #1 Interrupt Enable…….. : 0 (disabled)
– Threshold #2 Value………………. : 0
– Threshold #2 Interrupt Enable…….. : 0 (disabled)
– Power Limit Notification Enable…… : 0 (disabled)
IA32_THERM_STATUS…………….(0x19C) : 0x88400000
——————————————
– Thermal Status………………….. : 0
– Thermal Log…………………….. : 0
– PROCHOT # or FORCEPR# event………. : 0
– PROCHOT # or FORCEPR# log………… : 0
– Critical Temperature Status………. : 0
– Critical Temperature log…………. : 0
– Thermal Threshold #1 Status………. : 0
– Thermal Threshold #1 log…………. : 0
– Thermal Threshold #2 Status………. : 0
– Thermal Threshold #2 log…………. : 0
– Power Limitation Status………….. : 0
– Power Limitation log…………….. : 0
– Current Limit Status…………….. : 0
– Current Limit log……………….. : 0
– Cross Domain Limit Status………… : 0
– Cross Domain Limit log…………… : 0
– Digital Readout…………………. : 64
– Resolution in Degrees Celsius…….. : 1
– Reading Valid…………………… : 1 (valid)
MSR_THERM2_CTL……………….(0x19D) : 0x0
IA32_MISC_ENABLES…………….(0x1A0) : 0x850089
——————————————
– Fast-Strings……………………. : 1 (enabled)
– FOPCODE compatibility mode Enable…. : 0
– Automatic Thermal Control Circuit…. : 1 (enabled)
– Split-lock Disable………………. : 0
– Performance Monitoring…………… : 1 (available)
– Bus Lock On Cache Line Splits Disable : 0
– Hardware prefetch Disable………… : 0
– Processor Event Based Sampling……. : 0 (PEBS supported)
– GV1/2 legacy Enable……………… : 0
– Enhanced Intel SpeedStep Technology.. : 1 (enabled)
– MONITOR FSM…………………….. : 1 (MONITOR/MWAIT supported)
– Adjacent sector prefetch Disable….. : 0
– CFG Lock……………………….. : 0 (MSR not locked)
– xTPR Message Disable…………….. : 1 (disabled)
MSR_TEMPERATURE_TARGET………..(0x1A2) : 0x691400
——————————————
– Turbo Attenuation Units………….. : 0
– Temperature Target………………. : 105
– TCC Activation Offset……………. : 0
MSR_MISC_PWR_MGMT…………….(0x1AA) : 0x400000
——————————————
– EIST Hardware Coordination……….. : 0 (hardware coordination enabled)
– Energy/Performance Bias support…… : 1
– Energy/Performance Bias………….. : 0 (disabled/MSR not visible to software)
– Thermal Interrupt Coordination Enable : 1 (thermal interrupt routed to all cores)
MSR_TURBO_RATIO_LIMIT…………(0x1AD) : 0x23242525
——————————————
– Maximum Ratio Limit for C01………. : 25 (3700 MHz)
– Maximum Ratio Limit for C02………. : 25 (3700 MHz)
– Maximum Ratio Limit for C03………. : 24 (3600 MHz)
– Maximum Ratio Limit for C04………. : 23 (3500 MHz)
IA32_ENERGY_PERF_BIAS…………(0x1B0) : 0x5
——————————————
– Power Policy Preference……………: 5 (balanced performance and energy saving)
MSR_POWER_CTL………………..(0x1FC) : 0x14005F
——————————————
– Bi-Directional Processor Hot……….: 1 (enabled)
– C1E Enable……………………….: 1 (enabled)
MSR_RAPL_POWER_UNIT…………..(0x606) : 0xA1003
——————————————
– Power Units…………………….. : 3 (1/8 Watt)
– Energy Status Units……………… : 16 (15.3 micro-Joules)
– Time Units …………………….. : 10 (976.6 micro-Seconds)
MSR_PKG_POWER_LIMIT…………..(0x610) : 0x1AA580001A9F40
——————————————
– Package Power Limit #1…………… : 1000 Watt
– Enable Power Limit #1……………. : 1 (enabled)
– Package Clamping Limitation #1……. : 0 (disabled)
– Time Window for Power Limit #1……. : 13 (20480 milli-Seconds)
– Package Power Limit #2…………… : 1200 Watt
– Enable Power Limit #2……………. : 1 (enabled)
– Package Clamping Limitation #2……. : 0 (disabled)
– Time Window for Power Limit #2……. : 13 (20480 milli-Seconds)
– Lock…………………………… : 0 (MSR not locked)
MSR_PKG_ENERGY_STATUS…………(0x611) : 0xF7EB98FF
——————————————
– Total Energy Consumed……………. : 63467 Joules (Watt = Joules / seconds)
MSR_PKG_POWER_INFO……………(0x614) : 0xD000000600228
——————————————
– Thermal Spec Power………………. : 69 Watt
– Minimum Power…………………… : 0
– Maximum Power…………………… : 0
– Maximum Time Window……………… : 0
MSR_PP0_POWER_LIMIT…………..(0x638) : 0x1AA580
——————————————
– Power Limit…………………….. : 1200 Watt
– Enable Power Limit………………. : 1 (enabled)
– Clamping Limitation……………… : 0 (disabled)
– Time Window for Power Limit………. : 13 (81920 milli-Seconds)
– Lock…………………………… : 0 (MSR not locked)
MSR_PP0_ENERGY_STATUS…………(0x639) : 0x2E94CEA8
——————————————
– Total Energy Consumed……………. : 11924 Joules (Watt = Joules / seconds)
MSR_TURBO_ACTIVATION_RATIO…….(0x64C) : 0x0
MSR_CONFIG_TDP_NOMINAL………..(0x648) : 0x21
MSR_CONFIG_TDP_LEVEL1…………(0x649) : 0x60000000000000
MSR_CONFIG_TDP_LEVEL2…………(0x64a) : 0x60000000000000
MSR_CONFIG_TDP_CONTROL………..(0x64b) : 0x80000000
MSR_TURBO_ACTIVATION_RATIO…….(0x64c) : 0x0
MSR_PKGC3_IRTL……………….(0x60a) : 0x883B
MSR_PKGC6_IRTL……………….(0x60b) : 0x8850
MSR_PKGC7_IRTL……………….(0x60c) : 0x8857
MSR_PKG_C2_RESIDENCY………….(0x60d) : 0x173E022F205
MSR_PKG_C3_RESIDENCY………….(0x3f8) : 0xA9C023416D
MSR_PKG_C6_RESIDENCY………….(0x3f9) : 0x54B5BACDC0D
MSR_PKG_C7_RESIDENCY………….(0x3fa) : 0x0
IA32_TSC_DEADLINE…………….(0x6E0) : 0x19236F6185EB
CPU Ratio Info:
——————————————
Base Clock Frequency (BLCK)…………. : 100 MHz
Maximum Efficiency Ratio/Frequency…….: 16 (1600 MHz)
Maximum non-Turbo Ratio/Frequency……..: 33 (3300 MHz)
Maximum Turbo Ratio/Frequency…………: 37 (3700 MHz)
P-State ratio * 100 = Frequency in MHz
——————————————
CPU P-States [ (17) 22 36 ]
CPU C3-Cores [ 4 6 7 ]
CPU C6-Cores [ 2 3 4 5 6 7 ]
CPU C7-Cores [ 0 1 4 6 7 ]
CPU P-States [ (16) 17 22 25 36 ]
CPU C3-Cores [ 0 4 5 6 7 ]
CPU C6-Cores [ 0 2 3 4 5 6 7 ]
CPU C7-Cores [ 0 1 2 3 4 5 6 7 ]
CPU P-States [ 16 (17) 22 25 27 36 ]
CPU C3-Cores [ 0 1 4 5 6 7 ]
CPU C6-Cores [ 0 1 2 3 4 5 6 7 ]
CPU P-States [ 16 (17) 21 22 25 27 36 ]
CPU C3-Cores [ 0 1 2 3 4 5 6 7 ]
CPU P-States [ 16 17 21 22 25 27 (35) 36 ]
CPU P-States [ (16) 17 21 22 23 25 27 35 36 ]
CPU P-States [ 16 (17) 21 22 23 25 26 27 35 36 ]
CPU P-States [ 16 17 21 22 23 24 25 26 27 35 36 (37) ]
CPU P-States [ 16 17 21 22 23 24 25 26 27 29 35 (36) 37 ]
CPU P-States [ 16 17 21 22 23 24 25 26 27 29 33 35 (36) 37 ]
CPU P-States [ 16 17 21 22 23 24 25 26 27 29 30 33 35 (36) 37 ]
CPU P-States [ 16 17 21 22 23 24 25 26 27 29 30 31 33 (35) 36 37 ]
CPU P-States [ 16 17 21 22 23 24 25 26 27 29 30 31 32 33 35 (36) 37 ]
CPU P-States [ 16 17 21 22 23 24 25 26 27 29 30 31 32 33 (34) 35 36 37 ]
CPU P-States [ (16) 17 19 21 22 23 24 25 26 27 29 30 31 32 33 34 35 36 37 ]
CPU P-States [ (16) 17 18 19 21 22 23 24 25 26 27 29 30 31 32 33 34 35 36 37 ]
CPU P-States [ 16 (17) 18 19 21 22 23 24 25 26 27 28 29 30 31 32 33 34 35 36 37 ]
19 20 21 23 24 25 26 27 28 29 30 31 32 33 (34) 35 36 37 ]
CPU P-States [ 16 17 18 19 20 21 22 23 24 25 26 27 28 29 30 31 32 33 (34) 35 36 37 ]
And how to fix LFM to 1200MHz or lower such as 800MHz? Thanks.
Cheers,
Angel
The ASUS Z10PE absolutely loves Sierra only needing FakeSMC to get to the installer … It’s thereafter the trouble starts. No Haswell-E, Broadwell-E or IvyBridge-E support, along with an old Fermi based 660Ti has me bumping into a brick wall. MP6,1 killed by GFX1 with non of the usual fixes working for me – iMac14,2 works but then there are no Frequency Vectors for me to use with my v4 Xeons and of course no iMac ever came with dual processors so I was never comfortable with it. MP5,1 works really well, but then from my understanding the MP5,1 has 0 chance of supporting XCPM, is this true ? …. At least I can use a Quadro graphics card 🙂
Pike, in macOS Sierra 10.12.2 Beta (16C60b) searching in kernel I have found this:
_ml_init_max_cpus:
ffffff80003ffd90 55 pushq %rbp
ffffff80003ffd91 48 89 e5 movq %rsp, %rbp
ffffff80003ffd94 41 56 pushq %r14
ffffff80003ffd96 53 pushq %rbx
ffffff80003ffd97 48 83 ec 10 subq $0x10, %rsp
ffffff80003ffd9b 9c pushfq
ffffff80003ffd9c 5b popq %rbx
ffffff80003ffd9d 48 81 e3 00 02 00 00 andq $0x200, %rbx
ffffff80003ffda4 74 01 je 0xffffff80003ffda7
ffffff80003ffda6 fa cli
ffffff80003ffda7 8b 05 0b 75 6f 00 movl 0x6f750b(%rip), %eax
ffffff80003ffdad 83 f8 01 cmpl $0x1, %eax
ffffff80003ffdb0 74 56 je 0xffffff80003ffe08
ffffff80003ffdb2 48 8d 4f ff leaq -0x1(%rdi), %rcx
ffffff80003ffdb6 48 83 f9 3f cmpq $0x3f, %rcx
ffffff80003ffdba 77 12 ja 0xffffff80003ffdce
ffffff80003ffdbc 8b 0d 02 3b 61 00 movl _max_ncpus(%rip), %ecx
what is this?
ffffff80003ffdbc 8b 0d 02 3b 61 00 movl _max_ncpus(%rip), %ecx
Could be useful for people for dual cpu system with more of 64 cores?
your reboot fix patch and others are still working in this beta
Thank you
@Slice
Do you have a idea how I can use xcpm for ivy bridge on Sierra?
Because it seems apple removed the -xcpm flag.
Any ideas?
Sorry @ Pike 😉
I already tried different ssdt with ssdtprgen with xcpm set to 1.
On 10.11 there was no problem with the -xcpm flag.
Do you have a idea?
Thx to your work I did it…..
My Ivy Bridge is again working on Sierra.
Just edited the xcpm bootstrap.
And the reboot fix need also to be used.
Good job!
Why is it that enabling xcpm results in lower benchmarks/performance? I’m getting 15k vs 20k geekbench 4 on an i7-6800k @ 4.2 GHz using the method described here > https://www.tonymacx86.com/threads/guide-full-power-management-w-dual-single-broadwell-ep-haswell-ep-cpus-on-10-12.204468/
no official Broadwell-e in macs probably
@Ronald, you provably didn’t adjust your clock speed when you generated your SSDT. If you have overclocked your CPU, you need to tell ssdtprgen about the higher frequency, otherwise it will generate pstates for the stock clock speed of your CPU and you will effectively lose your overclock, but only in OSX. Just regenerate your SSDT with the correct frequency and that should restore your performance.
Yes Pike or it seems so reading binaries with an exadecimal editor app
initCPU string but maybe it is unrelated to maxCPU limit
Hi. Thank you so much for all the work.
AppleIntelInfo.kext is crashing my computer (Sierra 10.12.3). I followed the steps for XCPM in this thread: https://www.tonymacx86.com/threads/guide-full-power-management-w-dual-single-broadwell-ep-haswell-ep-cpus-on-10-12.204468/
I have a e5-2630v3. I am having a really hard time. My CPU says unknown in “About this mac” and I noticed the words “legacy shim” as I was booting. I tried patching vectors and everything, doesn’t seem like any power management is working at all.
Any suggestions please?
-> Github issues.
Downloaded beta versions of AppleIntelInfo and compiled myself, and it’s working.
Still no XCPM, and Intel Power Gadget won’t install because it says my CPU is unsupported. In About This Mac it says “Unknown”.
AppleIntelInfo gave this output:
AppleIntelInfo.kext v2.0 Copyright © 2012-2016 Pike R. Alpha. All rights reserved
enableHWP…………………………..: 0
Settings:
——————————————
logMSRs…………………………….: 1
logIGPU…………………………….: 0
logCStates………………………….: 1
logIPGStyle…………………………: 1
Warning: Clover hw.busfrequency error detected : 17d78400
InitialTSC………………………….: 0x2563339160ca (1712 MHz)
MWAIT C-States………………………: 8480
Processor Brandstring………………..: Intel(R) Xeon(R) CPU E5-2630 v3 @ 2.40GHz
Processor Signature………………… : 0x306F2
——————————————
– Family…………………………. : 6
– Stepping……………………….. : 2
– Model………………………….. : 0x3F (63)
Model Specific Registers (MSRs)
——————————————
MSR_CORE_THREAD_COUNT…………(0x35) : 0xFFFFFF8074B22100
——————————————
– Core Count……………………… : 8
– Thread Count……………………. : 16
MSR_PLATFORM_INFO…………….(0xCE) : 0x80C33F3811800
——————————————
– Maximum Non-Turbo Ratio………….. : 0x18 (2400 MHz)
– Ratio Limit for Turbo Mode……….. : 1 (programmable)
– TDP Limit for Turbo Mode…………. : 1 (programmable)
– Low Power Mode Support…………… : 1 (LPM supported)
– Number of ConfigTDP Levels……….. : 1 (additional TDP level(s) available)
– Maximum Efficiency Ratio…………. : 12
– Minimum Operating Ratio………….. : 8
MSR_PMG_CST_CONFIG_CONTROL…….(0xE2) : 0x8400
——————————————
– I/O MWAIT Redirection Enable……… : 1 (enabled, IO read of MSR(0xE4) mapped to MWAIT)
– CFG Lock……………………….. : 1 (MSR locked until next reset)
– C3 State Auto Demotion…………… : 0 (disabled/unsupported)
– C1 State Auto Demotion…………… : 0 (disabled/unsupported)
– C3 State Undemotion……………… : 0 (disabled/unsupported)
– C1 State Undemotion……………… : 0 (disabled/unsupported)
– Package C-State Auto Demotion…….. : 0 (disabled/unsupported)
– Package C-State Undemotion……….. : 0 (disabled/unsupported)
MSR_PMG_IO_CAPTURE_BASE……….(0xE4) : 0x10414
——————————————
– LVL_2 Base Address………………. : 0x414
– C-state Range…………………… : 1 (C6 is the max C-State to include)
IA32_MPERF…………………..(0xE7) : 0x12541C4B88
IA32_APERF…………………..(0xE8) : 0x12405ACD03
MSR_FLEX_RATIO……………….(0x194) : 0x0
——————————————
MSR_IA32_PERF_STATUS………….(0x198) : 0x1E5800001E00
——————————————
– Current Performance State Value…… : 0x1E00 (3000 MHz)
MSR_IA32_PERF_CONTROL…………(0x199) : 0x2000
——————————————
– Target performance State Value……. : 0x2000 (3200 MHz)
– Intel Dynamic Acceleration……….. : 0 (IDA engaged)
IA32_CLOCK_MODULATION…………(0x19A) : 0x0
IA32_THERM_INTERRUPT………….(0x19B) : 0x0
IA32_THERM_STATUS…………….(0x19C) : 0x88350000
——————————————
– Thermal Status………………….. : 0
– Thermal Log…………………….. : 0
– PROCHOT # or FORCEPR# event………. : 0
– PROCHOT # or FORCEPR# log………… : 0
– Critical Temperature Status………. : 0
– Critical Temperature log…………. : 0
– Thermal Threshold #1 Status………. : 0
– Thermal Threshold #1 log…………. : 0
– Thermal Threshold #2 Status………. : 0
– Thermal Threshold #2 log…………. : 0
– Power Limitation Status………….. : 0
– Power Limitation log…………….. : 0
– Current Limit Status…………….. : 0
– Current Limit log……………….. : 0
– Cross Domain Limit Status………… : 0
– Cross Domain Limit log…………… : 0
– Digital Readout…………………. : 53
– Resolution in Degrees Celsius…….. : 1
– Reading Valid…………………… : 1 (valid)
MSR_THERM2_CTL……………….(0x19D) : 0x0
IA32_MISC_ENABLES…………….(0x1A0) : 0x850089
——————————————
– Fast-Strings……………………. : 1 (enabled)
– FOPCODE compatibility mode Enable…. : 0
– Automatic Thermal Control Circuit…. : 1 (enabled)
– Split-lock Disable………………. : 0
– Performance Monitoring…………… : 1 (available)
– Bus Lock On Cache Line Splits Disable : 0
– Hardware prefetch Disable………… : 0
– Processor Event Based Sampling……. : 0 (PEBS supported)
– GV1/2 legacy Enable……………… : 0
– Enhanced Intel SpeedStep Technology.. : 1 (enabled)
– MONITOR FSM…………………….. : 1 (MONITOR/MWAIT supported)
– Adjacent sector prefetch Disable….. : 0
– CFG Lock……………………….. : 0 (MSR not locked)
– xTPR Message Disable…………….. : 1 (disabled)
MSR_TEMPERATURE_TARGET………..(0x1A2) : 0x570A00
——————————————
– Turbo Attenuation Units………….. : 0
– Temperature Target………………. : 87
– TCC Activation Offset……………. : 0
MSR_MISC_PWR_MGMT…………….(0x1AA) : 0x400000
——————————————
– EIST Hardware Coordination……….. : 0 (hardware coordination enabled)
– Energy/Performance Bias support…… : 1
– Energy/Performance Bias………….. : 0 (disabled/MSR not visible to software)
– Thermal Interrupt Coordination Enable : 1 (thermal interrupt routed to all cores)
MSR_TURBO_RATIO_LIMIT…………(0x1AD) : 0x1A1A1B1C1D1E2020
——————————————
– Maximum Ratio Limit for C01………. : 20 (3200 MHz)
– Maximum Ratio Limit for C02………. : 20 (3200 MHz)
– Maximum Ratio Limit for C03………. : 1E (3000 MHz)
– Maximum Ratio Limit for C04………. : 1D (2900 MHz)
– Maximum Ratio Limit for C05………. : 1C (2800 MHz)
– Maximum Ratio Limit for C06………. : 1B (2700 MHz)
– Maximum Ratio Limit for C07………. : 1A (2600 MHz)
– Maximum Ratio Limit for C08………. : 1A (2600 MHz)
IA32_ENERGY_PERF_BIAS…………(0x1B0) : 0x1
——————————————
– Power Policy Preference……………: 1 (highest performance)
MSR_POWER_CTL………………..(0x1FC) : 0x2904005B
——————————————
– Bi-Directional Processor Hot……….: 1 (enabled)
– C1E Enable……………………….: 1 (enabled)
MSR_RAPL_POWER_UNIT…………..(0x606) : 0xA0E03
——————————————
– Power Units…………………….. : 3 (1/8 Watt)
– Energy Status Units……………… : 14 (61 micro-Joules)
– Time Units …………………….. : 10 (976.6 micro-Seconds)
MSR_PKG_POWER_LIMIT…………..(0x610) : 0x78330001582A8
——————————————
– Package Power Limit #1…………… : 85 Watt
– Enable Power Limit #1……………. : 1 (enabled)
– Package Clamping Limitation #1……. : 1 (allow going below OS-requested P/T state during Time Window for Power Limit #1)
– Time Window for Power Limit #1……. : 10 (2560 milli-Seconds)
– Package Power Limit #2…………… : 102 Watt
– Enable Power Limit #2……………. : 1 (enabled)
– Package Clamping Limitation #2……. : 1 (allow going below OS-requested P/T state setting Time Window for Power Limit #2)
– Time Window for Power Limit #2……. : 3 (20 milli-Seconds)
– Lock…………………………… : 0 (MSR not locked)
MSR_PKG_ENERGY_STATUS…………(0x611) : 0x375F1A4
——————————————
– Total Energy Consumed……………. : 3543 Joules (Watt = Joules / seconds)
MSR_PKG_POWER_INFO……………(0x614) : 0x2F0550014002A8
——————————————
– Thermal Spec Power………………. : 85 Watt
– Minimum Power…………………… : 0
– Maximum Power…………………… : 170
– Maximum Time Window……………… : 0
MSR_PP0_POWER_LIMIT…………..(0x638) : 0x0
MSR_PP0_ENERGY_STATUS…………(0x639) : 0x0
MSR_TURBO_ACTIVATION_RATIO…….(0x64C) : 0x0
MSR_CONFIG_TDP_NOMINAL………..(0x648) : 0x18
MSR_CONFIG_TDP_LEVEL1…………(0x649) : 0xA00550001502A8
MSR_CONFIG_TDP_LEVEL2…………(0x64a) : 0xA0055000000000
MSR_CONFIG_TDP_CONTROL………..(0x64b) : 0x0
MSR_TURBO_ACTIVATION_RATIO…….(0x64c) : 0x0
MSR_PKGC6_IRTL……………….(0x60b) : 0x0
MSR_PKG_C2_RESIDENCY………….(0x60d) : 0x0
MSR_PKG_C6_RESIDENCY………….(0x3f9) : 0x0
IA32_TSC_DEADLINE…………….(0x6E0) : 0x256336C2D5B4
CPU Ratio Info:
——————————————
Base Clock Frequency (BLCK)…………. : 100 MHz
Maximum Efficiency Ratio/Frequency…….: 12 (1200 MHz)
Maximum non-Turbo Ratio/Frequency……..: 24 (2400 MHz)
Maximum Turbo Ratio/Frequency…………: 32 (3200 MHz)
P-State ratio * 100 = Frequency in MHz
——————————————
CPU P-States [ (12) 24 30 ]
CPU C6-Cores [ 1 2 4 6 10 12 14 ]
CPU P-States [ (12) 18 24 30 ]
CPU C6-Cores [ 0 1 2 4 6 8 10 12 14 ]
CPU P-States [ (12) 18 20 24 30 ]
CPU P-States [ (12) 18 20 23 24 30 ]
CPU C6-Cores [ 0 1 2 4 6 8 10 12 14 15 ]
CPU P-States [ (12) 18 19 20 23 24 30 ]
CPU C6-Cores [ 0 1 2 3 4 6 8 10 12 14 15 ]
CPU P-States [ 12 18 19 20 23 24 30 (32) ]
CPU P-States [ (12) 18 19 20 22 23 24 30 32 ]
CPU C6-Cores [ 0 1 2 3 4 6 8 9 10 12 13 14 15 ]
????
also not clear on what you meant by github issues
Didn’t you download and compile the source code of AppleIntelInfo.kext from Github? That is also where Github Issue can be found, and issues – aka bugs – should be files there.
Gotcha. I’m re reading everything, and for Haswell-E Xeon do I have to do a fakecpuID to a regular Haswell for this to work? Do I keep VoodooTSCSync.kext? I tried deleting both VoodooTSCSync and NullCPUPowerManagement on 10.12.3 and my CPU isn’t detected by the OS but as you can see from the output it’s being detected fine in the AppleIntelInfo.kext.
What can I do to troubleshoot?
Hello pikeralpha,
For anyone with 6950x I have instructions on how to boot with 0x040674, never found them on internet, but somehow they work for me on 10.12.3
Here is my plist and the special part – is kernel patches from 5960x that really work for me on Asus Deluxe 2
http://pastebin.com/WeHm9gMK
Currently I’m not able to get xcpm, despite of my FakeCPUID is 0x040674 (Broadwell E)
Here is the output from AppleIntelInfo
http://pastebin.com/U5TCzUL7
I have MSR 0xE2(if I don’t mistake) power management locked, I wasn’t able to upload patched BIOS 14.01 to Asus Deluxe 2 (error: not a proper BIOS)
In BIOS I have enabled C states, disabled EIST, set sync all cores with 45 ratio (stable for my liquid cooling)
Vectors set to Mac Pro 6,1, but they are not loaded
seven$ sysctl -n machdep.xcpm.vectors_loaded_count
0
kextstat | grep “X86PlatformPlugin”
80 1 0xffffff7f81c7d000 0x17000 0x17000 com.apple.driver.X86PlatformPlugin (1.0.0) 782DA5F5-F9BB-3B75-B40D-EC8EFFCE89A8
xcpm disabled
seven$ sysctl machdep.xcpm.mode
machdep.xcpm.mode: 0
SSDT was generated by
ssdtPRGen.sh -m “MacPro6,1” -target 3 -xcpm 1 -turbo 4500
Any help will be very appreciated because I spent lot of time and xcpm still wasn’t working
Currently I have turbo boost enabled and stable, frequency changes often from 1.2 to 2.79 in idle, and flat line with 4.5 Ghz when benchmarking
Screenshot from Intel Power Gadge: https://gyazo.com/956cb6d7669808ed68cd75e12b53c14a
What you need is the bold data in 0x040674. The rest is unimportant.
What exactly do you mean with: “Vectors set to Mac Pro 6,1, but they are not loaded“. Did you run freqVectorsEdit.sh?
Thank you for the quick response
I ran freqVectorsEdit.sh with MacPro 6,1 – no vectors are loaded
Then I tried with iMac 17,1 – also no vectors are loaded
Currently my FakeCPUID is 0x040674, but I don’t have XCPM working
The patches I applied are located here http://pastebin.com/WeHm9gMK
Anthony,
Have you checked the plist? Are the FrequencyVectors actually there?
pikeralpha, I don’t know how to verify that it is correct, but here is my plist
http://pastebin.com/WT6NaHMq
key FrequencyVectors exists there
Hi piker. I just implemented your patches and now my processor is running full speed. I was advised to delete null cpu power management and -xcpm and fake Cupid. What should I do?
I have no idea what Intel processor that you have, and what version of OS X that you are using. Also. The patch – as is in my blog post – may not work with newer versions of OS X, and I have no time to update in case it fails. The whole patching concept is explained, so you should be able to do that yourself.
Sorry I should’ve included those details. I have a 5820k and I’m running Sierra 10.12.2 I’ve looked around and seen that your patches have worked to Sierra. The only thing I’m not using is your cpu id patch as i thought it wasn’t needed in Sierra. Thank you.
Miles,
Would you please be so kind to verify that the patches are still matching with the macOS Sierra 10.12.2 binary. Thank you.
I’m sorry as this is probably a silly question. I’ve read your blog multiple times and I don’t understand how I would go about double checking the patches are for 10.12.2. Currently I’m using your Haswell-E CPUID patch (this should probably be disabled as it isn’t necessary in Sierra?), your xcpm-bootstrap, xcpm_core_scope, and xcpm idle. Btw, if you don’t mind me asking. Where did you acquire all the skills and knowledge to be able to work on operating systems at this level. I would really like to learn myself. thank you.
also wondering this. Not totally sure how to apply the patches to 10.12.2. I have a Haswell-E Xeon and looks like almost no one is running this processor either. P and C states look like they exist but can’t reach turbo speeds at all. TBH I would be down to run turbo speeds at all time, but my processor is NEVER reaching turbo speed.
Would love to know more in depth how to check these patches myself
Hi sorry. I’ve been asking for help in the past and I realize I haven’t been proactive on my part. Anyways, I’m wondering what to do now. I’m running Sierra 10.12.2 and I’ve implemented your patches I hope they’re the right ones as I can’t figure out how to check if they match the macOS Sierra 10.12.2 binary, sorry. Here is my readout from AppleIntelInfo.kext. I could be wrong but I feel like there are supposed to be more P and C states no? Thank you for your help.
AppleIntelInfo.kext v2.0 Copyright © 2012-2016 Pike R. Alpha. All rights reserved
enableHWP…………………………..: 0
Settings:
——————————————
logMSRs…………………………….: 1
logIGPU…………………………….: 0
logCStates………………………….: 1
logIPGStyle…………………………: 1
Warning: Clover hw.busfrequency error detected : 17d78400
InitialTSC………………………….: 0x87cb9c0f2cb (282 MHz)
MWAIT C-States………………………: 8480
Processor Brandstring………………..: Intel(R) Core(TM) i7-5820K CPU @ 3.30GHz
Processor Signature………………… : 0x306F2
——————————————
– Family…………………………. : 6
– Stepping……………………….. : 2
– Model………………………….. : 0x3F (63)
Model Specific Registers (MSRs)
——————————————
MSR_CORE_THREAD_COUNT…………(0x35) : 0xFFFFFF80786A2900
——————————————
– Core Count……………………… : 6
– Thread Count……………………. : 12
MSR_PLATFORM_INFO…………….(0xCE) : 0x20080C3BF3812100
——————————————
– Maximum Non-Turbo Ratio………….. : 0x21 (3300 MHz)
– Ratio Limit for Turbo Mode……….. : 1 (programmable)
– TDP Limit for Turbo Mode…………. : 1 (programmable)
– Low Power Mode Support…………… : 1 (LPM supported)
– Number of ConfigTDP Levels……….. : 1 (additional TDP level(s) available)
– Maximum Efficiency Ratio…………. : 12
– Minimum Operating Ratio………….. : 8
MSR_PMG_CST_CONFIG_CONTROL…….(0xE2) : 0x402
——————————————
– I/O MWAIT Redirection Enable……… : 1 (enabled, IO read of MSR(0xE4) mapped to MWAIT)
– CFG Lock……………………….. : 0 (MSR not locked)
– C3 State Auto Demotion…………… : 0 (disabled/unsupported)
– C1 State Auto Demotion…………… : 0 (disabled/unsupported)
– C3 State Undemotion……………… : 0 (disabled/unsupported)
– C1 State Undemotion……………… : 0 (disabled/unsupported)
– Package C-State Auto Demotion…….. : 0 (disabled/unsupported)
– Package C-State Undemotion……….. : 0 (disabled/unsupported)
MSR_PMG_IO_CAPTURE_BASE……….(0xE4) : 0x10414
——————————————
– LVL_2 Base Address………………. : 0x414
– C-state Range…………………… : 1 (C6 is the max C-State to include)
IA32_MPERF…………………..(0xE7) : 0x21474D0BC
IA32_APERF…………………..(0xE8) : 0x294CE8E25
MSR_FLEX_RATIO……………….(0x194) : 0xE0000
——————————————
MSR_IA32_PERF_STATUS………….(0x198) : 0x29E100002C00
——————————————
– Current Performance State Value…… : 0x2C00 (4400 MHz)
MSR_IA32_PERF_CONTROL…………(0x199) : 0xFF00
——————————————
– Target performance State Value……. : 0xFF00 (25500 MHz)
– Intel Dynamic Acceleration……….. : 0 (IDA engaged)
IA32_CLOCK_MODULATION…………(0x19A) : 0x0
IA32_THERM_INTERRUPT………….(0x19B) : 0x0
IA32_THERM_STATUS…………….(0x19C) : 0x88400000
——————————————
– Thermal Status………………….. : 0
– Thermal Log…………………….. : 0
– PROCHOT # or FORCEPR# event………. : 0
– PROCHOT # or FORCEPR# log………… : 0
– Critical Temperature Status………. : 0
– Critical Temperature log…………. : 0
– Thermal Threshold #1 Status………. : 0
– Thermal Threshold #1 log…………. : 0
– Thermal Threshold #2 Status………. : 0
– Thermal Threshold #2 log…………. : 0
– Power Limitation Status………….. : 0
– Power Limitation log…………….. : 0
– Current Limit Status…………….. : 0
– Current Limit log……………….. : 0
– Cross Domain Limit Status………… : 0
– Cross Domain Limit log…………… : 0
– Digital Readout…………………. : 64
– Resolution in Degrees Celsius…….. : 1
– Reading Valid…………………… : 1 (valid)
MSR_THERM2_CTL……………….(0x19D) : 0x0
IA32_MISC_ENABLES…………….(0x1A0) : 0x850089
——————————————
– Fast-Strings……………………. : 1 (enabled)
– FOPCODE compatibility mode Enable…. : 0
– Automatic Thermal Control Circuit…. : 1 (enabled)
– Split-lock Disable………………. : 0
– Performance Monitoring…………… : 1 (available)
– Bus Lock On Cache Line Splits Disable : 0
– Hardware prefetch Disable………… : 0
– Processor Event Based Sampling……. : 0 (PEBS supported)
– GV1/2 legacy Enable……………… : 0
– Enhanced Intel SpeedStep Technology.. : 1 (enabled)
– MONITOR FSM…………………….. : 1 (MONITOR/MWAIT supported)
– Adjacent sector prefetch Disable….. : 0
– CFG Lock……………………….. : 0 (MSR not locked)
– xTPR Message Disable…………….. : 1 (disabled)
MSR_TEMPERATURE_TARGET………..(0x1A2) : 0x690A00
——————————————
– Turbo Attenuation Units………….. : 0
– Temperature Target………………. : 105
– TCC Activation Offset……………. : 0
MSR_MISC_PWR_MGMT…………….(0x1AA) : 0x400000
——————————————
– EIST Hardware Coordination……….. : 0 (hardware coordination enabled)
– Energy/Performance Bias support…… : 1
– Energy/Performance Bias………….. : 0 (disabled/MSR not visible to software)
– Thermal Interrupt Coordination Enable : 1 (thermal interrupt routed to all cores)
MSR_TURBO_RATIO_LIMIT…………(0x1AD) : 0x22222C2C2C2C2C2C
——————————————
– Maximum Ratio Limit for C01………. : 2C (4400 MHz)
– Maximum Ratio Limit for C02………. : 2C (4400 MHz)
– Maximum Ratio Limit for C03………. : 2C (4400 MHz)
– Maximum Ratio Limit for C04………. : 2C (4400 MHz)
– Maximum Ratio Limit for C05………. : 2C (4400 MHz)
– Maximum Ratio Limit for C06………. : 2C (4400 MHz)
IA32_ENERGY_PERF_BIAS…………(0x1B0) : 0x7
——————————————
MSR_POWER_CTL………………..(0x1FC) : 0x2B04005B
——————————————
– Bi-Directional Processor Hot……….: 1 (enabled)
– C1E Enable……………………….: 1 (enabled)
MSR_RAPL_POWER_UNIT…………..(0x606) : 0xA0E03
——————————————
– Power Units…………………….. : 3 (1/8 Watt)
– Energy Status Units……………… : 14 (61 micro-Joules)
– Time Units …………………….. : 10 (976.6 micro-Seconds)
MSR_PKG_POWER_LIMIT…………..(0x610) : 0x79F40005B9F40
——————————————
– Package Power Limit #1…………… : 1000 Watt
– Enable Power Limit #1……………. : 1 (enabled)
– Package Clamping Limitation #1……. : 1 (allow going below OS-requested P/T state during Time Window for Power Limit #1)
– Time Window for Power Limit #1……. : 45 (40960 milli-Seconds)
– Package Power Limit #2…………… : 1000 Watt
– Enable Power Limit #2……………. : 1 (enabled)
– Package Clamping Limitation #2……. : 1 (allow going below OS-requested P/T state setting Time Window for Power Limit #2)
– Time Window for Power Limit #2……. : 3 (20 milli-Seconds)
– Lock…………………………… : 0 (MSR not locked)
MSR_PKG_ENERGY_STATUS…………(0x611) : 0x30252010
——————————————
– Total Energy Consumed……………. : 49300 Joules (Watt = Joules / seconds)
MSR_PKG_POWER_INFO……………(0x614) : 0x1280460
——————————————
– Thermal Spec Power………………. : 140 Watt
– Minimum Power…………………… : 0
– Maximum Power…………………… : 0
– Maximum Time Window……………… : 0
MSR_PP0_POWER_LIMIT…………..(0x638) : 0x0
MSR_PP0_ENERGY_STATUS…………(0x639) : 0x0
MSR_TURBO_ACTIVATION_RATIO…….(0x64C) : 0x0
MSR_CONFIG_TDP_NOMINAL………..(0x648) : 0x21
MSR_CONFIG_TDP_LEVEL1…………(0x649) : 0x940000001E0460
MSR_CONFIG_TDP_LEVEL2…………(0x64a) : 0x94000000000000
MSR_CONFIG_TDP_CONTROL………..(0x64b) : 0x0
MSR_TURBO_ACTIVATION_RATIO…….(0x64c) : 0x0
MSR_PKGC6_IRTL……………….(0x60b) : 0x0
MSR_PKG_C2_RESIDENCY………….(0x60d) : 0xAB97DDFE85
MSR_PKG_C6_RESIDENCY………….(0x3f9) : 0xDBCDC117F2
IA32_TSC_DEADLINE…………….(0x6E0) : 0x87CBB7AE51F
CPU Ratio Info:
——————————————
Base Clock Frequency (BLCK)…………. : 100 MHz
Maximum Efficiency Ratio/Frequency…….: 12 (1200 MHz)
Maximum non-Turbo Ratio/Frequency……..: 33 (3300 MHz)
Maximum Turbo Ratio/Frequency…………: 44 (4400 MHz)
P-State ratio * 100 = Frequency in MHz
——————————————
CPU P-States [ 41 (44) ]
CPU C6-Cores [ 0 2 4 6 8 10 ]
CPU P-States [ 38 41 (44) ]
CPU P-States [ 38 39 41 (44) ]
CPU P-States [ (12) 38 39 41 44 ]
CPU P-States [ 12 38 39 40 41 (44) ]
CPU C6-Cores [ 0 2 4 5 6 8 9 10 ]
CPU C6-Cores [ 0 2 4 5 6 7 8 9 10 ]
You see less P-States because you have set all cores to the same frequency. This is perhaps not what you want, but that’s what you get for using it. Ergo. You only see one instead of six P-States.
Also. I have said this many times before, but look at these two:
Package Power Limit #1…………… : 1000 Watt
Package Power Limit #2…………… : 1000 Watt
1000 Watt? Seriously? Now look at this:
– Target performance State Value……. : 0xFF00 (25500 MHz)
That is not right. You must be using a patch or kext to keep it that hight. I don’t know why, but hey… who am I to judge other peoples decisions. We’ll get to it when more people fry their CPU. And I have already seen a couple of reports where the CPU did actually burned up.
What do you want? A burned up CPU, or a CPU that runs slower in synthetic benchmarks, but will keep running for a much longer time? With your settings, it is time to start saving for a new motherboard and Intel Processor. Don’t say that I didn’t warn people about this, because I did.
Alright I think I know the patch that is doing it. However my CPU seems to stay around 40.0 Celsius according to Intel Speed gadget although that’s probabaly still too high. I just want my CPU to be cool and calm when I’m browsing the internet and rev up when I need it. I’ll stick to only your patches from now on. Thanks.
Hey Piker I don’t know if you would find this interested or not but after your comment I went through some digging in my bios. I found that the reason the package power limit was 1000 watts it was actually automatically set by my mobos auto overclock. I have no idea what Asrock was doing there. I set it to 250 watts which is still probably too high but I don’t know enough yet about wattage for cpus to know a better value. I’ll probably end up setting it 200. Further you pointed out before that the Target performance state value was 25500mhz which was crazy. I don’t know what was causing that but now I have a Target performance value of 0xC00 (1200 MHz). I don’t know if that is good or bad for a 5820k @ 4.4ghz.
The only things I need to change now is to make sure that SYNC all cores is disabled in Bios and that should give me the extra p-states.
I just have one last question. I can’t for the life of me get your frequency vectors loaded. I’ve run the script a lot of times using both iMac 14,2 and MacPro 6,1(I’m running X99 Sierra 10.12.2 so I thought these would be the rights board ids) with neither loading. I finally took the plist and manually copied it over to /System/Library/Extensions/IOPlatformPluginFamily.kext/Contents/PlugIns/X86PlatformPlugin.kext/Contents/Resources yet still sysctl -n machdep.xcpm.vectors_loaded_count returns 0. I believe this is the last step to get full power management because currently without the vectors loaded X86platform kext and AppleLPC are not loaded, sleep doesn’t work, and my multi-core performance isn’t doing too great according to GeekBench. I hope properly loaded vectors will fix all that, I just don’t know how to get them loaded.
Thank you in advance. All help is appreciated. Also, thanks for all the work you do for this community. I hope next year I’ll be more knowledgeable in UNIX and be able to give back as well.
Wait what? X86PlatformPlugin.kext isn’t loading? Please try this terminal command:
kextstat -lb com.apple.driver.X86PlatformPluginThat should show a line, but if it turns up blank, then your “plugin-type” property on CPU0 is missing. This is normally injected from a _DSM method. Please check this and fix in case it isn’t being set.
Sorry as this is maybe a silly question. I’ve been looking all over for ways to fix my issue and get working power management. I think I’ve finally settled on something that works that idles around 20 w. Anyways, I still don’t have X86PlatformPlugin loading and no vectors loaded despite using various board.plist like iMac 14,2, Mac Pro 6,1 (which I realize is the wrong chipset), and iMac 14,1. I don’t know how to inject a “plugin-type” with a _DSM method. It’s to my understanding that this should be done with an ssdt.aml generated with your script. Maybe I’m generating my script wrong. Either way I think I might soon revert back to NullCPU and FakeCPUID, I guess maybe not all boards work the same for some reason or another. I’ll try regenerating my ssdt.aml and try out some other plists in your freqVectorsedit.sh script. Thanks for all your hard work.
Hi Pike, in 10.12.4 beta 5 release seems that something is changed (reboot fix value):
find
554889E5415741564155415453504189D64189F74889FB4585FF0F84
replace
C34889E5415741564155415453504189D64189F74889FB4585FF0F84
reboot fix 10.12 © Pike R. Alpha
I have calculated it like you teached us some time ago
ffffff800041f139 leaq _xcpm_pkg_scope_msrs(%rip), %rdi
ffffff800041f140 movl $0x7, %esi
ffffff800041f145 xorl %edx, %edx
ffffff800041f147 callq 0xffffff800041ede0
and so we have this:
48c1e22089c04809d0482b0570ea660048890569ea6600b932060000
but replacing 48 with 53 does not help
Any advice?
Thank you for you research
Fabio,
Let us try to reproduce this. I think your search and replace data is wrong. Here is mine with macOS Sierra 10.12.4 Build 16E183b:
Step 1
otool -tV /S*/L*/Kernels/kernel > ~/Documents/kernel.txtStep 2
open ~/Documents/kernel.txtStep 3
Search for: “_xcpm_init:”
Step 4
Scroll down to:
ffffff800041f139 leaq _xcpm_pkg_scope_msrs(%rip), %rdi
ffffff800041f140 movl $0x7, %esi
ffffff800041f145 xorl %edx, %edx
ffffff800041f147 callq 0xffffff800041ede0
Step 5
Get file offset like this: 0x41ede0 – 0x200000 = 0x21ede0
Step 6
xxd -s 0x21ede0 -l 28 -psu /S*/L*/Kernels/kernelHere I get this:
554889e5415741564155415453504189d64189f74889fb4585ff0f84This is your Find pattern fro Clover. The Replace pattern is the same, but for this you replace the first 55 with C3 (example below):
C34889e5415741564155415453504189d64189f74889fb4585ff0f84I think that you got lost in step 5 because this data is wrong:
48c1e22089c04809d0482b0570ea660048890569ea6600b932060000You should use the find and replace data at the top of your post.
Note: This patch stops a reboot that will be triggered by writing to unsupported MSR’s in _xcpm_pkg_scope_msrs. Nothing else!
Thank you Pike, I did a path error in step 6
I have used this kernel patch together fakecpuid 0x040674 from macOS Sierra 10.12 to 10.12.3 with different x99 c610 motherboard with my xeon E5 2696 v4 (in single cpu mode and dual mode capping cores to 64)
with only your patch (_xcpm_pkg_scope_msrs) and a IOPCIFamily.kext patch (find 4881fb00 000040 replace 4881fb00 000080)
I have had a perfect and stable system
Now in beta 5 (10.12.4) it seems not working anymore
But thank you, now I know it is not “reboot fix patch” related
Hi. Fabio. I don’t think the hanging is related to XCPM patches. Just cuz the patches you used didn’t change. And also. There’s no new routine in XCPM related funcs.
Thank you Angel
With the help of PMHearth on Insanelymac forum I figured out that my system does not load fakesmc with 10.12.4 beta 3 or beta 5 but I don’t know why by now
You may not get that I’m actually PMheart @ InsanelyMac ! Anyway. The problem is now gone, isn’t it? 😇
Hi Angel, No I had not understood who you are in Insanely! 🙂
Yes, I solved with the same kext and config with only new boox64 and cloverx64 efi files
I’m PMheart! http://www.insanelymac.com/forum/user/1579222-pmheart/ 😇
Hey. At least I knew that 😉
How did you get that Pike 😜
By paying attention 😉
LOL
😂😂 I understand it when you said the first time PMHearth
Back to homework😊
No. Basically I have no homework to do now. 😊
nevermind..with beta 3 e 5 I have to update clover
old clover not works
Hi Pike,
I really need your help! I’m getting some random CPU kernel panics (before login and randomly after log in) and I can’t figure out why. I generated a SSDT via “sudo ~/ssdtPRGen.sh -x 1,” then injected 17,1 frequency vectors.
CPU: 6950x
Fake CPUID: 0x040674
KernelToPatch:
F: 554889e5 41574156 41554154 53504189d64189f7 4889fb45 85ff0f84
R: c34889e5 41574156 41554154 53504189d64189f7 4889fb45 85ff0f84
C: reboot fix 10.12 © Pike R. Alpha
XCPM mode seems to be working…
machdep.xcpm.epp_override: 0
machdep.xcpm.io_epp_boost_enabled: 0
machdep.xcpm.ring_boost_enabled: 0
machdep.xcpm.io_cst_control_enabled: 0
machdep.xcpm.io_filtered_reads: 0
machdep.xcpm.io_control_disengages: 1
machdep.xcpm.io_control_engages: 0
machdep.xcpm.io_thermal_level: 0
machdep.xcpm.gpu_thermal_level: 0
machdep.xcpm.cpu_thermal_level: 0
machdep.xcpm.deep_idle_total_stats: n/a
machdep.xcpm.deep_idle_last_stats: n/a
machdep.xcpm.deep_idle_count: 0
machdep.xcpm.qos_txfr: 1
machdep.xcpm.deep_idle_log: 0
machdep.xcpm.forced_idle_period: 30000000
machdep.xcpm.forced_idle_ratio: 100
machdep.xcpm.mbd_relaxations: 0
machdep.xcpm.mbd_applications: 1
machdep.xcpm.mbd_mode: 1
machdep.xcpm.maxintdelay: 0
machdep.xcpm.maxbusdelay: 71680
machdep.xcpm.ratio_changes_total: 5884
machdep.xcpm.ratio_change_ratelimit_ns: 500000
machdep.xcpm.vectors_loaded_count: 1
machdep.xcpm.tuib_ns: 0
machdep.xcpm.bootpst: 40
machdep.xcpm.bootplim: 0
machdep.xcpm.power_source: 0
machdep.xcpm.tuib_enabled: 0
machdep.xcpm.tuib_plimit_min_100mhz_ratio: 8
machdep.xcpm.tuib_plimit_max_100mhz_ratio: 40
machdep.xcpm.soft_plimit_min_100mhz_ratio: 8
machdep.xcpm.soft_plimit_max_100mhz_ratio: 40
machdep.xcpm.hard_plimit_min_100mhz_ratio: 8
machdep.xcpm.hard_plimit_max_100mhz_ratio: 40
machdep.xcpm.mode: 1
Here’s the KP right before it was going to boot into the OS: http://i.imgur.com/1KARBN8.png
Thank you for your time!
I’m not really in the mood to track down a KP. Sorry.
No worries, bud. I understand.
I updated my patched BIOS and found some Clover MSR patches that may have fixed the issue, but still need to test for a few weeks to be sure. When using these patches, in dmesg, I only see
“IOPPF: XCPM Mode”
“IOPPF: XCPM Mode”
Instead of:
“IOPPF: XCPM Mode”
“X86PlatformShim:: start – Failed to send to stepper”
followed by insta-KP or KP after 10 minutes or so.
Thank you for time!
Thanks. Not much time, but search ssdtPRGen.sh GitHub issues for the error: “X86PlatformShim:: start – Failed to send to stepper“. That should help.
I’m a little confused by what you mean by DP1, DP2, …, DP7. Can you give some tips on how to patch the kernel? what application we can use and so on..
Thanks
DP = Developer Preview.
Hi Piker,
Thank your for your contributions to this hobby.
I have a question for you. IPG (Intel Power Gadget) and HWmonitor never show idle below 30watts for my processor (5820K) and sysctl machdep.xcpm.mode returns 0 instead of 1.
However IGT and HWM both show 1.2GHZ at idle with 30watts usage (lowest I get)
During a cinebench CPU test IPG and HWM report 3.6GHZ (sustained during the test) at 104watts.
So I am assuming xcpm is working even with Sysctl coming back with 0. Also everything works for my build, sleep/wake,audio, USB etc.
I wanted to pick your brain in regards to sysctl machdep.xcpm,mode returning a 0 but xcpm apperantly working. I wonder if there is a correlation between it returning 0 and my idle wattage being high.
Thank you for your time.
No. XCPM isn’t working when machdep.xcpm.mode returns a zero.
Hi Piker,
Question 1: which frequency vectors did you inject for your i7-6850K? I have an i7-6950X and it seems that I do not find the appropriate resource file (plists) with the correct Frequency Vectors…
Question 2: xcpm_bootstrap apparently breaks XCPM in my case. Everything works fine, but as soon I implement this KernelToPatch entry, both “sysctl machdep.xcpm.mode” and “sysctl -n machdep.xcpm.vectors_loaded_count” suddenly reveal “0”.. Any idea what might go wrong?
Thanks for your time and cheers!
Hi PIke,
I have an intel i7 5820K on a Gigabyte X99 UD4 motherboard.
I also have the wrong values being displayed for Package Power Limit #1 and Package Power Limit #2 (3408 and 4090 Watts respectively) under the output of AppleIntelInfo.kext. I looked all over the comments (to this blog post, current and previous) and on Google but I couldn’t find any way to fix it. Can you help me with a link to the fix please? Thanks in advance.
HI, first thanks for your hard work. I tried to port this patch for Maverikcs. Please check this and let me know.
(__TEXT,__text) section
_vm_swapfile_for_handle:
ffffff8000202000 pushq %rbp
@@ -269392,7 +269392,11 @@
ffffff80002fa014 leaq _xcpm_SMT_scope_msrs(%rip), %rdi
ffffff80002fa01b movl $0x5, %esi
ffffff80002fa020 movl $0x1b0, %edx
ffffff80002fa025 popq %rbp
-ffffff80002fa026 jmpq 0xffffff80002fa340
+ffffff80002fa026 ret
+ffffff80002fa027 nop
+ffffff80002fa028 nop
+ffffff80002fa029 nop
+ffffff80002fa02a nop
ffffff80002fa02b nopl (%rax,%rax)
_xcpm_perf_bias_get:
ffffff80002fa030 pushq %rbp
@@ -269435,7 +269439,11 @@
ffffff80002fa0d3 movl $0x7, %esi
ffffff80002fa0d8 je 0xffffff80002fa0ed
ffffff80002fa0da xorl %edx, %edx
-ffffff80002fa0dc callq 0xffffff80002fa340
+ffffff80002fa0dc nop
+ffffff80002fa0dd nop
+ffffff80002fa0de nop
+ffffff80002fa0df nop
+ffffff80002fa0e0 nop
ffffff80002fa0e1 testb $0x1, %bl
ffffff80002fa0e4 jne 0xffffff80002fa107
ffffff80002fa0e6 jmp 0xffffff80002fa0f4
@@ -269448,11 +269456,19 @@
ffffff80002fa0f4 leaq _xcpm_core_scope_msrs(%rip), %rdi
ffffff80002fa0fb movl $0x3, %esi
ffffff80002fa100 xorl %edx, %edx
-ffffff80002fa102 callq 0xffffff80002fa340
+ffffff80002fa102 nop
+ffffff80002fa103 nop
+ffffff80002fa104 nop
+ffffff80002fa105 nop
+ffffff80002fa106 nop
ffffff80002fa107 leaq _xcpm_SMT_scope_msrs(%rip), %rdi
ffffff80002fa10e movl $0x5, %esi
ffffff80002fa113 xorl %edx, %edx
-ffffff80002fa115 callq 0xffffff80002fa340
+ffffff80002fa115 nop
+ffffff80002fa116 nop
+ffffff80002fa117 nop
+ffffff80002fa118 nop
+ffffff80002fa119 nop
ffffff80002fa11a testl %r14d, %r14d
ffffff80002fa11d je 0xffffff80002fa158
ffffff80002fa11f movq _xcpm_mchbar(%rip), %rax
The new macOS Sierra 10.12.5 Update for 2017 iMacs has native support for Kabylake but the _cpuid_set_info and _xcpm_bootstrap patch changed. The new patch:
_cpuid_set_info, replace XX
0F B6 C4 83 C0 E9 3D 87 00 00 00
to
0F B6 C4 83 C0 XX 3D 87 00 00 00
_xcpm_bootstrap, replace XX
8D 43 C4 83 F8 22
to
8D 43 XX 83 F8 22
Hi,
Thank you for the update!
Hi, On my journey of trying to port the patch to Mavericks 10.9.5 I found a problem.
Haswell -> Kabylake 3C -> 9E
_cpuid_set_info
ffffff80002ccf26 cmpl $0x3c, %eax
to
ffffff80002ccf26 cmpl $0x9e, %eax
74 11 83 F8 3C —-> 74 11 3D 9E 00 00 00
perl -pi -e ‘s|\x74\x11\x83\xF8\x3C|\x74\x11\x3D\x9E\x00\x00\x00|g’
_xcpm_bootstrap
ffffff80002fa97f cmpl $0x3c, %ebx
to
ffffff80002fa97f cmpl $0x9e, %ebx
EB 27 83 FB 3C —-> EB 27 81 FB 9E 00 00 00
perl -pi -e ‘s|\xEB\x27\x83\xFB\x3C|\xEB\x27\x81\xFB\x9E\x00\x00\x00|g’
But if I use the last otool to see the changes it gives me an error.
LLVM ERROR: truncated or malformed object (bad string index: 4294934528 for symbol at index 1)
Please help. Thanks
You need to change the find and replace patterns so that they are of the same length. Not you are injecting additional bytes.
I understand, but the following happens. I will take the _cpuid_set_info to explain me. If I change the pattern this way:
74 11 83 F8 3C —-> 74 11 83 F8 9E
I have the following result:
ffffff80002ccf26 cmpl $0x-62, %eax (Because 9E is greater than 3C?)
But it is not the result that I need.
The result I need is this:
ffffff80002ccf26 cmpl $0x9e, %eax
And the only way I have managed to get this result is by using the following pattern:
74 11 83 F8 3C —-> 74 11 3D 9E 00 00 00
I would need to check this, but I cannot do that right now.
If you want I can send you the 10.9.5 mach_kernel. Thank you for all your work. I’m a fan of you.
Any news concerning Broadwell and Haswell cpuid_set_info and bootstrap KerneltoPatch entries for macOS High Sierra Beta? Will ssdtPRGen.sh work with the actual Beta?
I think that it was Stinga who had something to say about this. Please verify.
Yes. The SSDT’s generated by the script work.
If I correctly understand, Stinga addresses the Kabylake update on 10.12.5. However, I would be interested in the actual Broadwell and Haswell KernelToPatch entries for macOS High Sierra Beta 10.13. Are you planning some update or modification of your “XCPM for unsupported Processor…” thread in the next time?
Hi,
No, but the trick is more or less the same:
_cpuid_set_info:
ffffff800039bd8c add al, 0x72
_xcpm_bootstrap:
ffffff80003cb825 add al, 0xc4
These are the locations and stock values of 10.13 build 17A291m. Does that help? People already know the value that they used to use for their unsupported processor, so there you have it 😉
10.13 public beta patches:
_cpuid_set_info, replace 72
04 72 3C D0
to
04 XX 3C D0
_xcpm_bootstrap, replace C4
89 D8 04 C4 3C 22
to
89 D8 04 XX 3C 22
_xcpm_SMT_scope_msrs(%rip), %rdi
BE 0B 00 00 00 5D E9 08 00 00 00
to
BE 0B 00 00 00 5D C3 90 90 90 90
_xcpm_pkg_scope_msrs(%rip), %rdi
BE 07 00 00 00 31 D2 E8 8F FC FF FF
to
BE 07 00 00 00 31 D2 90 90 90 90 90
_xcpm_core_scope_msrs(%rip), %rdi
BE 03 00 00 00 31 D2 E8 72 FC FF FF
to
BE 03 00 00 00 31 D2 90 90 90 90 90
_xcpm_SMT_scope_msrs(%rip), %rdi
BE 0B 00 00 00 31 D2 E8 5F FC FF FF
to
BE 0B 00 00 00 31 D2 90 90 90 90 90
Hi Piker,
I am configuring my E5-2648L v4 with Sierra now…
But the ssdtPrGen doesn’t support this CPU. I added one line in UserDefined.cfg:
‘E5-2648L v4’,75,1200,1800,2500,14,28,16,100
Should this suffice to generate a proper SSDT?
I just couldn’t get machdep.xcpm.mode to 1… Although I have tried different kernel patches..
Hi Teng,
Thanks. I added the missing processor data to: Data/Broadwell.cfg
Please remove ~/Library/ssdtPRGen and run the script again. Should work now.
Thanks piker!
One thing I’m confused is I’m still not able to get machdep.xcpm.mode = 1.
I have searched through the web, and you seems answered multiple times that it was because missing stepcontextdict in Mac-[board-Id].plist. However, I see from the github that you changed the behavior of freqVectorsEdit.sh by removing the stepcontextdict after patching frequencyVector.
I’m currently patched the macpro6,1 plist with your script, so there’s no stepcontextdict, and I receive the trace of “failed to send stepper”. I’m so confused… could you please enlighten me how it should work with the current methodology?(ssdt, frequencyvector, stepcontextdict).
Thanks!
I currently cannot check Github issues myself. Please provide links (per e-mail if you want). Are the FrequencyVectors added to the matching plist for your board-id?
Never mind. I figured it out. But another issue came up.
The xcpm works. mach.xcpm.mode == 1 and loadcount ==1.
But IA32_ENERGY_PERF_BIAS is 0.
I read your other article saying that this register should be [1,15]. How did you fixed it?
IA32_ENERGY_PERF_BIAS is normally initialised with the value from the resource plist. What resource plist (think board-id) are you using?
I use the macpro6,1 plist(Mac-F60DEB81FF30ACF6.plist), which was patched by your freqVectorsEdit.sh.
Model Specific Registers (MSRs)
MSR_CORE_THREAD_COUNT…………(0x35) : 0xFFFFFF806112A200
MSR_PLATFORM_INFO…………….(0xCE) : 0x20080C33FB811200
MSR_PMG_CST_CONFIG_CONTROL…….(0xE2) : 0x403
MSR_PMG_IO_CAPTURE_BASE……….(0xE4) : 0x10414
IA32_MPERF…………………..(0xE7) : 0x77D65E464
IA32_APERF…………………..(0xE8) : 0x7F25035D2
MSR_0x150……………………(0x150) : 0x0
MSR_FLEX_RATIO……………….(0x194) : 0x0
MSR_IA32_PERF_STATUS………….(0x198) : 0x197E00001200
MSR_IA32_PERF_CONTROL…………(0x199) : 0x1200
IA32_CLOCK_MODULATION…………(0x19A) : 0x0
IA32_THERM_INTERRUPT………….(0x19B) : 0x0
IA32_THERM_STATUS…………….(0x19C) : 0x883C0000
MSR_THERM2_CTL……………….(0x19D) : 0x0
IA32_MISC_ENABLES…………….(0x1A0) : 0x850089
MSR_TEMPERATURE_TARGET………..(0x1A2) : 0x601400
MSR_MISC_PWR_MGMT…………….(0x1AA) : 0x402000
MSR_TURBO_RATIO_LIMIT…………(0x1AD) : 0x1515151516171919
MSR_TURBO_RATIO_LIMIT1………..(0x1AE) : 0x1515151515151515
IA32_ENERGY_PERF_BIAS…………(0x1B0) : 0x0
MSR_POWER_CTL………………..(0x1FC) : 0x2904005B
MSR_RAPL_POWER_UNIT…………..(0x606) : 0xA0E03
MSR_PKG_POWER_LIMIT…………..(0x610) : 0x782D000158258
MSR_PKG_ENERGY_STATUS…………(0x611) : 0x53A1497
MSR_PKGC3_IRTL……………….(0x60a) : 0x0
MSR_PKGC6_IRTL……………….(0x60b) : 0x0
MSR_PKG_C2_RESIDENCY………….(0x60d) : 0xDAF7031E0
MSR_PKG_C3_RESIDENCY………….(0x3f8) : 0x0
MSR_PKG_C2_RESIDENCY………….(0x60d) : 0xDAF7031E0
MSR_PKG_C6_RESIDENCY………….(0x3f9) : 0x127E443EF8
IA32_TSC_DEADLINE…………….(0x6E0) : 0x113105B8CE867
That doesn’t tell me what I need to know – what board-id did you select to copy the data from when you ran the script?
You should run:
./freqVectorsEdit.sh -d 1This will show all the FrequencyVector related settings. Select/use a board-id that sets perf-bias (1).
Sorry misread your question.
I used this: it does have pref-bias(1)
Examining data of: Mac-031B6874CF7F642A.plist (iMac14,1) …
—————————————————————–
Converted to: /tmp/Mac-031B6874CF7F642A.bin (7668 bytes)
Settings: BACKGROUND, KGROUND, REALTIME_SHORT
hard-rt-ns (4000000), ubpc (1), off (0), on (0), perf-bias (1), utility-tlvl (79), non-focal-tlvl (250)
Ok. Check the debug output of the script; the output of the Mac-F60DEB81FF30ACF6.plist and Mac-031B6874CF7F642A.plist should be the same.
They look same…
Examining data of: Mac-031B6874CF7F642A.plist (iMac14,1) …
—————————————————————–
Converted to: /tmp/Mac-031B6874CF7F642A.bin (7668 bytes)
Settings: BACKGROUND, KGROUND, REALTIME_SHORT
hard-rt-ns (4000000), ubpc (1), off (0), on (0), perf-bias (1), utility-tlvl (79), non-focal-tlvl (250)
Examining data of: Mac-F60DEB81FF30ACF6.plist (MacPro6,1) …
—————————————————————–
Converted to: /tmp/Mac-F60DEB81FF30ACF6.bin (7668 bytes)
Settings: BACKGROUND, KGROUND, REALTIME_SHORT
hard-rt-ns (4000000), ubpc (1), off (0), on (0), perf-bias (1), utility-tlvl (79), non-focal-tlvl (250)
Is HWP enabled? The complete output of AppleIntelInfo.kext – latest version – would be very handy.
Herer’s the output.
I can see P state. Does this mean HWP enabled?
Ah. Ok. Well. No. There’s no HWP for your Xeon processor. That explains a lot.
Thanks! But is it because the CPU doesn’t have HWP or my BIOS didn’t enable it?
The output of the HWP related MSR’s is missing so no. It’s not supported by your processor.
If HWP is supported, but disabled, then the kext will also output a line.
Thanks for the explanation.. interesting..
Did you see other Xeon CPUs that don’t support the HWP? Or it’s because this CPU is an “L” version..
Pingback: Support for Skylake X found in High Sierra DP-X – Pike's Universum
Pike I have an Skylake-X i7-7800X and xcpm works just brilliant and splendid with SMBIOS iMac17,1.. But this SMBIOS has no entry in the IOUSBHostFamily.kext, thus USB3.0 write speeds are lousy, i.e.10MB/s…. None of the Skylake models you support, i.e. iMac17,1 , MacBook 9,1 , and MacBookPro13,1 , MacBookPro13,2, MacBookPro13,2 are part of IOUSBHostFamily.kext… If I however choose a compatible model like the MacPro6,1 I get the error message in my post above when executing ssdtPRGen.. What to do? What would be the -target option for MacPro6,1?
I really would appreciate a quick reply…
Thanks,
KGP
Sorry, but I am traveling. No time to look into things.
Tip: You could as well inject the missing USB data – copied from the correct profile – with help of a plist only kext.
Hi Pike, first of all thanks for your great articles and all your work.
I have a question, basically I’m not able to read the output of AppleIntelInfo kext to understand if I configured my xcpm correctly.
I’m running an x99 with an i7-5820k, stock speed (3.3 GHz) with Sierra 10.12.6.
The computer is working fine, it’s smooth and by looking at Intel Power Gadget it looks like that xcpm works, it ranges between 1.2GHz and 3.4GHz (never reached the max of 3.60 tho). I’m not sure it is 100% correct because of a couple of things:
1. Random kernel panics at boot (it looks like it only happens at boot and like once every 5-10 boots approximately)
2. Benchmark scores are really low compared to other benchmarks I see on the web with the same processor at stock speed.
For example, my current Geekbench 4 score is 3964 for single core and 14453 for multi-core, and I see this one where the multi-core is 5000+ higher
https://browser.geekbench.com/geekbench3/4264684
There are other which are even higher (~27000) but this one was run at 3.30 GHz like mine.
Same thing happens with other benchmarking software, mine is always lower.
I used the freqVectorsEdit.sh to inject what I thought were the correct frequencies for my processor by using this entry:
[ 28 ] Mac-B809C3757DA9BB8D.plist (iMac17,1)
I also used ssdtPRGen to enable turbo frequencies up to 4GHz with this command (In case I’ll decide to overclock once it’s stable at stock speed):
./ssdtPRGen.sh -a CPU0 -turbo 4000
Plugin-type value is set to 1:
Property: Type: Value:
plugin-type Number 0x1
Vectors are loaded:
sysctl -n machdep.xcpm.vectors_loaded_count
1
So the question is, can you tell me if my xcpm is configured correctly? Is xcpm in this configuration stable and not too stressful for the processor?
I read that package power limits are wrong and dangerous but I don’t know why.
Also, how can I know if the amount p states for my processor is correct? I mean, how can I know how much p states should it have to properly work?
Here are the kernel patches I’m using:
F: 83c0e9 R: 83c0e6 Haswell-E CPUID Patch 10.12
F: 20b9e200 00000f30 R: 20b9e200 00009090 xcpm_idle patch by Pike R. Alpha
F: 83c3c483 fb22 R: 83c3bc83 fb22 xcpm_bootstrap Sierra © Pike R. Alpha
F: be020000 0031d2e8 6cfcffff R: be020000 0031d290 90909090 xcpm_core_scope_msrs © Pike R. Alpha
F: be070000 0031d2e8 94fcffff R: be070000 0031d290 90909090 xcpm_pkg_scope_msr © Pike R. Alpha
F: 554889e5 41574156 41554154 53504189d64189f7 4889fb45 85ff0f84 R: c34889e5 41574156 41554154 53504189d64189f7 4889fb45 85ff0f84 reboot fix 10.12 © Pike R. Alpha
I don’t have FakeCPUID set and the cpu is recognized as a 3.3Ghz 6-core Intel Xeon
SMBIOS: Mac Pro 6,1
Here is the output of AppleIntelInfo.kext:
1.) There are three board-id’s for the iMac17,1 (with the same logic board):
Mac-B809C3757DA9BB8D:iMac17,1
Mac-65CE76090165799A:iMac17,1
Mac-DB15BD556843C820:iMac17,1
They all are using the same data so the one that you are using is fine. No matter what ill-informed people may say/think about it.
2.) The number of p-states can change only due to BIOS settings. Not just by running ssdtPRGen.sh
3.) Turbo is disabled in your BIOS, and this is the one that you should enable.
Note: The text: “Maximum Ratio/Frequency………………:” in the output will then also change into: “Maximum Turbo Ratio/Frequency…………:”.
Thank you for your reply, I enabled turbo in my BIOS and now, as you said, AppleIntelInfo shows me the correct line
Maximum Turbo Ratio/Frequency…………: 36 (3600 MHz)
However the benchmark has not improved, I’m still stuck between 12000 and 14000 for multi-core while the single-core is approximately the same.
Is this score related to wrong bios settings? What should I do to improve it?
Another question, are these two line correct? Or it’s something I shouldn’t care about?
– Package Power Limit #1…………… : 3408 Watt
– Package Power Limit #2…………… : 4090 Watt
i post on your blog Mac-B809C3757DA9BB8D:iMac17,1 with i7 6700k wit all informations