Jump to content

Testers needed: CPU Power Management for SB and IB Xeon(s) or i7-39xx on X79 or C60x chipset


omni
 Share

494 posts in this topic

Recommended Posts

To address some of your questions:

 

- AICPM I provided you with *is* from MacPro6,1 build, there's no other that you can get or need - this one detects the IB Xeon in MacPro6,1, all others don't.

 

- I *have* a Xeon E5 2.6GHz 8-core (two of them) and with it and my patched AICPM, my GB3 scores double. This tells me that at least my Turbo works. But also my CPU packages idles down to 10-11W which also tells me that it steps down, too. I also have a physical temperature monitor on my fan controller and it shows a drop when running this patch vs. no patch.

 

- I never said there are any guarantees this would work, especially with i7 Core-E chips. My guesstimate was (and still is) that because the cores of those chips are very close to Xeon ones (they lack SMP) it should basically be the same deal.

 

 

@yangshun1029

 

Your Max Efficient Power State is the lowest clock state your CPU can do, so look at your SSDT.

Link to comment
Share on other sites

Hi Omni,

 

I'm sorry. Apparently i missunterstood stomething. I tought you took the "old" AICPUPM and patched it, so that it recognizes the xeon e5.

 

I also have the Xeon E5-2690 v2. With your patched versions i also have fully working turbo states. Only downclocking and C-States are not working.

Could you provide an unpatched version(stock kext from macpro6,1)?

 

Do you have working p-state reporting in HWmon and/or AICPUPMI by Pike? I especially mean between 1200Mhz and stock frequency. (e.g. 3000Mhz).

 

Doing crazy things with my SSDT i also get very low power/wattage and working turboboost. But the P and C States aren't correctly recognized by HWmon and AICPUPMI, but with IPG. (I posted a screenshot).

 

Could you tell us, what and why you patched the stock kext? Did it kernel panic?

 

I think with the right SSDT the stock version shouldn't panic. therefore i would like to try the stock kext.

 

Could you also upload your SSDT? Did you use pike's script to generate it?

Link to comment
Share on other sites

I'm a loss for understanding. With Rampage IV Extreme BIOS 4206 I had to use CodeRush's fix to unlock MSR E2 and the v2 patched AICPUPM worked without KP but v3 gave me KP within 2-3 seconds. I updated to BIOS 4802 (MSR E2 is unlocked from Asus in this version) and I can now use v3 without KP. In IOReg the X86PlatformPlugin is loaded with all CStates enabled and all PStates listed but my 3930K is locked at 12x and HWmonitor reports CPU Temp=23c & CPU Package Cores=3.92W @ idle. Under load the CPU Temp=35c & CPU Package Cores=24.58W Geekbench=6,763 so it's definitely not stepping from 12. 

Plugin Type1
Console:
1/22/14 2:22:11.000 AM kernel[0]: AppleACPICPU: ProcessorId=0 LocalApicId=0 Enabled
1/22/14 2:22:11.000 AM kernel[0]: AppleACPICPU: ProcessorId=2 LocalApicId=2 Enabled
1/22/14 2:22:11.000 AM kernel[0]: AppleACPICPU: ProcessorId=4 LocalApicId=4 Enabled
1/22/14 2:22:11.000 AM kernel[0]: AppleACPICPU: ProcessorId=6 LocalApicId=6 Enabled
1/22/14 2:22:11.000 AM kernel[0]: AppleACPICPU: ProcessorId=8 LocalApicId=8 Enabled
1/22/14 2:22:11.000 AM kernel[0]: AppleACPICPU: ProcessorId=10 LocalApicId=10 Enabled
1/22/14 2:22:11.000 AM kernel[0]: AppleACPICPU: ProcessorId=1 LocalApicId=1 Enabled
1/22/14 2:22:11.000 AM kernel[0]: AppleACPICPU: ProcessorId=3 LocalApicId=3 Enabled
1/22/14 2:22:11.000 AM kernel[0]: AppleACPICPU: ProcessorId=5 LocalApicId=5 Enabled
1/22/14 2:22:11.000 AM kernel[0]: AppleACPICPU: ProcessorId=7 LocalApicId=7 Enabled
1/22/14 2:22:11.000 AM kernel[0]: AppleACPICPU: ProcessorId=9 LocalApicId=9 Enabled
1/22/14 2:22:11.000 AM kernel[0]: AppleACPICPU: ProcessorId=11 LocalApicId=11 Enabled
1/22/14 2:22:11.000 AM kernel[0]: ACPI: sleep states S3 S4 S5
1/22/14 2:22:11.000 AM kernel[0]: pci (build 18:41:15 Dec 15 2013), flags 0x61008, pfm64 (44 cpu) 0xfff80000000, 0x80000000
1/22/14 2:22:11.000 AM kernel[0]: CPUSensors: CPU family 0x6, model 0x2d, stepping 0x7, cores 6, threads 12, TJmax 91
1/22/14 2:22:11.000 AM kernel[0]: CPUSensors: setting platform keys to [k62     ]
1/22/14 2:22:11.000 AM kernel[0]: CPUSensors: base CPU multiplier is 32
1/22/14 2:22:11.000 AM kernel[0]: CPUSensors: RAPL units power: 0x3 energy: 0x10 time: 0xa
1/22/14 2:22:11.000 AM kernel[0]: AppleIntelCPUPowerManagement: Turbo Ratios CCCCCC
1/22/14 2:22:11.000 AM kernel[0]: AppleIntelCPUPowerManagement: (built 20:35:09 Oct 17 2013) initialization complete
1/22/14 2:22:11.000 AM kernel[0]: mcache: 12 CPU(s), 64 bytes CPU cache line size
1/22/14 2:22:11.000 AM kernel[0]: AppleIntelCPUPowerManagementClient: ready
1/22/14 2:22:13.000 AM kernel[0]: IOPPF: AppleIntelCPUPowerManagement mode
AICPUPMI:
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MWAIT C-States.....................: 135456
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_CORE_THREAD_COUNT......(0x35)  : 0x6000C
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PLATFORM_INFO..........(0xCE)  : 0xC0070012000
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PMG_CST_CONFIG_CONTROL.(0xE2)  : 0x1E000403
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PMG_IO_CAPTURE_BASE....(0xE4)  : 0x414
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: IA32_MPERF.................(0xE7)  : 0x8AB0AA0
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: IA32_APERF.................(0xE8)  : 0x3402FDF
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_FLEX_RATIO.............(0x194) : 0xE0000
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_IA32_PERF_STATUS.......(0x198) : 0x1B6100000C00
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_IA32_PERF_CONTROL......(0x199) : 0xC00
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: IA32_CLOCK_MODULATION......(0x19A) : 0x0
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: IA32_THERM_STATUS..........(0x19C) : 0x88360000
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: IA32_MISC_ENABLES..........(0x1A0) : 0x850089
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_MISC_PWR_MGMT..........(0x1AA) : 0x400001
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_TURBO_RATIO_LIMIT......(0x1AD) : 0x23232C2C2C2C2C2C
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: IA32_ENERGY_PERF_BIAS......(0x1B0) : 0x0
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_POWER_CTL..............(0x1FC) : 0x2504005B
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_RAPL_POWER_UNIT........(0x606) : 0xA1003
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKG_POWER_LIMIT........(0x610) : 0x69F4000E0B200
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKG_ENERGY_STATUS......(0x611) : 0x1D6A5D2A
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKGC3_IRTL.............(0x60a) : 0x0
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKGC6_IRTL.............(0x60b) : 0x0
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKGC7_IRTL.............(0x60c) : 0x0
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PP0_CURRENT_CONFIG.....(0x601) : 0x141494800007F8
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PP0_POWER_LIMIT........(0x638) : 0x0
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PP0_ENERGY_STATUS......(0x639) : 0xAFA0254
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PP0_POLICY.............(0x63a) : 0x0
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKG_C2_RESIDENCY.......(0x60d) : 0x606BDB5380
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKG_C3_RESIDENCY.......(0x3f8) : 0x6779400
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKG_C6_RESIDENCY.......(0x3f9) : 0x1E6CD84440
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKG_C7_RESIDENCY.......(0x3fa) : 0x0
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: Low Frequency Mode.................: 1200 MHz
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: Clock Speed (Max. Non-Turbo Freq.).: 3200 MHz
Jan 22 02:26:25 11s-Hac-Pro kernel[0]: AICPUPMI: Maximum Turbo Frequency............: 4400 MHz
Jan 22 02:26:26 11s-Hac-Pro kernel[0]: AICPUPMI: CPU P-States [ (12) ]
Jan 22 02:26:26 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C3-Cores [ 0 3 7 ]
Jan 22 02:26:26 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C6-Cores [ 0 1 3 10 11 ]
Jan 22 02:26:27 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C3-Cores [ 0 3 4 5 6 7 ]
Jan 22 02:26:27 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C6-Cores [ 0 1 2 3 6 7 8 9 10 11 ]
Jan 22 02:26:27 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C3-Cores [ 0 1 3 4 5 6 7 ]
Jan 22 02:26:27 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 ]
Jan 22 02:26:28 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C3-Cores [ 0 1 2 3 4 5 6 7 ]
Jan 22 02:26:28 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C3-Cores [ 0 1 2 3 4 5 6 7 8 ]
Jan 22 02:26:29 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C3-Cores [ 0 1 2 3 4 5 6 7 8 10 11 ]
Jan 22 02:26:30 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C3-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 ]

However, if I do anything that causes X86PlatformPlugin not to load like change the Plugin type to Zero, @ idle,  I jump around between 32-44x, CPU Temp=24c & CPU Package Cores=6.3-7W. Under load I stick at 44x, CPU Temp=55c & CPU Package Cores=53W and my GeekBench goes back up to 23,854. Of course the X86PlatformPlugin isn't loaded so there is no C or PStates reported in IOReg.

Type0
Console:
1/22/14 3:01:14.000 AM kernel[0]: AppleACPICPU: ProcessorId=0 LocalApicId=0 Enabled
1/22/14 3:01:14.000 AM kernel[0]: AppleACPICPU: ProcessorId=2 LocalApicId=2 Enabled
1/22/14 3:01:14.000 AM kernel[0]: AppleACPICPU: ProcessorId=4 LocalApicId=4 Enabled
1/22/14 3:01:14.000 AM kernel[0]: AppleACPICPU: ProcessorId=6 LocalApicId=6 Enabled
1/22/14 3:01:14.000 AM kernel[0]: AppleACPICPU: ProcessorId=8 LocalApicId=8 Enabled
1/22/14 3:01:14.000 AM kernel[0]: AppleACPICPU: ProcessorId=10 LocalApicId=10 Enabled
1/22/14 3:01:14.000 AM kernel[0]: AppleACPICPU: ProcessorId=1 LocalApicId=1 Enabled
1/22/14 3:01:14.000 AM kernel[0]: AppleACPICPU: ProcessorId=3 LocalApicId=3 Enabled
1/22/14 3:01:14.000 AM kernel[0]: AppleACPICPU: ProcessorId=5 LocalApicId=5 Enabled
1/22/14 3:01:14.000 AM kernel[0]: AppleACPICPU: ProcessorId=7 LocalApicId=7 Enabled
1/22/14 3:01:14.000 AM kernel[0]: AppleACPICPU: ProcessorId=9 LocalApicId=9 Enabled
1/22/14 3:01:14.000 AM kernel[0]: AppleACPICPU: ProcessorId=11 LocalApicId=11 Enabled
1/22/14 3:01:14.000 AM kernel[0]: ACPI: sleep states S3 S4 S5
1/22/14 3:01:14.000 AM kernel[0]: pci (build 18:41:15 Dec 15 2013), flags 0x61008, pfm64 (44 cpu) 0xfff80000000, 0x80000000
1/22/14 3:01:14.000 AM kernel[0]: CPUSensors: CPU family 0x6, model 0x2d, stepping 0x7, cores 6, threads 12, TJmax 91
1/22/14 3:01:14.000 AM kernel[0]: CPUSensors: setting platform keys to [k62     ]
1/22/14 3:01:14.000 AM kernel[0]: CPUSensors: base CPU multiplier is 32
1/22/14 3:01:14.000 AM kernel[0]: CPUSensors: RAPL units power: 0x3 energy: 0x10 time: 0xa
1/22/14 3:01:14.000 AM kernel[0]: AppleIntelCPUPowerManagement: Turbo Ratios CCCCCC
1/22/14 3:01:14.000 AM kernel[0]: AppleIntelCPUPowerManagement: (built 20:35:09 Oct 17 2013) initialization complete
1/22/14 3:01:14.000 AM kernel[0]: mcache: 12 CPU(s), 64 bytes CPU cache line size
1/22/14 3:01:14.000 AM kernel[0]: AppleIntelCPUPowerManagementClient: ready
1/22/14 3:01:23.945 AM SystemUIServer[213]: MenuCracker: Allowing "MenuMeterCPUExtra".
1/22/14 3:01:24.276 AM SystemUIServer[213]: MenuMeterCPU loaded.
AICPUPMI:
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MWAIT C-States.....................: 135456
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_CORE_THREAD_COUNT......(0x35)  : 0x6000C
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PLATFORM_INFO..........(0xCE)  : 0xC0070012000
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PMG_CST_CONFIG_CONTROL.(0xE2)  : 0x1E000403
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PMG_IO_CAPTURE_BASE....(0xE4)  : 0x414
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: IA32_MPERF.................(0xE7)  : 0x39C3290
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: IA32_APERF.................(0xE8)  : 0x4F2839D
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_FLEX_RATIO.............(0x194) : 0xE0000
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_IA32_PERF_STATUS.......(0x198) : 0x29F400002C00
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_IA32_PERF_CONTROL......(0x199) : 0x2C00
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: IA32_CLOCK_MODULATION......(0x19A) : 0x0
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: IA32_THERM_STATUS..........(0x19C) : 0x88350000
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: IA32_MISC_ENABLES..........(0x1A0) : 0x850089
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_MISC_PWR_MGMT..........(0x1AA) : 0x400001
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_TURBO_RATIO_LIMIT......(0x1AD) : 0x23232C2C2C2C2C2C
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: IA32_ENERGY_PERF_BIAS......(0x1B0) : 0x0
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_POWER_CTL..............(0x1FC) : 0x2504005B
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_RAPL_POWER_UNIT........(0x606) : 0xA1003
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKG_POWER_LIMIT........(0x610) : 0x69F4000E0B200
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKG_ENERGY_STATUS......(0x611) : 0x346B0945
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKGC3_IRTL.............(0x60a) : 0x0
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKGC6_IRTL.............(0x60b) : 0x0
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKGC7_IRTL.............(0x60c) : 0x0
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PP0_CURRENT_CONFIG.....(0x601) : 0x141494800007F8
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PP0_POWER_LIMIT........(0x638) : 0x0
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PP0_ENERGY_STATUS......(0x639) : 0x17DA0F34
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PP0_POLICY.............(0x63a) : 0x0
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKG_C2_RESIDENCY.......(0x60d) : 0xBFC7EF1AE0
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKG_C3_RESIDENCY.......(0x3f8) : 0x42ED9820
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKG_C6_RESIDENCY.......(0x3f9) : 0x386CE2FBE0
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: MSR_PKG_C7_RESIDENCY.......(0x3fa) : 0x0
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: Low Frequency Mode.................: 1200 MHz
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: Clock Speed (Max. Non-Turbo Freq.).: 3200 MHz
Jan 22 03:07:57 11s-Hac-Pro kernel[0]: AICPUPMI: Maximum Turbo Frequency............: 4400 MHz
Jan 22 03:07:58 11s-Hac-Pro kernel[0]: AICPUPMI: CPU P-States [ (44) ]
Jan 22 03:07:58 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C3-Cores [ 1 2 3 8 9 ]
Jan 22 03:07:58 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C6-Cores [ 1 6 7 ]
Jan 22 03:07:59 11s-Hac-Pro kernel[0]: AICPUPMI: CPU P-States [ (12) 44 ]
Jan 22 03:07:59 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C3-Cores [ 0 1 2 3 6 7 8 9 10 ]
Jan 22 03:07:59 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C6-Cores [ 1 4 6 7 8 9 ]
Jan 22 03:07:59 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C3-Cores [ 0 1 2 3 4 6 7 8 9 10 11 ]
Jan 22 03:07:59 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C6-Cores [ 1 4 5 6 7 8 9 11 ]
Jan 22 03:08:00 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C6-Cores [ 0 1 4 5 6 7 8 9 11 ]
Jan 22 03:08:00 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C3-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 ]
Jan 22 03:08:00 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C6-Cores [ 0 1 2 4 5 6 7 8 9 10 11 ]
Jan 22 03:08:01 11s-Hac-Pro kernel[0]: AICPUPMI: CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 ]

Any suggestions?

I'm not sure it's correct to do this, but try adding APLF to your SSDT. 

 

On another note, I noticed  AICPUPMI reads your max turbo freq correctly.  Mine reads 5700 no matter what I do.  Can you attach your SSDT for me to check out if you don't mind?

Link to comment
Share on other sites

I'm not sure it's correct to do this, but try adding APLF to your SSDT. 

 

On another note, I noticed  AICPUPMI reads your max turbo freq correctly.  Mine reads 5700 no matter what I do.  Can you attach your SSDT for me to check out if you don't mind?

I have tried APLF 1 & 0 with the same results. I have tried manny ssdt variations including modifying the R4E 4802 stock with correct freq vectors and the same with the stock ssdt from MP6,1. Currently I'm using non turbo values up to 3200 from stock R4E (it had nothing above 3201 in it) with turbo values (only above 3200 up to 4401) from pike in pikes SSDT. I had to add 4401 above 4400 to match what the stock r4e values were doing at the top. I don't think we're dealing with an ssdt problem. According to omni, the patched kext may not work for our 3930.
Link to comment
Share on other sites

I have tried APLF 1 & 0 with the same results. I have tried manny ssdt variations including modifying the R4E 4802 stock with correct freq vectors and the same with the stock ssdt from MP6,1. Currently I'm using non turbo values up to 3200 from stock R4E (it had nothing above 3201 in it) with turbo values (only above 3200 up to 4401) from pike in pikes SSDT. I had to add 4401 above 4400 to match what the stock r4e values were doing at the top. I don't think we're dealing with an ssdt problem. According to omni, the patched kext may not work for our 3930.

I used APLF 0x04 because I read somewhere that it equates to 1200. I'm not overclocked though so perhaps that makes a difference. Right now my system idles at 32x but with wattage equal to what I'd get at 12x and turbo up to 38x. My geekbench score equals what it did before on bios 4206 and nullcpu at 38x max turbo. I'll do some testing tonight with an OC and check the results.

Link to comment
Share on other sites

@magnfico,

 

Don't despair, there's still a possibility to get it working because the CPU cores aren't really that different.

 

 

I posted a "virgin" version of AICPM for anyone who wants to help out in the OP.

 

What I patched were routines for processor matching and then some init methods that were causing KPs because my Xeons don't have those MSRs. Obviously, if IB Xeon has them they could be used, but I'm not sure it would make a difference really.

 

The work is not done, as there's probably more things to be possibly patched, I just have limited time at hand.

 

We need to get someone with a real MacPro6,1 (nMP) to run Pike's AICPMI so we can see its MSRs and then compare to what we have.

 

Link to comment
Share on other sites

@magnfico,

 

Don't despair, there's still a possibility to get it working because the CPU cores aren't really that different.

 

 

I posted the "virgin" version of AICPM for anyone who wants to help out in the OP.

 

What I patched were routines for processor matching and then some init methods that were causing KPs because my Xeons don't have those MSRs. Obviously, if IB Xeon has them they could be used, but I'm not sure it would make a difference really.

 

The work is not done, as there's probably more things to be possibly patched, I just have limited time at hand.

 

We need to get someone with a real MacPro6,1 (nMP) to run Pike's AICPMI so we can see its MSRs and then compare to what we have.

 

 

now I do it to my brother to buy a Mac Pro.....lol

Link to comment
Share on other sites

@kundica

Our stock ssdt starts at 1200. Pikes starts at 800 and adds one extra state between all the others. That's why I used stock states up to turbo and 0. With values starting at 1200, 0=1200

Yeah, I'm using stock states, there's 27 from 1200 to 3800.  My APLF is actually "One" so I was wrong about what I said before.  

 

I just tested with an OC to 40x and it behaves like it does with turbo at 38x.  System idles at 32x but with very low wattage as if it were at 12x (like 17W) and if I stress the system it'll move up to 40x. HWMonitor at times occasionally shows idle freq jump down to 12 but barely.  IGP shows the system idle at 3.2ish.

Link to comment
Share on other sites

Does it show Unknown CPU though in the kernel log or not?

 

There's more lower level debugging needed to figure it out, most likely remote kernel debugging of the kext itself to see what values are passed around etc.

Link to comment
Share on other sites

the cpu gets recognized, so NO "unknown cpu"

 

Turboboost even works for me, without registered P-States in ioreg. 

So it seems, even if they are registered correctly, they aren't really used by AICPUPM. Who knows why not...

 

IOreg tells me, my bus-frequency is 0xfffffff. the macpro61 ioreg has a normal value for that... may that be an issue?

Link to comment
Share on other sites

Yeah, I'm using stock states, there's 27 from 1200 to 3800. My APLF is actually "One" so I was wrong about what I said before.

 

I just tested with an OC to 40x and it behaves like it does with turbo at 38x. System idles at 32x but with very low wattage as if it were at 12x (like 17W) and if I stress the system it'll move up to 40x. HWMonitor at times occasionally shows idle freq jump down to 12 but barely. IGP shows the system idle at 3.2ish.

THIS IS WRONG: AFAIK, One=1300. I think it counts up from Zero.

 

I think Zero was locking me at 1200. If I add 1100 and use One, it works like yours. AICPUPM reports 12, 32 & 44.

 

Omni's patched AICPUPM is definitely doing something good for us (3930K) in the OS. If I use stock AICPUPM I'm locked at 32x, 62w & 43c. If I add NullCPUPM to stock AICPUPM I get 12 and 32 (4802 BIOS) Idle @ 10.6w & 29c. Load @ 101w 53c and geekbench 17.4k. Similar to 4206 with NullCPU but without the turbo OC. According to wattage, heat and GeekBench, Omni's patch allows Turbo and OC to work in BIOS 4802 (Nothing else does) and as a matter of fact it appears that CPUPM is fully working BUT ONLY if I keep the X86PlatformPlugin from loading: AICPUPMI reports only 12 and 44 (HWMon only shows 32-44 & bounces around allot), idle @ 6w & 23c. Load (Prime95 torture test) @ 164-181w (from IPG because HWMon cant show anything this high), 58c & geekbench=24k. Again, this is ONLY WITHOUT X86PlatformPlugin loaded and my logic says it should be opposite.

Edited by shilohhh
Link to comment
Share on other sites

Interesting, my bus-frequency is also 0xffffffff on all CPUs in the IODeviceTree, but I wonder if this is because something is missing in SMBios.plist or not.

 

X86Plugin looks at that P-States array to at least determine the maxGuaranteedTurboPState so it must be in use.

Link to comment
Share on other sites

hm.. I even added SMbusspeed to smbios.plist but this didn't change anything. Also i tried to edit the chameleon source code, but I'm not a good coder at all..

 

bdmesg tells me:

AppleProcessorBusSpeed:

ProcessorBusSpeed (QPI): 25.4GT/s

 

But this should be 8 GT/s for my CPU i think


oh my gosh... i just installed chimera boot loader, which gives my a correct value in ioreg for bus-frequency and look:

23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MWAIT C-States.....................: 4384
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_CORE_THREAD_COUNT......(0x35)  : 0xA0014
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_PLATFORM_INFO..........(0xCE)  : 0xC10EC811E00
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_PMG_CST_CONFIG_CONTROL.(0xE2)  : 0x403
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_PMG_IO_CAPTURE_BASE....(0xE4)  : 0x10414
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: IA32_MPERF.................(0xE7)  : 0x3BB0FA4
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: IA32_APERF.................(0xE8)  : 0x44BC624
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_FLEX_RATIO.............(0x194) : 0x0
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_IA32_PERF_STATUS.......(0x198) : 0x233A00002100
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_IA32_PERF_CONTROL......(0x199) : 0x2100
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: IA32_CLOCK_MODULATION......(0x19A) : 0x0
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: IA32_THERM_STATUS..........(0x19C) : 0x88340000
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: IA32_MISC_ENABLES..........(0x1A0) : 0x850089
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_MISC_PWR_MGMT..........(0x1AA) : 0x400001
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_TURBO_RATIO_LIMIT......(0x1AD) : 0x2121212121222324
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: IA32_ENERGY_PERF_BIAS......(0x1B0) : 0x0
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_POWER_CTL..............(0x1FC) : 0x2104005B
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_RAPL_POWER_UNIT........(0x606) : 0xA1003
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_PKG_POWER_LIMIT........(0x610) : 0x684E0005E8410
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_PKG_ENERGY_STATUS......(0x611) : 0x288626B5
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_PKGC3_IRTL.............(0x60a) : 0x0
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_PKGC6_IRTL.............(0x60b) : 0x0
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_PP0_CURRENT_CONFIG.....(0x601) : 0x14149480000520
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_PP0_POWER_LIMIT........(0x638) : 0x80000000
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_PP0_ENERGY_STATUS......(0x639) : 0x1CAACC3E
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_PP0_POLICY.............(0x63a) : 0x0
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_PKG_C2_RESIDENCY.......(0x60d) : 0x17CD0EBFD4
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_PKG_C3_RESIDENCY.......(0x3f8) : 0x0
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: MSR_PKG_C6_RESIDENCY.......(0x3f9) : 0x228D462260
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: Low Frequency Mode.................: 1200 MHz
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: Clock Speed (Max. Non-Turbo Freq.).: 3000 MHz
23.01.14 04:02:31,000 kernel[0]: AICPUPMI: Maximum Turbo Frequency............: 3600 MHz
23.01.14 04:02:32,000 kernel[0]: AICPUPMI: CPU P-States [ 33 (35) ]
23.01.14 04:02:32,000 kernel[0]: AICPUPMI: CPU C3-Cores [ 5 10 11 ]
23.01.14 04:02:32,000 kernel[0]: AICPUPMI: CPU C6-Cores [ 0 3 4 5 6 7 8 9 10 11 12 14 ]
23.01.14 04:02:33,000 kernel[0]: AICPUPMI: CPU C3-Cores [ 4 5 10 11 ]
23.01.14 04:02:33,000 kernel[0]: AICPUPMI: CPU C6-Cores [ 0 1 3 4 5 6 7 8 9 10 11 12 14 15 16 ]
23.01.14 04:02:33,000 kernel[0]: AICPUPMI: CPU C6-Cores [ 0 1 3 4 5 6 7 8 9 10 11 12 14 15 16 ]
23.01.14 04:02:34,000 kernel[0]: AICPUPMI: CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 ]
23.01.14 04:02:35,000 kernel[0]: AICPUPMI: CPU C6-Cores [ 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 ]
23.01.14 04:02:41,000 kernel[0]: AICPUPMI: CPU P-States [ (12) 33 35 ]
23.01.14 04:02:42,000 kernel[0]: AICPUPMI: CPU P-States [ 12 (30) 33 35 ]
23.01.14 04:03:20,000 kernel[0]: AICPUPMI: CPU P-States [ 12 30 33 (34) 35 ]
23.01.14 04:03:27,000 kernel[0]: AICPUPMI: CPU C3-Cores [ 0 1 4 5 10 11 ]
23.01.14 04:03:59,000 kernel[0]: AICPUPMI: CPU C3-Cores [ 0 1 4 5 10 11 14 15 ]

Suddenly reported C3 cores and C6 package residency!!! Also power wattage and temperature went further down! 

 

But, my cpu speed is not properly recognized. 3,1Ghz instead of 3,0. I will now try some older chameleon versions...

 

Edit:  Chameleon r2269 works. correct speed detection and bus frequency.

Link to comment
Share on other sites

Thank you Klonkrieger2,

 

please, could you indicate which flag i have to select in Chameleon Wizard?

I think he's saying to use Chameleon Wizard to change versions. I wouldn't use it to edit your smbios. You can, but you're better off just editing it yourself to ensure it's correct.

 

Link to comment
Share on other sites

maaannn...i don't know what has happened, but after shutting down my pc yesterday I don't have c3 states anymore.

 

But if I boot in safe mode (-x) I do have c3 states??? What kind of sorcery is this?

 

Even C6 states on core 0 and 1 are not guaranteed... I have no clue what was going on yesterday.

 

Hm.. after installing chimera and rebooting, my system automatically booted in safe mode and did reset some things. (icon/application link and the warning when starting a application, that it's from the internet)

Has someone an idea what did happen??

Link to comment
Share on other sites

@omni: With patched AICPM, there is a small lag between the CPU power jumps, with the latest OG jump is much faster, also for some odd reason MSR 0xE2 is now  0x403 instead of 404, weird.
 
@fabiosun: I can bet x86Platform didn't load :)

Link to comment
Share on other sites

Did anyone else notice this ioreg entry in the original macpro61 ioreg?

"PLimitVersion"=3

I don't have this in my own ioreg, so maybe thats something important, we miss yet?

 

I disassembled the x86platformplugin and found the ascii string "pliimitversion" in here. Maybe someone with more disassembling skills could have a look at it. 

Link to comment
Share on other sites

 Share

×
×
  • Create New...