ergot Posted September 1, 2016 Share Posted September 1, 2016 Ergot, if you are using the method I wrote out, then you are not using AICPUPM, so Null is doing nothing. You can keep it removed. If you also unlocked the MSR, then you do not need VoodooTSCSync to wake from sleep (VodooTSCSync has not been necessary to boot since 10.10.3). Thanks!! I think my MSR is locked because system does not wake up from sleep just restart so i need to use the voodooTSCSyncIs their a patch to unlock MSR or it is MB BIOS related? freqVectors is not loaded, I used your method for SMBIOS Mac Pro 6,1 using iMac 14,2 I assumed my GB results are low due to ES version of the CPU... Link to comment Share on other sites More sharing options...
ekwipt Posted September 2, 2016 Share Posted September 2, 2016 Thanks!! I think my MSR is locked because system does not wake up from sleep just restart so i need to use the voodooTSCSyncIs their a patch to unlock MSR or it is MB BIOS related?freqVectors is not loaded, I used your method for SMBIOS Mac Pro 6,1 using iMac 14,2I assumed my GB results are low due to ES version of the CPU... Sorry what's the process for smbios macpro 6,1 using iMac 14,2?I used the frequencyvector script made by pikeralpha for macpro 6.1 but not sure if it worked or not (is there anyways to test it) Link to comment Share on other sites More sharing options...
chemary85 Posted September 2, 2016 Share Posted September 2, 2016 What I appreciate with the patch PikerAlpha a better ability to OC (Turbo) a stability to the entire ordeal Hi mate, can you post your patch of 5820K? i can't reach 18K or 19K in geekbench. Thanks. Link to comment Share on other sites More sharing options...
ergot Posted September 2, 2016 Share Posted September 2, 2016 Sorry what's the process for smbios macpro 6,1 using iMac 14,2?I used the frequencyvector script made by pikeralpha for macpro 6.1 but not sure if it worked or not (is there anyways to test it) there is no process just use SMBIOS for Mac pro and when running freqvector use SMBIOS for iMac 14,2 to check if freqvector is working run sysctl machdep.xcpm in terminal and make sure you see as below: machdep.xcpm.vectors_loaded_count: 1 1 Link to comment Share on other sites More sharing options...
ekwipt Posted September 2, 2016 Share Posted September 2, 2016 great thanks Ergot I've tried your method but the frequency vector doesn't load when checking with: sysctl machdep.xcpm pretty sure the file is locked, i've tried unlocking via right click info, but i don't think it's working?? Could someone send a copy because i think i've broke the kext now Link to comment Share on other sites More sharing options...
ergot Posted September 3, 2016 Share Posted September 3, 2016 ekwipt, my system also not loading freqvectors...but it doesn't mean that the system in not working well..from my point of view 1 Link to comment Share on other sites More sharing options...
ekwipt Posted September 3, 2016 Share Posted September 3, 2016 ekwipt, my system also not loading freqvectors...but it doesn't mean that the system in not working well..from my point of view Thanks Ergot, i can't even get Pikeralpha xclm patches to work, still can't get past the +++++ sign, not sure what's wrong, theres couple of people that have broadwell-e chips working with the patch, but i know it's not properly supported yet, so maybe it just comes down to the motherboard type etc... or even the top of broad well chip itself. unsure at this point Edit i've got Pikeralpha XCPM patches working, i've read that my FAKECPUID should actually be sysctl machdep.xcpm still gives me: machdep.xcpm.vectors_loaded_count: 0 So same as you, not sure what is happening. An interesting not that when i ran ssdtPRGen.sh I now get an warning of Warning: 'cpu-type' may be set improperly (0x0402 instead of 0x0a02) Warning: 'system-type' may be set improperly (1 instead of 3) so not sure if its wrong or the script is wrong, but i will test and try again and also write to P.A. on his blog Link to comment Share on other sites More sharing options...
The Real Deal Posted September 10, 2016 Share Posted September 10, 2016 Hi, Is : <key>Boot</key> <dict> <key>Arguments</key> <string>-xcpm</string> </dict> needed with 10.12 ? I am building a rig with X99M Killer and 5820K. Thanks Link to comment Share on other sites More sharing options...
ekwipt Posted September 10, 2016 Share Posted September 10, 2016 Hi, I have a Broadway EP cpu and is working with this fakecpuid 0x040674 See in Pike blog for infos then it is mandatory for me to have 5960x brumbaer patch and a kernel patch that suggest maleorderbride search for reboot fix in clover you can solve those warning in this way set type to 0x0a02 in clover cpu part (Thanks to Pike and Needy) and the SmarUPS flagged for system type warning (Thank to TheRaceMaster) Great thanks Fabiosun that did the trick for the errors Link to comment Share on other sites More sharing options...
Absolutelight Posted September 18, 2016 Share Posted September 18, 2016 Hello all! I have the same issue how ergot " the issue with the kernel patches is sleep...when even i wake the system it auto restart, when using voodoo kext system wake up normally and still Intel power gadget show me nice PM states..., Geekbence results are same with the patch and without the patch, bottom line is that OK to leave the hack with voodoo instead of patches? or there is a patch to fix the auto restart...? " How to fix it? Or Need I UEFIPatch 0.3.9 ? Link to comment Share on other sites More sharing options...
maleorderbride Posted September 22, 2016 Share Posted September 22, 2016 Hello all! I have the same issue how ergot " the issue with the kernel patches is sleep...when even i wake the system it auto restart, when using voodoo kext system wake up normally and still Intel power gadget show me nice PM states..., Geekbence results are same with the patch and without the patch, bottom line is that OK to leave the hack with voodoo instead of patches? or there is a patch to fix the auto restart...? " How to fix it? Or Need I UEFIPatch 0.3.9 ? After testing, patching the BIOS does not help with random restarts. You can stay with Voodoo if you prefer. Link to comment Share on other sites More sharing options...
Fljagd Posted September 22, 2016 Share Posted September 22, 2016 Hi all, Just to say that everything rolls on my X99 with Sierra Methode en Français 3 Link to comment Share on other sites More sharing options...
maleorderbride Posted September 22, 2016 Share Posted September 22, 2016 Yes, mine is also fine with Sierra. This is an i7-6850K @ 4.3GHz. The second picture is an i7-6900K @ 4.0GHz. 4 Link to comment Share on other sites More sharing options...
ergot Posted September 22, 2016 Share Posted September 22, 2016 For me Sierra is also working fine (10.12.0) looks smoother and faster than 10.11.6 just waiting for CUDA support Link to comment Share on other sites More sharing options...
Fljagd Posted September 22, 2016 Share Posted September 22, 2016 Yes, mine is also fine with Sierra. This is an i7-6850K @ 4.3GHz. 5820K =>25937, 6850K =>27653 : 6850K => +6,60% 5820K => 409€ , 6850K => 690€ : 6850K => +68,7% a gain of 6.6% in performance for a higher price of 68.7%. it's expensive 6.6% Link to comment Share on other sites More sharing options...
maleorderbride Posted September 22, 2016 Share Posted September 22, 2016 5820K =>25937, 6850K =>27653 : 6850K => +6,60% 5820K => 409€ , 6850K => 690€ : 6850K => +68,7% a gain of 6.6% in performance for a higher price of 68.7%. it's expensive 6.6% It is ;p But I also need the PCI-e lanes. Thanks Intel ;( Link to comment Share on other sites More sharing options...
Fljagd Posted September 22, 2016 Share Posted September 22, 2016 It is ;p But I also need the PCI-e lanes. Thanks Intel ;( 5820K => 28 lanes, is more than enough, even for 2 CG Link to comment Share on other sites More sharing options...
maleorderbride Posted September 23, 2016 Share Posted September 23, 2016 5820K => 28 lanes, is more than enough, even for 2 CG Sure, but not enough for this workstation. Link to comment Share on other sites More sharing options...
Fljagd Posted September 24, 2016 Share Posted September 24, 2016 Sure, but not enough for this workstation. Wahou!!! I'm curious What use for your workstation Link to comment Share on other sites More sharing options...
giacomoleopardo Posted September 24, 2016 Share Posted September 24, 2016 Yes, mine is also fine with Sierra. This is an i7-6850K @ 4.3GHz. The second picture is an i7-6900K @ 4.0GHz. This is interesting. Would you share your EFI? Link to comment Share on other sites More sharing options...
bobenia Posted September 27, 2016 Share Posted September 27, 2016 hi all, is there anything special to do while upgrading from 10.11.6 for x99 i7 5930k? procedure? Link to comment Share on other sites More sharing options...
t1ago Posted October 4, 2016 Share Posted October 4, 2016 Thanks, perhaps you can go back and edit your original post with a succinct numbered list of what needs to be done? i.e. 1. Three patches xcpm_bootstrap 10.11.6 © Pike R. Alpha find 83C3BB83FB09 replace 83C3B883FB09 cpuid_set_info 10.11.6 © Pike R. Alpha find 5D7F288D48BB83F902 replace 5D7F288D48B883F902 reboot fix 10.11.6 © Pike R. Alpha find 554889E5415741564155415453504189D64189F74889FB4585FF0F84 replace C34889E5415741564155415453504189D64189F74889FB4585FF0F84 2. FrequencyVectorEdit.sh (on a MacPro6,1 SMBIOS targeting iMac14,2, I assume?) https://github.com/Piker-Alpha/freqVectorsEdit.sh 3. Remove NullCPUPowerManagement & any FakeCPUID. 4. Disable EIST in BIOS (Ai Tweaker, Internal CPU Management, Disable Intel Enhanced Speedstep) 5. Oh, and have Clover 36xx? installed. What is the minimum version? I used 3688: https://sourceforge.net/projects/cloverefiboot/ 6. Have a CPU SSDT from SSDTPRGen: https://github.com/Piker-Alpha/ssdtPRGen.sh (not strictly necessary so long as you have x86 Platform and Shim loading from some method) Edit: Ah, found another one. 7. in the BIOS, set Ati Tweaker->CPU Core Ratio to Sync All Cores, then Auto for the snyc'ed value (or if overclocking, set to your value). If you do not do this, then no turbo (when not overclocking) frequencies will be hit. With those changes I now get completely flat IPG readouts (!) while benchmarking, but my scores are still low. Multi-core are still 25% lower on Cinebench/GB with this method versus my own earlier method. Presumably my SSDT needs some love? No, I needed to do #7. It was pegged at non-turbo states without it. Edit: After finding #7, I now have the flatest performance in Cinebench/GB with IGP that I have seen to date with Haswell-EP in MacOS. Performance in Cinebench is identical to my previous method, but GB went up by 1800. I was already using a method that did 2-7, but Pike's three patches greatly aid in keeping turbo constant and flat. Thanks Pike! Hope those random reboots are gone with this method. Temps seem ~5C cooler as well. did the steps above and issue (kernel panic)... can anyone help? See picture: 1 Link to comment Share on other sites More sharing options...
maleorderbride Posted October 5, 2016 Share Posted October 5, 2016 @T1ago, that information is outdated now anyway. If you are using Haswell-E/P in 10.11.6 then I recommend the one patch method: http://www.insanelymac.com/forum/topic/315579-el-capitan-xcpm-for-x99-one-patch-solution/ Also, use MausiEthernet instead of IntelE1000e. I have yet to find an equally straightforward and robust solution for Broadwell-E/P. Since they are all a mess of kernel patches that seem likely to break with each OS update, I have fallen back to using FakeCPUID & Null with EIST disabled in the BIOS and C-states Enabled. If someone has a simple (and reliable through updates) XCPM guide for Broadwell-E/P then it would be great if they would create a thread specifically for Broadwell since what works for one does not work for another (and this thread, is meant for Haswell ;p) For those of you who PM'ed me about my 10.12 screenshots, that was the above FakeCPUID+Null+AICPUPM method that has worked since 10.9.5 with both Haswell-E/P and now Broadwell-E/P. 1 Link to comment Share on other sites More sharing options...
maleorderbride Posted October 5, 2016 Share Posted October 5, 2016 Hi So also with broad well EP you have a good lower steps and Higher steps IPG graphics with fakecpuid and Null? Yes, but I only have K-series Broadwell-EP's right now, not any Broadwell-E Xeons to test with. Make certain you have CPU Core Ratio set to Sync All Cores (or an overclock), EIST Disabled, and C-States Enabled while using FakeCPUID+Null AICPUPM method. Wahou!!! I'm curious What use for your workstation Video editing and coloring. It just has a lot of PCI-e devices. With multiple NVMe drives, two GPUs, a Decklink, a RAID card, and sometimes a RR-X, I really do need at least 36 lanes. 1 Link to comment Share on other sites More sharing options...
t1ago Posted October 6, 2016 Share Posted October 6, 2016 @T1ago, that information is outdated now anyway. If you are using Haswell-E/P in 10.11.6 then I recommend the one patch method: http://www.insanelymac.com/forum/topic/315579-el-capitan-xcpm-for-x99-one-patch-solution/ Also, use MausiEthernet instead of IntelE1000e. I have yet to find an equally straightforward and robust solution for Broadwell-E/P. Since they are all a mess of kernel patches that seem likely to break with each OS update, I have fallen back to using FakeCPUID & Null with EIST disabled in the BIOS and C-states Enabled. If someone has a simple (and reliable through updates) XCPM guide for Broadwell-E/P then it would be great if they would create a thread specifically for Broadwell since what works for one does not work for another (and this thread, is meant for Haswell ;p) For those of you who PM'ed me about my 10.12 screenshots, that was the above FakeCPUID+Null+AICPUPM method that has worked since 10.9.5 with both Haswell-E/P and now Broadwell-E/P. I applied the patch but I am low score See: https://browser.geekbench.com/v4/cpu/653574 Link to comment Share on other sites More sharing options...
Recommended Posts