Jump to content


  • Content Count

  • Joined

  • Last visited

  • Days Won


c0ercion last won the day on July 11 2013

c0ercion had the most liked content!

About c0ercion

  • Rank
    InsanelyMac Protégé
  1. It's on the previous page. With regard to that MST solution... it works, but I lose the ability to run multiple displays. If I put 2 or more GPUS in the machine and hook displays up to them it's random which one actually works when I get to the log in prompt. The GPUs show up in system profiler but only one of them will indicate a connected display.
  2. A few weeks ago I pulled the EDID out of that IOReg from the previous page, and it is different from my own up2414q EDID. It may be worth popping that into a display override, I simply haven't had the time or motivation to play with it lately. There are other differences in IOReg as well that may be worth reproducing in your own DSDT.
  3. c0ercion

    GTX 680 GraphicsEnabler=No

    Does it work with GraphicsEnabler=no? If so, there is no reason to turn it on. I believe GraphicsEnabler simply injects device information into ACPI tables so OS X can use cards it doesn't officially support. If you're using two cards and the other requires GraphicsEnabler you should be able to manually add it's info to your DSDT.
  4. Thanks for this. Btw, I have been struggling with speedstep for a while now as well. I have the MX version of that board and I get the same exact pstates with my 4770k. My registry dump shows that the OS is recognizing the states I generated with Pike's ssdt script (I can dig up the link if anyone needs it) but is apparently not using them. I've tried using a patched AppleIntelCPUPowerManagement kext, and it removes some errors at boot time, but doesn't fix the states. The PMPatch utility doesn't work on my BIOS... but I've loaded it up in a disassembler and found the location where I believe it is locking MSR 0xE2 (the reason unpatched AICPM.kext causes a panic). Hopefully I will have successful patch within the next few days (limited hacking time with work/gf/etc) and will report back with my results.
  5. c0ercion

    FakeSMC w/ Haswell CPU Sensor for 10.8/9

    I uploaded a new file. The first one in the list (the larger of the two) was compiled for 10.8 and does not work on 10.9. I don't have 10.8 installed on my hackintosh so I have not tested it.
  6. Thanks for the link, I'll play around with it when I get home. That still doesn't really explain why my AICPM.kext won't load properly does it? Either way I've already tried at least a dozen different smbios configs without much luck. I'm not too familiar with how SMC works though so I look forward to toying with it none the less.
  7. c0ercion


    Hey Real Deal, have you had success with that kext? Someone else with a 4670K just reported failure on the download page... so I'm curious... I might look at exploring other haswell holes in the code this evening. Unfortunately I only have the one CPU to test on. I only just noticed CPUSensors reporting 16 cores & 16 threads for your cpu... I have a few ideas of where to look. Edit: well apparently the other guy with a 4670K got it working and I just misunderstood his first post.
  8. c0ercion

    FakeSMC w/ Haswell CPU Sensor for 10.8/9

    Thanks for the feedback. FakeSMC still has some work to do to completely support Haswell according to all the to-do comments throughout the code. I'll take at look at the cpuid stuff and see if anything has changed for haswell... not only is your temperature borked, it thinks you have 10 cores. You're using a 4670K right? Are you on 10.9 DP3?
  9. I got rid of nullpm several days ago. Not in my cache or anything, checked kextstat to be sure it wasn't loaded. I meant to mention that in the original post, but I'm a little short on sleep.
  10. Is this supposed to stay loaded? If I manually load it, it only sticks around for a minute or so. It appears to load during boot but doesn't show up in kextstat immediately afterwards. I'm trying to get pstates working. I've compiled an ssdt with Pike's ssdtPRGen script, but to no avail. I only get two pstates, 8 and 35. I heard people had luck with ssdtPRGen and AICPM in combination. I even tried a kext patched for 10.9 and it doesn't seem to do anything. Is there a blacklist somewhere that I'm not aware of? Could any other kexts be preventing it from sticking around? The only mention I see of it in my logs is the following: ACPI_SMC_PlatformPlugin : :startForReal - waitForService(resourceMatching(AppleIntelCPUPower Management) timed out I have no other major problems (besides the nvidia drivers not playing very well with my gtx780... nothing serious, just getting 50fps in games I should be getting 150fps in) but other than that, my system boots fine. Oh my system log does get spammed with SMC::smcGetKeyFromIndexAction ERROR: failed on index 901 if that is relevant in any way.
  11. c0ercion


    My 4770K is also model 3C. I looked through the FakeSMC source and found that it only has model ids for ultrabook and laptop haswell cpus, not the regular LGA1150 desktop. Here's a kext compiled for 10.9 with model 3c support. http://www.insanelymac.com/forum/files/file/107-fakesmckext/
  12. Version


    Fixes CPU temperature and multiplier readings in HWMonitor and iStat on model 0x3C (LGA1150 Desktop Form) Haswell CPUs, including i7-4770K and i5-4670K. The larger of the two downloads was compiled on 10.8.4 with the 10.8 sdk. I have not tested this build. The second and smaller file was compiled on 10.9 with the 10.9 sdk and has been confirmed to work in Mavericks with i7-4770K and i5-4670K. Kozlek has since merged this patch with his master branch, you can download and compile it yourself at https://github.com/kozlek/HWSensors Wait for hwmonitor to prompt you to update, or wait for binaries to be available at https://bitbucket.org/kozlek/hwsensors/downloads diff --git a/CPUSensors/CPUSensors.cpp b/CPUSensors/CPUSensors.cpp index 3af1039..54b9d60 100755 --- a/CPUSensors/CPUSensors.cpp +++ b/CPUSensors/CPUSensors.cpp @@ -363,6 +363,7 @@ bool CPUSensors::start(IOService *provider) readTjmaxFromMSR(); break; + case CPUID_MODEL_HASWELL_DT: case CPUID_MODEL_HASWELL_MB: case CPUID_MODEL_HASWELL_ULT: case CPUID_MODEL_HASWELL_ULX: diff --git a/Shared/cpuid.h b/Shared/cpuid.h index 4bcba2f..9c17cb1 100755 --- a/Shared/cpuid.h +++ b/Shared/cpuid.h @@ -151,6 +151,7 @@ #define CPUID_MODEL_NEHALEM_EX 0x2E #define CPUID_MODEL_WESTMERE_EX 0x2F #define CPUID_MODEL_IVYBRIDGE 0x3A +#define CPUID_MODEL_HASWELL_DT 0x3C #define CPUID_MODEL_HASWELL_MB 0x3F /* Haswell MB */ //#define CPUID_MODEL_HASWELL_H 0x?? // Haswell H #define CPUID_MODEL_HASWELL_ULT 0x45 /* Haswell ULT */ @@ -457,6 +458,7 @@ static void cpuid_update_generic_info() cpufamily = CPUFAMILY_INTEL_IVYBRIDGE; break; + case CPUID_MODEL_HASWELL_DT: case CPUID_MODEL_HASWELL_MB: case CPUID_MODEL_HASWELL_ULT: case CPUID_MODEL_HASWELL_ULX:
  13. c0ercion

    QE/CI Glitchy Graphics issues on 10.9 GTX 770 and Haswell

    Web drivers can't be installed in 10.9. I have a 780 and I run with graphics enabler off, and pciroot is 0 (i don't set it, i'm pretty sure it defaults to this) and I have no glitches whatsoever. Graphics acceleration is working perfectly fine, the only problem is the drivers don't seem to fully support the card so you're not going to have a great time gaming. I can get 100+ fps in a bunch of games, and while it's playable there are some serious stutter problems and in some situations my fps is less than 1/2 or 1/3 of what it is in windows. I had GE enabled at first and experienced lots of glitches as well so I recommend turning it off. In DP1 I had to set npci=0x2000 but I leave that off in DP2 now.
  14. c0ercion

    GTX 780

    Are you running 10.9 or 10.8.4? 313.01.02f01 won't link with the 10.9 kernel. You can try to load them manually and see the linking errors in /var/log/system.log. APIs were changed, symbols were changed... it's going to require potentially serious reverse engineering and binary rewriting so I'm more interested in getting the stock drivers to recognize it properly. The stock 10.8.4 drivers are supposed to work with the 780, so I'm confused as to why I can't get the stock 10.9 drivers to recognize it properly.
  15. -f just ignores caches... my kernel cache was updated as soon as I installed the kext. It worked fine for 10 or 20 reboots (Rebooting a lot to test DSDTs) then randomly bugged out one time. It's been fine for another dozen or so reboots since. I very rarely use -f, pretty much only if I can't boot and have to boot from usb to install a kext in S/L/E.