Jump to content

eierfrucht

Members
  • Content count

    4
  • Joined

  • Last visited

About eierfrucht

  • Rank
    InsanelyMac Protégé
  1. Three cheers to Bronya, best AMD kernel so far! And huge thanks to Andy and Sinetek for their immense contributions! Just in case, Andy's latest power management pack for AMD is available in this post: http://www.insanelymac.com/forum/topic/281450-mavericks-kernel-testing-on-amd-formerly-mountain-lion-kernel-testing-on-amd/?p=1965971 If it's possible, I would much prefer a VoodooPState.kext that would work without PStateChanger.app, just assume a default governor of 'on demand' and utilize whichever P-States were generated by the bootloader. PStateChanger is great for testing but keeping it in the tray isn't a good idea for novice users who may accidentally switch it to Conservative. P.S. Regarding VoodooTSCSyncAMD.kext from Andy, do I need to patch it manually depending on how many cores my AMD CPU has? Vanilla VoodooTSCSync.kext for Intel CPUs comes in three flavors for 2, 4 and 8 kernels...
  2. Reporting GREAT SUCCESS (as Borat would say) with Bronya's RC7_FIX kernel (version for older CPUs) on Asus М5А78L-M-USB3 / AMD Phenom II x4 925 / ATI Radeon HD5750, running Mavericks 10.9.4 perfectly with no glitches with colors / playback / clock / etc. I use Clover to patch DSDT and kexts on the fly, had some issues but finally sorted them out. Also added a couple of third party kexts to get LAN, QE/CI and onboard audio running. Sleep, reboot, shutdown all work well due to the proper real-time patches with Clover. Some curious observations: 1. Latest FakeSMC works fine, but installing HWSensors immediately results in a KP. Tried various FakeSMC / HWSensors bundles, it's all the same. Is there a way to get HWSensors running with Bronya's latest kernel? If the KP is caused by an incompatible sensor like IntelCPUSensor, how to prevent it from being installed? The system crashes right after running the HWSensors installer and keeps crashing until I restore FakeSMC.kext. 2. TSCSyncAMD.kext from Andy's AMD_PM_PACK_REL2 gets loaded and causes no trouble, but I'm not sure if it does what it's supposed to. So far no sudden lag or wildly behaving clock. 3. Andy's VoodooPState.kext and System.kext from AMD_PM_PACK_REL2 cause an endless loop of "invalid user opcode 64" during boot. Is it because bronya's kernel needs its own version of System.kext? Where could I get one?
  3. Hi! I'm preparing to install Mavericks 10.9.4 on an Asus M5A78M-L-USB3 / Phenom II x4 925 / Radeon HD5750 combo. So far I got a more or less sorted out clover.plist with the proper DSDT patches, bronya's RC7 kernel reported to run 10.9.4 on a similar CPU, a 10.9.4-specific kext for my video card and the latest fixed IOPCIFamily.zip. I would like to ask Andy, Bronya, Sinetek and the rest of the gurus a couple of stupid questions: 1. Several months ago Andy suggested merging his power management support into bronya's kernel and even submitted a diff patch; does this mean that bronya's RC7 kernel will hook up with the custom VoodooPStates.kext, System.kext and VoodooTSCSync.kext from this post: http://www.insanelymac.com/forum/topic/281450-mavericks-kernel-testing-on-amd-formerly-mountain-lion-kernel-testing-on-amd/?p=1965971 2. Do I ever need VoodooPStates.kext to get P-States working? 3. Will AppleIntelCPUPowerManagement.kext work for me? If so, should I retain both AICPM and VoodooPStates or pick either? 4. Will I be able to ditch NullCPU or whatever blocks AICPM? So far only one kernel ran for me without NullCPU on AMD but it had serious issues. 5. I have Asus Q-Fan, C1E State and AMD Cool'N'Quiet enabled in my test rig's BIOS. Booting to Linux, both CPU and chassis fans perfectly scale their RPMs according to current P-State / C-State / temperature / whatever. This is not the case with Mavericks and FakeSMC and HWSensors do not help much, but so far I only tried outdated (10.9.0-1.9.1) builds with somewhat lame kernels. AMD Cool'N'Quiet has been reported to work years ago on 10.6.2 with an older version of VoodooPState.kext, and Asus Q-Fan reportedly scales with the CPU's current P-State without any need for OS-side apps or drivers at all (provided that the CPU state tables in ACPI namespace are properly generated and the kernel is aware of them) Which option (VoodooPState vs. AICPM) should I pick to get at least some form of CPU fan scaling in Mavericks? I'm not asking for the same I'm getting on Linux, but CPU fan running at full speed is driving me crazy. Clover devs suggested that certain Macs like MacPro mostly rely on P-States for power management while iMacs use C-states, and Hackintoshes behave depending on their SMBios model ID. Suggestions were made like totally disabling either P- or C- state generation in Clover to see if that helps get native power management working as supposed. But all this was said regarding Intel CPUs... I'm totally confused. I spent hours on the web but was never able to find any technical explanations regarding AMD Cool'n'Quiet and Asus Q-Fan, I can't even say for sure which value either is supposed to scale according to! Am I right to suggest that Cool'N'Quiet is what actually throttles down the CPU and Q-Fan is what should scale the fans' RPM accordingly whenever a CPU throttle-up or throttle-down is detected? And C1E is simply another (and not universally supported) CPU C-State just to save a tad more power should the OS kernel know how to properly handle it? I second that question! But could it be "from be8f0000 to be6f0000" then "00ba7000 to 00ba9000" like two consequent patches, and not a single one? If I correctly understand there are two unrelated blocks of bytes to patch, either should be looked up and replaced individually. I also saw guys mostly use BASE64 when specifying kext patch binary data in Clover.plist, are you sure your current Clover build won't misinterpret plain hex? Clover is notorious for often getting tiny undocumented changes with huge consequences. The most current Clover manual from Slice is somewhere around 2706, and the current Clover itself is 2770-something. Just for giggles, try two separate patches for IOPCIFamily noted as below: replace vo8AAAA= with vm8AAAA= replace unAAAAA= with upAAAAA=
  4. AMD Kernel For Mavericks 10.9/10.9.1/10.9.2/10.9.3

    Hi! Am I right to assume that any AMD kernel that supports SSE4 emulation can run with the native AppleIntelCPUPowerManagement.kext? Does your latest kernel support it? I'm about to give it a try with a Phenom II x4 925 on an ASUS M5A78L-M/USB3 motherboard and so far I have found the proper configuration for the Clover bootloader so it produces a nice DSDT, but I desperately need the native power management to control the CPU fan rate. Riz_vista claims his kernel works well with the M5A78L-M and Phenom II x4, but I'm not sure if it supports native power management... Can I use your AMD kernel for Mavericks without NullCpuPowerManagement.kext and with the native AppleIntelCPUPowerManagement.kext?
×