bastt41 Posted January 15, 2015 Share Posted January 15, 2015 (edited) Hi Everyone, I've been playing with OSX on PCs for some years. Recently I ran into the following issue, maybe some of the pros here can spark some ideas or other ways to address it. My build is a HP z620 workstation with dual Xeon E5-2630v2 / 64 GB / MSI GTX 760 2GB / 2 SSDs for OS' and HDD for some extra storage / latest bios. Had to patch MSI GTX760 bios with a UEFI enabled one to make things fly. No legacy boot used now. I'm on 10.10.2 Beta, latest clover, booting with darkwake=0 npci=0x2000 nvda_drv=1 kext-dev-mode=1. Not dropping any tables. Kexts loaded through clover besides the abvious fakesmc et al: VoodooTSCSync, VoodooHDA, AppleIntelE1000, GenericUSBXHCI System definition used is macpro5.1 and patched AICPM to get turbo/speedstep as I cannot get XCPM to work for now with macpro6,1... but that's something else. Now for the issue. With one CPU everything is smooth. When I add the 2nd CPU to the box I am getting freezes, pretty much similar to what can be found when searching for the OpenCL Nvidia Issue / GPU resets. I've tried different Nvidia web drivers up to the latest beta for 10.10.2 (even joined apple beta program for this), booting with CPUs=1, injecting things through clover options, more or less everything I could think of. No help - if I remove the 2nd CPU - all fixed. Here's the log (from trying to start Safari) 5/01/15 09:22:43,000 kernel[0]: NVDA: Unable to make context resident prior to start of command buffer processing!15/01/15 09:22:43,000 kernel[0]: NVDA: Calling glrKillClient for task 0xffffff80962036d015/01/15 09:22:43,000 kernel[0]: process com.apple.WebKit[476] caught causing excessive wakeups. Observed wakeups rate (per sec): 7203; Maximum permitted wakeups rate (per sec): 150; Observation period: 300 seconds; Task lifetime number of wakeups: 5703615/01/15 09:22:43,132 ReportCrash[428]: Invoking spindump for pid=476 wakeups_rate=7203 duration=7 because of excessive wakeups15/01/15 09:22:43,000 kernel[0]: NVDA(OpenGL): Channel exception! Exception type = 0x8 DMA Engine Error (FIFO Error 8)15/01/15 09:22:43,000 kernel[0]: NVDA: Channel Info: [9, 0xd, 0x9, 0x0]15/01/15 09:22:43,000 kernel[0]: NVDA: Version Info: [com.apple.GeForce, 10.0.2, 0x746c6943485b52795b4a534e5b54 , 19056461, 310.41.15f01, 1] Any ideas? Already on my list is: - Try opencl framework from 10.9.5 - edit: Not succesful - Analyse SSDT/DSDT differences between single and dual cpu - edit: Pending - Try different GFX card (not available right now) - edit:Tried a GigaByte GTX 750 TI - not addressing the issue. - Disable OpenCL (not good) - See if newer VBIOS is available (unlikely) - edit: not available and mentioning OSX with MSI not helpful (not supported). - Try different PCIe slot edit: no change as expected Rgds Bast Edited January 21, 2015 by bastt41 Quote Link to comment Share on other sites More sharing options...
bastt41 Posted January 21, 2015 Author Share Posted January 21, 2015 In the meantime I narrowed down the issue further. It seems that is has purely to do with the 2nd CPU being added - what exact issue, not sure. Maybe TSC - I tried tweaking voodootscsync to the right core#, even recompiled it myself with a diff from the source site... also, CPUs=1 do not make a different when booting with the 2nd CPU. However not only the GFX is hanging, with the 2nd CPU also VoodooHDA stops playing sound after 3-4 seconds, no log entries. Further ideas welcome... Quote Link to comment Share on other sites More sharing options...
ldis Posted May 1, 2015 Share Posted May 1, 2015 I'm using a similar system. Did you get VoodooTSCSync working for the dual CPUs? So far I can only boot with 'cpus=1' which is a bit of a waste with this system! Thanks. Quote Link to comment Share on other sites More sharing options...
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.