Jump to content


  • Content count

  • Joined

  • Last visited

About oliveiro

  • Rank
    InsanelyMac Geek

Profile Information

  • Gender
  • Location
  • Interests

Recent Profile Visitors

The recent visitors block is disabled and is not being shown to other users.

  1. Glad it worked for you, Спасибо for your post !
  2. my guess about the cpu is your power management is probably not implemented, and windowserver crashing and restarting uses the cpu a lot. another thing you can try to neutralize nvidia is renaming IONDRVsupport.kext to .bak in SLE. It should prevent nvidia drivers from loading, even earlier in boot process than nv_disable=1 does.
  3. yep "Metal WindowServer" is the name of that fundamental level you're talking about. And don't worry, I have about the same approach you have. I keep a dedicated partition for testing, and use a usb key for alternative clover booting, so that my workstation is always 100% functionnal for my daily work including safe-untouched-rocksolid bootloader settings for 10.11 and 10.12 (!)
  4. @theconnactic hey I just re-checked the config file in one of your first posts, but I find it to be a bit confusing with all those commented lines (the file you attached doesn't include Intel Injection, nor igplatformid). Did you try to recreate a simple config.plist according to the following recommandations ? Then changing only the smbios to 15.1, without any AGDC patch/edit or any Lilu kexts ? try removing anything but fakesmc.kext. One final thought, you didn't mentionned bios settings : you have Intel IGPU enabled, primary, and usual DVMT settings, right ?
  5. Same here ! respectfully !
  6. actually I mentionned smbios15.1 without patched AGDP, because on my system, from 10.13.0 to 10.13.3, it was one easy way to actually reach desktop. I know it is strange enough, because at the exact opposite of usual recommandations, but I could not boot with (my usual) smbios 14.2, facing the same symptoms Theconnactic described (windowserver crash, black screen + mouse pointer starting then crashing repeatedly), and 15.1 + patched AGDP would lead to the same situation. Smbios15.1 is using config2 in AGDP ("GFX0 unload = No", implying the unloading of all other graphics) which in that case unloads nvidia devices, leaving HD4600 alone. (main condition being that HD4600 is still named GFX0) but that's just an idea to help isolate problems, and find out whether HD4600 DVI output is a problem itself.
  7. what about smbios 15.1 ? I mean : try changing smbios to something that is actually using a config set in AGDC/AGDP. Make sure before that, that you're not using a custom DSDT (GFX renaming to IGPU for instance) and not using Lilu and its graphicsfixups neither...
  8. WindowServer now relying on Metal, we're back to the stone age of multi-gpu hackintoshes. Did you try nv_disable ? And switching smbios --assuming you're on a 14.2-- to 15.1 ? it should unload nvidia gfx, then you'll know whether DVI output is a problem in itself with your HD4600. My humble understanding is that Metal WindowServer is not very happy with nvidia drivers for older devices, even if they're properly loaded and registered, and AppleDisplay instance created. Especially with Intel IGPU. Also I'm not an expert with clover config files, I keep mine as simple as possible : ScreenResolution 1920x1080 Inject Intel Yes Ig-platform-id 0x0d220003 kernelandkextpatches AICPUPm / RTC / KernelPm / Trim BooterConfig 0x28 CsrActiveConfig 0x67 Smbios iMac 14.2 InjectKexts Yes if you need nvidia webdrivers for your 9600GT, you'll want nvda_drv=1 as a boot argument and NvidiaWeb Yes in Systemparameters, and you probably already know webdrivers have to be installed with SIP enabled. EmuVariableUefi driver is also required. To add to Cyberdevs answer, I never was aware of any kind of VGA support on real macs, but I used it on a 9600GT under Yosemite. Since you had it working with Sierra, maybe it's not the heart of the problem.
  9. Hi, I started a post yesterday about multiple gpus. I think it might be related (HD4600 based). I had same symptoms till 13.3. Did you try a clean install of 13.4 ? Also, can you check your logs : on my build the disappearing/reappearing mouse pointer was related to a windowserver crash with an error about no MTLDevice. on 13.4 it's different, as detailed on my post, HD4600 hdmi output works.
  10. Hi community ! Is anybody successful with 10.13.x and multiple GPUs on a desktop hack ? My main rig (see in signature) HD4600 + GTX550Ti + GTX560 is working perfectly under 10.11.6 and 10.12.6. Very simple to install, Intel IGPU set to active and primary in bios, inject intel and igplatformid in bootloader (enoch or clover), no nividia injection, since Fermi GPUs are natively supported. Smbios set to iMac14.2 for my Haswell CPU, and to avoid AppleGraphicsDeviceControl devices unloading. But from 10.13.0 to 10.13.3, system boots fine until login screen where WindowServer crashes with a "no MTLDevice" error (MeTaL device I guess). I could workaround that error by unloading nvidia devices in AppleGraphicsDeviceControl, but that kind of defeats the purpose, since only the IntelHD4600 can then be used. 10.13.4 brings a little improvement, desktop can be reached, all 3 gpus active (AppleDisplay instances appear in IORegistryExplorer for Intel and Nvidias) but only Intel HDMI output shows desktop correctly, both nvidias (HDMI or DVI output) only show a black desktop where the mouse pointer can be moved normally. Displays can be arranged, orientation and resolution can be changed, as if everything was fine, but it is not. Console shows the following repeated messages : WindowServer (Skylight) : Unable to composite display surface due to null compositor. WindowServer (CoreDisplay) : [ERROR] - Attempting to get capabilities from capabilities with no devices Does anybody have an idea on how to avoid or workaround these errors ? Thanks ! O. Things I've tried: Bios Graphic Devices order changes (breaks everything unless IGPU is primary) nvidia injection in bootloader (fails in enoch, disables secondary nvidia GPU outputs in clover) Smbios changes and/or AppleDeviceControlPolicy plist edits (simply enables or disables nvidia outputs) nvidia official or web drivers (no changes at all, even though webdrivers are properly loaded ; requires nvram emulation with enoch) enoch or clover (no difference except clover boots a bit faster - than enoch with nvram - thanks to proper UEFI boot I guess) lilu intel and/or nvidia and/or coredisplay graphicsfixups (no visible changes but graphics devices names, and thus enables or disables AGPM) use binaries from Sierra 10.12.6 for CoreDisplay and/or Skylight frameworks (never reached desktop) NVIDIAeGPUSupport (no changes)
  11. oliveiro

    Intel HD Graphics (2500) Ivy Bridge macOS HighSierra

    hey ! (salut à Rennes !) Don't know whether it's related to the changes in the site interface, but all the screen captures I'd added to my post seem to be broken links. I don't have my laptop with me now, I'll repost the screen captures; Let me know what you think can be relevant (I m not much of a gamer, mainly use the rig for music production, internet ...) macOS interface is extra fluid, video playing too (web or local), openGL tests went fine till 4.2 if I remember
  12. oliveiro


    Well thanks too I was about to try rolling back to a Sierra version of AppleAHCIPort.kext when I found your post. The still-buggy kext (macOS 13,3 here) was misloading my JMicron JMB363 ports (setting them as ALPM instead of eSATA) thus not mounting connected disks and delaying boot by about 5 minutes... verbose boot showing "IOAHCIBlockStorage Driver could not recover SATA HDD after 5 attempts"
  13. I know veterans from the hackintosh community claimed that Intel HD 2500 is not supported and doesn't work, but mine (mobile version) works with High Sierra, as it did back with every versions since Mountain Lion. I formerly started a topic here for ElCapitan : http://www.insanelymac.com/forum/topic/316981-intel-hd-graphics-ivy-b-hd-2500-qeci-for-osx-1011/ This should apply to any laptop with following CPU (mobile Ivy Bridge) Intel Pentium 2030M, Intel Pentium 2020M, Intel Celeron 1020E, Intel Celeron 1020M, Intel Pentium 2127U, Intel Celeron 1005M, Intel Pentium 2117U, Intel Celeron 1000M, Intel Celeron 1037U, Intel Celeron 1017U, Intel Celeron 1007U, Intel Celeron 1047UE, Intel Pentium 2129Y, Intel Celeron 1019Y, Intel Celeron 927UE I guess following Desktop CPUs owners should be able to make it work (It probably requires DSDT edits to enable the proper output, on my laptop I only use and tested LVDS) Intel Core i5-3550, Intel Core i5-3470, Intel Core i3-3220 My Laptop is an Asus (F301A-RX229H) with an Intel Pentium Mobile 2020M, which includes an Intel HD Graphics, not labelled 2500, but it has 6 execution Units like the 2500, and approximately same clock settings (650-1100) instead of 650-1150 for HD2500. You can check specs here and here. Most importantly, it is identified by bootloader and MacOsX/macOS as "Intel HD Graphics 2500" and uses HD4000 kexts. Few Graphic Glitches (windows title bar, dock...) solved by reducing transparency in Accessibility Preferences Panel.
  14. IORegistryExplorer is a useful tool that anyone should use before trying anything. Many laptops, mac or not, have keyboard and trackpad connected to an internal USB bus, so modifying system USB kexts can prevent their drivers to load. While experimenting, keep a standard USB mouse nearby, plug it and use keyboard viewer to reverse things if you loose your internal keyboard/trackpad. Should be faster than reinstalling if you backed things up as advised. Use IORegistryExplorer to find what driver is used for your keyboard/trackpad : keep in mind that the workarounds described in this post simply revert USB drivers in Sierra or ElCapitan to Yosemite versions. You just need to find the matching Yosemite drivers for your keyboard/trackpad. It's probably not that complicated.