search

osy / HaC-Mini

Intel NUC Hades Canyon Hackintosh support
MIT License
1.22k stars 159 forks source link

VRAM maxed out after idling for hours; choppy UI animations and video playback #75

Closed Malumen closed 4 years ago

Malumen commented 4 years ago

On restart/warm boot from Windows OS, or after hours of idling, system will max out available VRAM.

I expect the VRAM to clear quickly to free up for future use. (this could be a macOS thing)

Configuration

Screen Shot 2019-10-15 at 10 21 57 PM

osy commented 4 years ago

Likely an OSX bug.

KrisCris commented 4 years ago

I also found this issue https://github.com/osy86/HaC-Mini/issues/251. Weird though, at most times it won't lead to choppy animations, but I did experience that at some point.

osy commented 4 years ago

I've been trying to figure out if there's any incompatibilities with the GPU drivers but so far haven't found any issues. It seems like the Windows and OSX AMD drivers share most of their code so there isn't that much missing features in OSX even though Vega M isn't officially supported.

Malumen commented 4 years ago

Just adding more info: this is still happening. With the new 2.10 release, the random flickering has been fixed by switching to the ADM Codec... The problem remains of watching a few YouTube videos or Netflix or any light gaming results in what looks like the VRAM becoming maxed out and then the entire system suffers (audio and video stutters in sync, the system animations become choppy).

@osy86 if this is not happening to you, can you share your RAM specs? I don't want to but maybe I need to switch to 2x16GB of G.Skill RAM or something...

Configuration I am now using a different monitor, same problem

osy commented 4 years ago

I have 1x16GB ram. I haven’t ran into this but I also haven’t used the same workload (I don’t use Netflix or YouTube or game often). Maybe others can comment.

KrisCris commented 4 years ago

1*16GB ram.

For gaming, I have played the Minecraft, which made other apps laggy, even when scrolling webpages and typing words. (But the game runs alright)

I have experienced that several times without playing games, but they're hard to reproduce.

Youtube videos hardly make the system lag in my case, unless too many of them are playing at the same time. But opening the Chrome for too long sometimes make my system stuttering, so I have to re-open it.

Malumen commented 4 years ago

@osy86

What SSD are you using? I am wondering if it is a problem with me using an Intel 760p 1TB SSD, as this model has an 8GB SPAN as opposed to entire-disk SPAN found on Samsung 970 PRO SSDs.

osy commented 4 years ago

I have a 860 Evo SATA M2

KrisCris commented 4 years ago

I don't think this is caused by Ram nor storage. Otherwise, more people in the Hackintosh community would report it. 😂

Malumen commented 4 years ago

I may have found a fix for this for my own instance: NVMEfix

I will re-run the installer and report my findings

Malumen commented 4 years ago

After re-running the installer and adding the NVME Kernel Panic Fix, things are way worse. I tried to re-run the installer yet again but cannot get it back to the way it was. System is choppy and moving the mouse across the screen shows severe slowdown.

Malumen commented 4 years ago

Looks like at fresh boot, the SSD sits around 25-28C temps, then somehow spikes up 10C or more and THEN I see the degraded performance. Weird is that previously it would take days/hours before the behaviour showed up... Now it is within an hour or a few minutes of boot.

Also: pretty much any attempt at shutting down or restarting from MacOS results in a kernel panic; the system will reboot (instead of shut down) and then if I pick the MacOS install again, it will prompt me with the kernel panic page and ask if I want to send the info to Apple...

The kernel panic info is very similar to #22 except this only happens on a restart or shutdown. Otherwise during use it has never kernel panicked...

Using an Intel 760p 1TB NVMe SSD.

Also when the problem starts to arise, the choppy behaviour etc. the SSD read/write speeds drop to single or double digits in BlackMagic Disk test...

(Could be SSD failure but Windows HDSentinel says the drive is OK)

Malumen commented 4 years ago

Worse: kernel panics during normal operation now. Really unsure what is causing it. It's been normal forever and now suddenly the SSD temp spikes, causes the slowdown, and kernel panics every shutdown and every reboot and now random KPs during use....

Malumen commented 4 years ago

@osy86 Used the discord and CorpNewt pointed me to dump my DSDT using Hackintool, run SSDTTime and then check for IRQ conflicts and correct them. This concluded in an SSDT-HPET.aml file and some additional OpenCore patches.

Screen Shot 2020-10-10 at 10 13 46 PM

What do you think? is this something that can be added to your HaC mini ?

osy commented 4 years ago

I'm curious why it's not an issue for everyone (including me). Does this fix the issue and not introduce any new issue?

Malumen commented 4 years ago

I'm curious why it's not an issue for everyone (including me).

It must relate to the SSD; if data interactions are limited to <8GB SPAN found on my SSD, then it stays OK. Do you remember #22 ?

Does this fix the issue and not introduce any new issue?

Sorry @osy86 but how can I add them?

Also: are you booting second SSD as Windows 10? Is it normal for the Windows 10 second SSD to have no EFI folder, and instead Windows EFI is found on the MacOS drive?

osy commented 4 years ago

So you have not tested it? If you don't know how to add them, you should ask around or maybe someone else here can help.

I do not have a second SSD. I have heard countless issues others have with two SSDs. Maybe try disabling one and seeing if that fixes it.

Malumen commented 4 years ago

@osy86 I have not tested it as I am unaware how to properly add it in (precise ordering to cross over the .plists patches and add branches, and I assume I just drop in the SSDT-HPET.aml file into ACPI folder).

Here is the Kernel Panic report similar to #22

Panic Report panic(cpu 6 caller 0xffffff7f8d8ecb30): nvme: "Fatal error occurred. CSTS=0xffffffff US[1]=0x0 US[0]=0x77 VID/DID=0xf1a58086 . FW Revision= PSF121C\n"@/BuildRoot/Library/Caches/com.apple.xbs/Sources/IONVMeFamily/IONVMeFamily-387.270.1/IONVMeController.cpp:5334 Backtrace (CPU 6), Frame : Return Address 0xffffffa3d5033ae0 : 0xffffff800a9ad58d 0xffffffa3d5033b30 : 0xffffff800aae9145 0xffffffa3d5033b70 : 0xffffff800aada87a 0xffffffa3d5033be0 : 0xffffff800a95a9d0 0xffffffa3d5033c00 : 0xffffff800a9acfa7 0xffffffa3d5033d20 : 0xffffff800a9acdf3 0xffffffa3d5033d90 : 0xffffff7f8d8ecb30 0xffffffa3d5033ef0 : 0xffffff7f8d8ecdb7 0xffffffa3d5033f30 : 0xffffff800b0bb18e 0xffffffa3d5033f70 : 0xffffff800b0baf08 0xffffffa3d5033fa0 : 0xffffff800a95a0ce Kernel Extensions in backtrace: com.apple.iokit.IONVMeFamily(2.1)[E109699D-6257-3176-B081-4CC8B1C181AB]@0xffffff7f8d8df000->0xffffff7f8d91efff dependency: com.apple.driver.AppleMobileFileIntegrity(1.0.5)[1AD7D9F4-24B5-354F-BD01-C301F58FAA52]@0xffffff7f8ba8d000 dependency: com.apple.iokit.IOPCIFamily(2.9)[EF12A360-E92B-3407-8080-E4889F8AAC97]@0xffffff7f8b295000 dependency: com.apple.driver.AppleEFINVRAM(2.1)[32B99D26-4CD1-3CE5-8856-D2659CCA4861]@0xffffff7f8bb0b000 dependency: com.apple.iokit.IOStorageFamily(2.1)[DFD9596C-E596-376A-8A00-3B74A06C2D02]@0xffffff7f8b887000 dependency: com.apple.iokit.IOReportFamily(47)[769D4408-2D1B-3B65-89D1-4C3C547099E3]@0xffffff7f8b471000

BSD process name corresponding to current thread: kernel_task Boot args: alcid=11 agdpmod=ignore -disablegfxfirmware shikigva=32 shiki-id=Mac-BE088AF8C5EB4FA2 igfxmetal=1 -v

Mac OS version: 18G6020

Kernel version: Darwin Kernel Version 18.7.0: Thu Jun 18 20:50:10 PDT 2020; root:xnu-4903.278.43~1/RELEASE_X86_64 Kernel UUID: BB828521-3DD9-3ECC-9FA2-3C3ED1848228 Kernel slide: 0x000000000a600000 Kernel text base: 0xffffff800a800000 __HIB text base: 0xffffff800a700000 System model name: Macmini8,1 (Mac-7BA5B2DFE22DDD8C)

System uptime in nanoseconds: 262467827250

EOF Model: Macmini8,1, BootROM 1554.40.18.0.0, 4 processors, Intel Core i7, 3.1 GHz, 32 GB, SMC Graphics: AMD Radeon RX Vega M GH, AMD Radeon RX Vega M GH, spdisplays_pcie_device, 4 GB Memory Module: BANK 0/ChannelA-DIMM0, 16 GB, DDR4, 2400 MHz, Kingston, KHX2400C14S4/16G Memory Module: BANK 2/ChannelB-DIMM0, 16 GB, DDR4, 2400 MHz, Kingston, KHX2400C14S4/16G Bluetooth: Version 6.0.14d11, 3 services, 26 devices, 1 incoming serial ports Network Service: Ethernet, Ethernet, en0 USB Device: USB 3.0 Bus USB Device: USB Receiver USB Device: Bluetooth HCI USB Device: XBOX ACC USB Device: USB 3.1 Bus Thunderbolt Bus:

Malumen commented 4 years ago

Osy can you take a moment when you have time and tell me what your uncore voltage setting is at in BIOS? If offset is + or - ?

osy commented 4 years ago

My settings are stock except for what's highlighted in the guide.

Malumen commented 4 years ago

My settings are stock except for what's highlighted in the guide.

I understand, but I tried resetting my BIOS to stock and it jumped the Uncore voltage offset to +100mV which is weird as normal stock systems use 0 as stock uncore voltage offset...

Malumen commented 4 years ago

@osy86 I am truly an idiot: I went to the last resort of cloning my drive to another SSD (which I thought was an Intel 600p 1TB). Upon opening the USB-C enclosure... turns out I put the Intel 600p SSD in the HaC mini and the 760p in the enclosure...

I formatted the 760p as APFS, used Carbon Copy Cloner and then copied over my EFI. I cleared the NVram and then used various high-IO tests for the MacOS drive... and it works. It also starts at a higher temp (33C) and under load gets to ~52C with no slowdowns or other detrimental effects. I'll keep an eye open for any other kernel panics and apply the NVMefix if necessary but for now:

consider the problem solved.

My deepest thanks for assisting me Osy, and thank you again for putting up with my many questions and continuing this amazing project.