WhatEverGreen Support Topic

telepati · September 3, 2019

@jsl2000 I tried cables but it didn't help.

I think this is an EDID problem. I create a new EDID and choose Injection Apple iMac Retina Display (16:9) 1080p for my HP 2711x non K monitor. Now, this boot glitch is gone. I am not sure this is a proper way to fix but it helped.

exquirentibus · September 3, 2019

23 hours ago, bobpedro said:

@HighSierraWannabe I ordered this adapter from Club3D as it is one of the only adapters available which is capable of a full 2560x1600 resolution. It arrived today, and after some configuration I was able to get the full resolution to work. The only issue I noticed is some blue flickering on black screens (noticeable on boot and youtube)... I think I will give the Apple (MB571Z/A) Adapter another try (I can lend one from work) as it might be superior to the Club3D adapter... but then an additional miniDP -> DP adapter is necessary - not sure if its a good idea...

I am on Mojave 10.14.5 now, but I want to perform the update in the next few weeks.

I think the problem exists for all RX cards, as @jsl2000 has it with the RX560, I have it with RX570 and @TheBloke with a RX580.

EDIT1: After uninstalling SwitchResX the blue flickering is gone, might have been related to this. Not sure if its worth trying the Apple (MB571Z/A) Adapter then...

EDIT2: I noticed some small red dot flickering in terminal window, verified by making it fullscreen. Seems like the adapter has some issues with black or the disorder is only noticeable there...

EDIT3: Today I tried the Apple (MB571Z/A) adapter at from with a miniDP -> DP adapter and it works perfectly. No more flickering and a lot more supported resolutions to choose from. The Club3D Adapter goes back, MB571Z/A ordered on eBay.

Excellent! So there is a DVI solution.

Any idea of the best RX 570 or RX 580 that has two DP's or a DP & mini DP? I'll also need a working HDMI - or yet another DP or mini DP for 3 monitors.

Or, should I go with two video cards in my Hackintosh (again, Which card is 'best')?

I just 'repurposed' two nVidia GTX 1050's that I had intended to use in a Hackintosh - to a Win10 box. Running two cards is no problem at all in Win10, but ... I have no idea of the best solution for a Hackintosh.

Mojo28 · September 8, 2019

Hi all,

I am struggling now for two weeks to get my card working properly - I have Sapphire HD7850 2GB OC and cannot seem to get hardware acceleration working. Screenshot attached together with my clover.plist.

I am also attaching report from Lilu and WhateverGreen debug versions.

These are my specs:

Asrock Z77 Pro3
i5-2500k
Sapphire HD7850 2GB OC

I am using these kexts:

WhateverGreen
Lilu
AppleALC
RealtekRTL8111
USBInjectAll
VirtualSMC

I tried all ports (2 mDP, HDMI, DVI), tried ati injector, without ati injector, radpg boot flag but nothing seems to work.

Do you have any idea what else could I try?

config.plist

debug.log

Edited September 8, 2019 by Mojo28

Slice · September 9, 2019

On 9/8/2019 at 9:30 AM, Mojo28 said:

Hi all,

I am struggling now for two weeks to get my card working properly - I have Sapphire HD7850 2GB OC and cannot seem to get hardware acceleration working. Screenshot attached together with my clover.plist.

I am also attaching report from Lilu and WhateverGreen debug versions.

These are my specs:

Asrock Z77 Pro3

i5-2500k

Sapphire HD7850 2GB OC

I am using these kexts:

WhateverGreen

Lilu

AppleALC

RealtekRTL8111

USBInjectAll

VirtualSMC

I tried all ports (2 mDP, HDMI, DVI), tried ati injector, without ati injector, radpg boot flag but nothing seems to work.

Do you have any idea what else could I try?

config.plist

debug.log

See messages in your log

2019-09-01 23:51:10.568411+0200  localhost kernel[0]: (kernel) WhateverGreen:     weg @ IGPU device (00:02.00) has device-id 0x0112, you should change it to 0x0116
2019-09-01 23:51:10.575741+0200  localhost kernel[0]: (kernel) WhateverGreen:     weg @ IMEI device (00:16.00) has device-id 0x1E3A, you should change it to 0x1C3A

Mojo28 · September 9, 2019

The IGPU message seems was coming from igfxsnb=1 boot argument...

I removed it and also seems that I managed to get rid of the other warning:

2019-09-09 18:25:55.761803+0200 localhost kernel[0]: (kernel) WhateverGreen: weg @ IMEI device (00:16.00) has device-id 0x1E3A, you should change it to 0x1C3A

2019-09-09 18:28:35.605336+0200 localhost kernel[0]: calling mpo_policy_init for Lilu

2019-09-09 18:28:35.606464+0200 localhost kernel[0]: Security policy loaded: Lilu Kernel Extension 1.3.8 DEBUG build (Lilu)

2019-09-09 18:28:37.649370+0200 localhost kernel[0]: (kernel) WhateverGreen: weg @ IMEI device (00:16.00) has device-id 0x1C3A0000, you should change it to 0x1C3A

2019-09-09 18:33:43.605995+0200 localhost kernel[0]: calling mpo_policy_init for Lilu

2019-09-09 18:33:43.607123+0200 localhost kernel[0]: Security policy loaded: Lilu Kernel Extension 1.3.8 DEBUG build (Lilu)

Seems that from my last boot (18:33) I don't get any warning anymore.

Didn't help however

Edited September 9, 2019 by Mojo28

TheBloke · September 10, 2019

On 9/2/2019 at 9:06 PM, bobpedro said:

I think the problem exists for all RX cards, as @jsl2000 has it with the RX560, I have it with RX570 and @TheBloke with a RX580.

Actually I'm on an AMD 7970Ghz / R9 280X. I was thinking about getting an RX580 but haven't so far.

Also, in my case it is not specifically related to the DVI ports. My card has 2 x DVI + 4 x DP ports. I have 5 monitors connected - 2 x DVI, 3 x DP, and I always get a picture on only two monitors. Even if I unplug both DVI ports, I would still only get a picture on 2 out of the 4 DP ports. The monitors with picture are the same two monitors that show the BIOS/Clover boot screens. Also, since 10.14, monitors without picture now connect/disconnect rapidly, and show weird grey 'static'. In 10.13 they were just black.

Now I am thinking of getting a Vega 64 instead, as they have 6 outputs (3 x HDMI, 3 x DP), and are faster than the 580. I am currently researching to see if such a card will work on my legacy boot system. I fully anticipate that I will have the same problem, and require to sleep & wake after boot. But as long as I can do that and get all monitors working, that will be OK.

It's so weird that we all have this problem, with varying cards. I was sure it was related to legacy boot, but you are not on legacy. And I really don't think every RX owner has this issue, else it would be discussed far more, surely?

Also, regarding your use of DP->DVI adapters - is this necessary just because you need dual-link DVI, with higher than 108p or 1920x1200 resolution? If you had a lower resolution monitor, would HDMI -> DVI passive cable work?

I ask because the Vega 64 I am thinking of getting has 3 x HDMI + 3 x DP, and I was hoping to use it with 3 x DP monitors + 2 x HDMI->DVI passive cables. I have two monitors that are DVI only, both 1920x1200 resolution. I do own an active DP -> DVI adapter, but last time I tried it, it only worked intermittently; the monitor would work for a while, then suddenly rapidly disconnect/connect/disconnect.

I had more success with an active DP -> HDMI adapter, which worked fine. But it's DVI I would need. However, two of my monitors support both DP and HDMI, so I would have the option to run: 1 x DP, 2 x HDMI, 2 x DP -> DVI adapter, if I can find one that doesn't disconnect as my (cheap, off Amazon) one did. Maybe I should try one of the ones you mention.

How is that you can use the Apple mini DP adapter? I thought all RX580s had full sized DP ports? Do you first have a full size DP to mini DP cable, then put the mini DP into that?

Edited September 10, 2019 by TheBloke

kuruu · September 12, 2019

I am struggling here. I have a ASRock Z390 ITX/AC with a Vega 64 reference card. All works normally with displayport and a 4k@60 monitor, except that occasionally, maybe once or twice a week, the screen will suddenly go black.

When this happened this morning, I was just browsing, no heavy load. Side panel happened to be off, so getting plenty of air.

When this happens I can still connect to network services like ssh, but screen sharing does not work indicating that the display driver has shutdown.

I have whatevergreen 1.3.2 and Lilu 1.3.8 active.

Troubleshooting has been a challenge because I cannot trigger it and it occurs so rarely, so any fresh thoughts would be appreciated.

Is there a boot argument or a setting in my config.plist that I am missing? Please see attached config.

FWIW, if I boot with an HDMI cable, it goes black after the apple logo, though HDMI works for windows gaming.

config.plist

maclinuxG4 · September 14, 2019

strange, but with nojpeg, i have no troubleshoot with jpeg.

when WEG merge he kext, it doesn 't provide all functionnality.

so i propose a test to debut to ensure, that the activation is running to a test.

Thank to propose an arg on boot and see on a log the activation or running.

Edited September 14, 2019 by maclinuxG4

TheBloke · September 15, 2019

On 9/10/2019 at 6:10 PM, TheBloke said:

Now I am thinking of getting a Vega 64 instead, as they have 6 outputs (3 x HDMI, 3 x DP), and are faster than the 580. I am currently researching to see if such a card will work on my legacy boot system. I fully anticipate that I will have the same problem, and require to sleep & wake after boot. But as long as I can do that and get all monitors working, that will be OK.

So I bought the Gigabyte Vega 64 (GV-RXVEGA64GAMING OC-8GD). I have spent approximately 8 hours today trying to get it to work. And I have.. mostly. And in a somewhat annoying way.

The TLDR of this is: On my legacy boot, no-iGPU system, Vega 64 still requires sleep&wake to get a picture on monitors, and it is even worse than with R9 280X because now I can only sleep if I boot macOS with no monitors connected at all. But if I do boot without monitors connected, then log in via SSH and put the system to sleep, then attach monitors and wake, then I do get a picture on all 5 monitors.

Here is the full story of my many, many hours of testing, constant reboots, crawling around connecting/reconnecting cables, and generally pulling my hair out..

Booting into macOS with any monitors connected will not work. All displays are always black. They may or may not have signal, depending on the exact settings tried. Sometimes they have signal then flick on and off.
This is quite similar to my situation with the AMD 7970Ghz (R9 280X), except with the 280X I could usually get a picture on two monitors, and I could then sleep & wake and immediately get a picture on all monitors.
I tested with many permutations of settings/options: WEG on and off, Clover InjectATI on and off, different AGDP modes, and more.
With the Vega 64, the only way I have found to do a successful sleep is if I boot into macOS with zero monitors connected.
1. Otherwise, the Console logs are full of many errors
  1. the "CRITICAL ERROR : VBLANK interrupt has not been generated in time!" that I also got with the 2980X. But also:
  2. kernel[0:79f] (AMDSupport) --> AMD Resyncing LB:1. Iteration: 2.
  3. kernel[0:7a9] (IOAcceleratorFamily2) AMDRadeonAccelerator [2]: transaction ID (3) generated error (0xe00002d5). dirtyBits=0x80, options=0x0
  4. kernel[0:472] (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
  5. kernel[0:5e9] (IOAcceleratorFamily2) virtual void IOAccelEventMachineFast2::checkGPUProgress() - Signaling hardware error on channel 5..
  6. kernel[0:5e9] (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartSignaled stampIdx=5 type=2 prevType=0 numStamps=22
  7. kernel[0:5e9] (AMDRadeonX5000HWLibs) AMD Cail: [6:0:0] GPU HangState 0x00000040, HangFlags 0x00000005: IndividualEngineHang 1, NonEngineBlockHang 0, FenceNotRetired 1, PerEngineReset 1, FullAsicReset 0
  8. And more like like that.
2. As a result, I cannot put the machine to sleep. Accessing via SSH and running "pmset sleepnow" will put the displays to sleep, but the PC does not shutdown, until eventually (after 2-3 minutes) I can tell that it has crashed (I hear a noise as my USB sound card shuts down), and if I try to ping the PC or SSH in again, it fails.
If I boot with no monitors connected, then I can SSH in, put the machine to sleep, reconnect monitors, wake up, and then I get a picture on all.
I found that I needed to add -rad24 boot argument, otherwise my monitors were detected as having 30-bit colour. This was fine for most of them, but my 4K monitor would only connect at either 2048 x 1080, or else it would connect at 4K but only at 30Hz. Adding -rad24 fixed this issue.
The good news: if I do this long, annoying bootup process, then I get a picture on all 5 monitors; 3 x DP, and 2 x HDMI -> DVI (using passive HDMI->DVI cables).
1. I also have -raddvi boot argument specified, which I believe is needed but I haven't 100% confirmed (when I was doing earlier testing, before I got things working, I only got a signal on the DVI monitors with this flag added.)
GPU is detected in About->System, showing Metal support and showing 8GB RAM.
GeekBench 4 Metal score is 147k, which I think is OK? Not 100% sure on that yet, but looking at the Geekbench Browser, most scores for the Vega 64 are around this level (although there are also some much higher.. I need to investigate that.)
EDIT: My highest GeekBench 4 Metal score is 203k, and CL is 196k. It seems that enabling Clover's InjectATI has made a big difference to the score. When booted with only WEG, score was 147k as mentioned above. I tried rebooting with WEG + InjectATI, and this had four effects:
1. About This Mac shows GPU is listed as "AMD Radeon Vega 64 8GB" (WEG+InjectATI) instead of "Radeon Vega 64 8GB" (WEG only)
2. About This Mac shows says "Metal: supported" (WEG+InjectATI) instead of "Metal: Supported, feature set macOS GPUFamily 2 v1" (WEG only)
3. Geekbench 4 score increased significantly, now to 203k in Metal benchmark (WEG+InjectATI), compared to 147k (WEG only).
4. After the fans first spin up (eg after doing Geekbench benchmark), they then stay at 100% permanently. This has been a known issue with the Vega 64 in the past, but I had read that it was meant to be fixed by 10.14.5, which I am on. So I don't know why I still have this issue - another thing I need to research! I have read that maybe the Gigabyte Vega 64 OC cards have a different fan controller, which could perhaps explain the difference.
I don't quite understand why InjectATI seems to have made such a difference to performance?

This has been a long and frustrating day, but I do now have a mostly working system.

I do still have one issue though - other than the very annoying boot up routine: I do not have any hardware h264 or HEVC/h265 encode/decode. I've checked this with a couple of apps such as MacX Video Converter, which show no HW support. And when I play videos in VLC or in the web browser, I can see very high CPU usage - eg 300-500% CPU for a 4K video. I didn't have working h264 with my R9 280X either, but I didn't know at the time if it was possible to have that. I know for sure the Vega 64 should provide this in macOS, from info I see from other people.

I have not yet started to research to see if I can fix h264/h265 HW encode/decode. If anyone has any info on getting that (in a legacy boot system with no iGPU), I would be very grateful for any info!

Edited September 16, 2019 by TheBloke

TheBloke · September 16, 2019

Regarding the "no picture until sleep & wake" issue we've been talking about: I have another instance of it.

Now that I have my Vega 64 in my main hack, I put my 7970Ghz (R9 280X) in my second hack, which is an H77 system (Asus P8H77-M) with UEFI boot.

And I got exactly the same issue. On boot, I can only get a picture on two of the four DP ports, and not either of the DVI ports - monitors on other ports show signal, but are black. Console logs shows "CRITICAL ERROR : VBLANK interrupt has not been generated in time!" and other errors. Activity Monitor shows kernel with high CPU usage, 100+%, making the system sluggish.

After sleep & wake, all of these issues go away: I can get a picture on all monitors connected; Console errors stop; high CPU usage stops.

So yes, definitely not related to legacy boot as I once thought. I did try playing around in the BIOS on the H77 machine, changing settings relating to UEFI/legacy boot, eg I changed it to look for only UEFI OpRoms, not legacy. But that made no difference.

So now I have two systems that I have to SSH into and put to sleep before I can see anything!

It's a bit of a mystery: several of us have this issue where we must sleep & wake after boot, with a variety of AMD cards (eg R9 280X, RX580, Vega 64). But most users don't have this issue, and so it is not much talked about. I will continue to research when I have time.

Edited September 16, 2019 by TheBloke

dragonmel · September 16, 2019

@TheBloke

I am having many of the same issues

I am trying to upgrade my system from a fully working and more stable than my real macs setup using clover in the 4300 range... with a patched DSDT that was good enough for me to run essentially with all options off in clover 2 monitors 1 a 4k@60 TV .. it all just works great..

upgrade my gtx600 for a MSI VEGA 56 airboost and its been a fight from the word go

initial install I believe was 10.14.5 or base .6... I started getting some success but after either the .6 and or the supplemental updates its been a bigger train wreck...

most time booting to black screen. with HDMI and DP to my 4k .. nothing at all..

the update took forever , hung many times , might have even panicked a few times..

I can't get ANY picture on the HDMI , even if its the only one connected now..

I use a Mac Pro 5,1 smbios as my hardware is almost identical..

I have seen posts elsewhere of real Mac Pro 5,1 users having all kinds of issues like this as well with the latest updates

I think this is another example of apple bricking or making hardware seem outdated JUST BEFORE launching a new mac pro to force people to open their wallets...

I have a real macbook 5,1 that I was installing mojave on with dosdudes solution.. again.. 10.14.5 worked great.. updated , and my airport card is bricked.. shows as no hardware installed and nothing gets it back

no sleep, no display sleep, running really hot fans going like crazy.. tried running the combo updating on it again.. no go.. still running terrible.. downgraded back to el cap and its much better, but airport card still bricked... none of the dozens of fixes that usually bring it back are working.. by all accounts its like its not even installed..

I am going to try an run the installer for 10.14.5 to see if it downgrades my install on the mac pro 5,1 hack... I think the g95 supplemental is beyond fixing.. and I doubt with catalina just days away we will see any more point updates on mojave.. forcing people to upgrade to catalina, get away from 32bit arch, and having to buy lots of hardware and software along the way

TheBloke · September 17, 2019

So I spoke too soon when I said I had a working solution with my Vega 64..

My main 4K monitor will now no longer connect at 4K 60hz. It's either 4K 30hz, or 2048 x 1080. In other words, it seems like it's connecting as DisplayPort 1.0 or 1.1, not 1.2. I've confirmed the monitor connects fine at 4K 60hz in Windows 10.

I had this problem when I first managed to get a picture, but it went away as soon as I added -rad24, which made me think it was related to the GPU wrongly detecting the displays as 30 bit. Now suddenly the problem has come back, and won't go away, and I really don't understand it.

Last night when I went to bed, I decided that rather than leave my PC on all night as I usually do, I would try sleeping it for the night. I don't normally do that because I've found it just wakes up. But last night I tried again, and it did go to sleep fine. I came back this morning, woke it up.. and found my 4K monitor would now only connect at 2048x1080. I've rebooted multiple times, I've tried power cycling the monitors, I've tried connecting the 4K to all three of the DP ports on the GPU. Nothing will get it back to 4K 60hz.

I just don't understand why it worked fine for 48 hours - including over several reboots - now suddenly won't work at all. This is driving me crazy!

Right now I'm trying upgrading to 10.14.6 (on a backup SSD) to see if that makes any difference, but I hold no hopes at all at this stage.

dragonmel · September 17, 2019

@TheBloke

I have found that any updates past 10.14.5 are only going to make those problems worse

I kinda had my dual monitor (HDMI 1980x1200 + DP 4k@60) working in 10.14.5 with sleep and wake. most of the time

upgrading past that introduced a host of issues as listed in my previous post

again the real mac pro 5,1 crowd using vega are having same issues with the 10.14.6+ supplementals .. the update comes up in black screen due to the loginwindow/setup having code issues

then after using an old graphics card to complete the update.. going back to the vega only to find that HDMI no longer works at all and a host of other issues..

its almost like they are trying to brick the 5,1 so that they will buy the new 'cheese grater' in October

TheBloke · September 17, 2019

I've finally got it back at 4K@60hz. This has been one of the most annoying problems yet.

I don't know what causes it to fail to offer 4K@60. I don't know what causes it to start working again. I'd reboot five times in a row and it would always come up at 2840x1080, with no option for 4K. Then the next time it would be 4K@30hz. Then eventually it would come up in 4K @60hz.

At first I thought this was being affected by things I was testing. For example, I booted with -wegoff and I immediately got 4K@60. I rebooted with WEG back on, and it was back to 2840x1080. I rebooted with -wegoff again, and it went back to 4K. So it seemed like -wegoff made it work. Until some time later, I booted with -wegoff and got no 4K.

10.14.6 made no difference, either positive or negative. I also tried switching to the MacPro 6.1 SMBIOS. Actually the first time I booted into MP 6.1, I got 4K. But I haven't tested further with that, so it's quite likely that it would have broken again later.

Actually it feels a bit like that making any change will bring back 4K - ie if I kept rebooting when it wasn't working, it would never start working. But as soon as I changed something, like turning WEG off, or switching SMBIOS, or even enabling InjectIntel in Clover (which I did by mistake, and shouldn't affect my system at all as I have no iGPU), the next boot I would have 4K@60 again.

Right now it's working and so I'm simply not going to turn the PC off until I absolutely have to. It's hardly a great solution, but right now it's the only one I have..

TheBloke · September 17, 2019

50 minutes ago, dragonmel said:

@TheBloke

I have found that any updates past 10.14.5 are only going to make those problems worse

I kinda had my dual monitor (HDMI 1980x1200 + DP 4k@60) working in 10.14.5 with sleep and wake. most of the time

upgrading past that introduced a host of issues as listed in my previous post

again the real mac pro 5,1 crowd using vega are having same issues with the 10.14.6+ supplementals .. the update comes up in black screen due to the loginwindow/setup having code issues

then after using an old graphics card to complete the update.. going back to the vega only to find that HDMI no longer works at all and a host of other issues..

its almost like they are trying to brick the 5,1 so that they will buy the new 'cheese grater' in October

OK thanks for the info @dragonmel . I guess it's good to know that real Mac users have issue as well.

However I know plenty of people are using Vega 64s and don't have these problems, so I really wish I understood what the cause of it all was. It's definitely not related to the MP 5.1 SMBIOS specifically, as I just tried 6.1 and I still didn't get any pictures on any monitors from boot, nor did I have working h264/h265 HW accel. Don't know yet if it made any difference to the 4k@60 issue; first time (and only) time I booted with 6.1 SMBIOS, I did get 4k@60, but as it didn't seem to improve on anything else, I haven't yet tested further.

Right now my theory is that the reason I have all these problems is perhaps because I have no iGPU. Most people these days have an internal GPU, and most Macs expect them to be there.

@bobpedro any update from issues with the RX580? And could you tell me what system you're running? Do you have an internal GPU?

dragonmel · September 17, 2019

@TheBloke

well thats the thing really...

keep in mind that apple is having issues with its own hardware.. and its native drivers.. hence the update

then compound apples issues that they can't get right with a controlled enviromentent of hardware that they control and select

then apple having to go in a make highly specific improvements and optimizations to get THEIR hardware combos to work... in the process likely taking more generic drivers found in the earlier builds and making them more apple hardware specific

considering that hackintoshing is all about masquerading hardware to look like something else.. further compounds the issues

searching outside the hackentosh sphere ... you can quite clearly see many many threads of windows and real mac pro users having issues with VEGA cards and in lesser cases 580s etc.. VEGA has not exactly been a trouble free product for AMD

more troubling is apples whole attitude to OS X since the tick/tock adoption of OS X releases...

now that catalina is on the cusp... Mojave will be abandoned.. a handful of security patches done by a skeleton crew of less capable programmer will continue for a time.. but their attitude at this point even if this latest release is garbage will be ... well they can upgrade to Catalina ... and if their hardware doest support it.. win for apple ... as you will just upgrade

cook has been taking apple (as a bean counter would) much more toward a microsft mentality of .. bugs are good for business and programmed obsolescence is now pretty much industry standard across many many products

apple use to build fantastic bullet proof hardware... I am writing this on a 2008 macbook pro.. still as good today running 10.11 as it was running what like 10.5 when it came out.. nothing wrong with it.. nothing faded, rubbed off, broken, .. I could wipe it down, put it back in the box (i still have) and could practically sell it as new

now MacBooks built in the COOK age.. well lets just add it up... horrible thermals, motherboard and GPU failure rates through the roof, BATTS, CPU, RAM, and for fucsake STORAGE---STORAGE.. welded to the board... have a failing SSD... loosing all your data.. too bad.. pay up for a new computer...

my macbook I have upgraded ram, replaced batteries and removed my optical drive to run dual SSDs in.. try that with todays $3000 mac pro.. you cant

additions of T2 and other encryption and security chips will soon (if not already) begin closing the loopholes.. dosdude has a patch right now for MacBooks like mine that want to run mojave that will help exclude some T2 code check in the latest updates... and thats on real macs... they have been been pushing more and more firmware updates lately with software updates again closing the environment down bit by bit...

TheBloke · September 18, 2019

PROGRESS!

I posted about my many issues on the Olarilla forum, and @MaLd0n has immediately made things a lot better for me, at least on my X58 system.

The answer: setting my SMBIOS to iMacPro 1.1. It honestly would never have occurred to me to try an SMBIOS as late as this on my old X58 system. But in hindsight, it makes sense: the iMacPro has no iGPU, and is probably pretty unique in that nowadays. And the more I think about it, the more I am sure that the lack of iGPU in both my X58 and H77 systems is the key to all these problems.

Anyway, changing my X58 to iMacPro 1.1 has made the following improvements:

I now have HW encode/decode, on both h264 and h265. Playing a 4K video in VLC now uses 15% CPU, instead of 500% !!
If I boot with only one monitor connected to DP1, that monitor will work perfectly - no black screen, no errors in Console, no high CPU usage, sleep/wake works fine. Absolutely perfect.
- So for anyone who only has one monitor, this would be an entirely working situation.
Unfortunately I still cannot successfully boot with more than 1 monitor connected. If any other monitor is plugged in, even if the monitor is off, then I get black on all screens, and I cannot put the system to sleep.
As before, once I have done sleep & wake once, I can then use all ports; all monitors get a picture.
Therefore, my bootup process is now:
- At Clover screen, unplug all monitor cables except DP1
- Boot into macOS
- Sleep
- Reconnect my other 4 monitors
- Wake
- Then everything is perfect.
My other issue was the 4K @ 60hz monitor, and that also seems to be mostly resolved. Since I switched SMBIOS I haven't ever had it come up at only 30hz or only 2840x1080.
- I have had a different issue, where the monitor is detected at 4K @ 60hz but I get no picture. But I think I've resolved that now by putting the 4K monitor on DP1, so it's also the boot up monitor. Even if the issue does happen, it seems to be resolvable by unplugging and replugging the monitor a few times.

My new boot up procedure still sounds quite time consuming, and it is. But it is hugely better than the situation I had before, where I could not get a picture on any monitor. Now at least I can easily put the system to sleep with the mouse, rather than needing to SSH in. And applying upgrades and things like that is so much easier when you can actually see a picture on one monitor!

So I have gone from majorly annoyed to now being quite happy with my Vega 64 setup. In particular, getting HW encode/decode is a major step forward. I've literally never had that working in macOS before.

Unfortunately this SMBIOS change did not make any difference on my second system, the H77 with my old 7970Ghz / R9 280X GPU. On that system I need to sleep & wake in order to get a picture on both monitors (though at least I can see a picture on one from boot). And sadly I still don't have h264 encode/decode available. EDIT: further research has revealed that it's impossible to get HW accel on the R9 280X. The drivers just do not support it. Source is a long MacRumours thread where real Mac Pro users are using Lilu and Whatevergreen to enable HW accel on recent AMD GPUs. Their testing shows that the earliest GPUs with supported HW accel is Polaris / RX 400 series.

I will continuing researching on black screen on boot errors and will post again if I do get any further.

Edited September 18, 2019 by TheBloke

maclinuxG4 · September 18, 2019

i get this: mojave 10.14.6

l) WhateverGreen: weg @ failed to apply agdp Piker-Alpha's patch 7

Edited September 19, 2019 by maclinuxG4

bobpedro · September 19, 2019

On 9/18/2019 at 2:11 AM, TheBloke said:

PROGRESS!

I posted about my many issues on the Olarilla forum, and @MaLd0n has immediately made things a lot better for me, at least on my X58 system.

The answer: setting my SMBIOS to iMacPro 1.1.   It honestly would never have occurred to me to try an SMBIOS as late as this on my old X58 system. But in hindsight, it makes sense: the iMacPro has no iGPU, and is probably pretty unique in that nowadays.   And the more I think about it, the more I am sure that the lack of iGPU in both my X58 and H77 systems is the key to all these problems.

Anyway, changing my X58 to iMacPro 1.1 has made the following improvements:

I now have HW encode/decode, on both h264 and h265. Playing a 4K video in VLC now uses 15% CPU, instead of 500% !!

If I boot with only one monitor connected to DP1, that monitor will work perfectly - no black screen, no errors in Console, no high CPU usage, sleep/wake works fine. Absolutely perfect.
So for anyone who only has one monitor, this would be an entirely working situation.

Unfortunately I still cannot successfully boot with more than 1 monitor connected. If any other monitor is plugged in, even if the monitor is off, then I get black on all screens, and I cannot put the system to sleep.

As before, once I have done sleep & wake once, I can then use all ports; all monitors get a picture.

Therefore, my bootup process is now:
At Clover screen, unplug all monitor cables except DP1

Boot into macOS

Sleep

Reconnect my other 4 monitors

Wake

Then everything is perfect.

My other issue was the 4K @ 60hz monitor, and that also seems to be mostly resolved. Since I switched SMBIOS I haven't ever had it come up at only 30hz or only 2840x1080.
I have had a different issue, where the monitor is detected at 4K @ 60hz but I get no picture. But I think I've resolved that now by putting the 4K monitor on DP1, so it's also the boot up monitor.   Even if the issue does happen, it seems to be resolvable by unplugging and replugging the monitor a few times.

My new boot up procedure still sounds quite time consuming, and it is. But it is hugely better than the situation I had before, where I could not get a picture on any monitor. Now at least I can easily put the system to sleep with the mouse, rather than needing to SSH in. And applying upgrades and things like that is so much easier when you can actually see a picture on one monitor!

So I have gone from majorly annoyed to now being quite happy with my Vega 64 setup. In particular, getting HW encode/decode is a major step forward. I've literally never had that working in macOS before.

Unfortunately this SMBIOS change did not make any difference on my second system, the H77 with my old 7970Ghz / R9 280X GPU.    On that system I need to sleep & wake in order to get a picture on both monitors (though at least I can see a picture on one from boot). And sadly I still don't have h264 encode/decode available. EDIT: further research has revealed that it's impossible to get HW accel on the R9 280X. The drivers just do not support it. Source is a long MacRumours thread where real Mac Pro users are using Lilu and Whatevergreen to enable HW accel on recent AMD GPUs. Their testing shows that the earliest GPUs with supported HW accel is Polaris / RX 400 series.

I will continuing researching on black screen on boot errors and will post again if I do get any further.

Hey @TheBloke great to hear from your success on then new vega using another smbios. My hack runs on a mobo without an on-board gpu, so the strange problem might be related to the missing iGPU. In my case, as I only need to run 2 monitors simultaneously, I am quite happy with the adapter work-around using DP and HDMI ports. But I will try changing the smbios too, maybe I can get rid of the adapter, I will report my progress.

Mojo28 · September 19, 2019

On 9/9/2019 at 12:58 PM, Slice said:

See messages in your log


2019-09-01 23:51:10.568411+0200  localhost kernel[0]: (kernel) WhateverGreen:     weg @ IGPU device (00:02.00) has device-id 0x0112, you should change it to 0x0116
2019-09-01 23:51:10.575741+0200  localhost kernel[0]: (kernel) WhateverGreen:     weg @ IMEI device (00:16.00) has device-id 0x1E3A, you should change it to 0x1C3A

No ideas?

dragonmel · September 21, 2019

On 9/17/2019 at 8:11 PM, TheBloke said:

PROGRESS!

I posted about my many issues on the Olarilla forum, and @MaLd0n has immediately made things a lot better for me, at least on my X58 system.

The answer: setting my SMBIOS to iMacPro 1.1.   It honestly would never have occurred to me to try an SMBIOS as late as this on my old X58 system. But in hindsight, it makes sense: the iMacPro has no iGPU, and is probably pretty unique in that nowadays.   And the more I think about it, the more I am sure that the lack of iGPU in both my X58 and H77 systems is the key to all these problems.

Anyway, changing my X58 to iMacPro 1.1 has made the following improvements:

I now have HW encode/decode, on both h264 and h265. Playing a 4K video in VLC now uses 15% CPU, instead of 500% !!

If I boot with only one monitor connected to DP1, that monitor will work perfectly - no black screen, no errors in Console, no high CPU usage, sleep/wake works fine. Absolutely perfect.
So for anyone who only has one monitor, this would be an entirely working situation.

Unfortunately I still cannot successfully boot with more than 1 monitor connected. If any other monitor is plugged in, even if the monitor is off, then I get black on all screens, and I cannot put the system to sleep.

As before, once I have done sleep & wake once, I can then use all ports; all monitors get a picture.

Therefore, my bootup process is now:
At Clover screen, unplug all monitor cables except DP1

Boot into macOS

Sleep

Reconnect my other 4 monitors

Wake

Then everything is perfect.

My other issue was the 4K @ 60hz monitor, and that also seems to be mostly resolved. Since I switched SMBIOS I haven't ever had it come up at only 30hz or only 2840x1080.
I have had a different issue, where the monitor is detected at 4K @ 60hz but I get no picture. But I think I've resolved that now by putting the 4K monitor on DP1, so it's also the boot up monitor.   Even if the issue does happen, it seems to be resolvable by unplugging and replugging the monitor a few times.

My new boot up procedure still sounds quite time consuming, and it is. But it is hugely better than the situation I had before, where I could not get a picture on any monitor. Now at least I can easily put the system to sleep with the mouse, rather than needing to SSH in. And applying upgrades and things like that is so much easier when you can actually see a picture on one monitor!

So I have gone from majorly annoyed to now being quite happy with my Vega 64 setup. In particular, getting HW encode/decode is a major step forward. I've literally never had that working in macOS before.

Unfortunately this SMBIOS change did not make any difference on my second system, the H77 with my old 7970Ghz / R9 280X GPU.    On that system I need to sleep & wake in order to get a picture on both monitors (though at least I can see a picture on one from boot). And sadly I still don't have h264 encode/decode available. EDIT: further research has revealed that it's impossible to get HW accel on the R9 280X. The drivers just do not support it. Source is a long MacRumours thread where real Mac Pro users are using Lilu and Whatevergreen to enable HW accel on recent AMD GPUs. Their testing shows that the earliest GPUs with supported HW accel is Polaris / RX 400 series.

I will continuing researching on black screen on boot errors and will post again if I do get any further.

I have exactly the same thing going on.

I tried a fresh install with clover on 10.14.6 g87 I think before the supplementaries

on my DX58so booting UEFI I tried native nvram (which works and holds testvar) and using RC scripts

same issues if I only have my 4k plugged in at boot at DP with a 1.2 active to HDMI adapter, its fine.. but if I try to boot with or attach in any way my HDMI monitor.. no go ... black screen most of the time with the system running as I can ssh in.

I tried my DSDT, and Madl0ns and no sleep, display would go dark but system stayed up. an epic fail all around.. latest WEG installed

went back to another install from last month that was a combination of installing mojave fresh and using mac migration to bring in my El Cap volume that has worked rock solid on this board.. only issues since upgrading to mojave and replacing the gtx660 which did multimonitor, sleep etc rock solid for years but could just not dial in the black screening with mojave

well I went back to clover 4932, UEFI boot, native nvram, my old DSDT and no dsdt patching turned on in clover, and now I have the same as bloke for the most part

I can boot into 4k@60 on the DP to HDMI regularly, and sleep, wake seems to work. If I try to boot with HDMI or plug it in during or after boot, black screen with the system up behind it

if I boot 4k monitor and have the other monitor off (plugged in but not standby), get to the login screen, sleep, turn on both monitors and wake, I have both monitors and so far can sleep and wake them

I am using a macpro5,1 smbios since the board chipset and w3680 are almost identical to Mac Pro . I get full power management on the GPU and CPU, stepping and GPU fan works as advertised

I use no darkwake in my boot args (in the past had darkwake=no and darkwake=0 seemed to work) shikigva-96, shikiboardid for imacpro, -rad24 (both monitors would come up as 30bit and they are not )

I have been trying mostly with ATI deinit deselected in clover as I was told that was now in WEG

using the boot args, I can get hardware accelerated h264 encode/decode and while not showing up in the videoproc h265 decode for sure is working since sony HDR wordsmith plays back fine with almost no CPU and the GPU ramps up... witthout boot args, no hardware accel on decode or encode

I would love to use iMac pro since it seems to play better with the GPU but obviously I get no CPU power management when I use it my cores stay locked at a 12 multiplier and don't step, using Mac Pro it steps to boost of 25-27 depending on thread involvement

in SUM

this card / apples drivers are not syncing the monitors properly or possibly at all.. it seems hit or miss

the iMac pro is advertised to drive its internal plus up to 4 external 4k but over thunderbolt 3

I think that most of the tuning in WEG has been concentrated on the newer chipsets, could someone take a look at this on more legacy hardware?

when my system is booting into black screen and up behind it.. my console crash and spin dump section is filling up with GPU.RESTART entrys... I can post them up here later, sent them to @slice in another thread last night

incidentally, with my DSDT edits, I can get GFX0 to bring up the VEGA without WEG, with HDMI audio, but am using AppleALC with LILU as well for system sound

TheBloke · September 21, 2019

@dragonmel I've just realised you must be the duece guy I've been speaking to on mald0n's forum?

When you mentioned your mobo and CPU I thought it was a coincidence, especially when you say here you'd love to use iMacPro SMBIOS, and over there you said you wanted MP 5.1. But you've just written about the speed x12 multiplier on both forums, so, must be you.

Anyway, yes I have the same problem with iMacPro 1.1 and x12 multiplier. But it is resolved by sleep & wake. Given I have to sleep & wake anyway to use more than one port, this is not the end of the world.

So right now I am still in iMacPro 1.1. From boot I get a picture on DP1, then I put the system to sleep, plug in the other monitors, then wake. At that point I will have all monitors detected, and proper power management, with x12 - x24 multipliers and normal benchmark scores.

However, my 4K @ 60 monitor often will show no picture. It's detected at 4K@60, and shows in Displays, but it's just black. I have to unplug/replug it a couple of times, and/or power it on and off, before eventually it shows a picture. This is a big improvement on the MacPro 5.1 SMBIOS, where it normally only connects at 2084x1080, or sometimes at 4K@30, but almost never at 4K@60.

I noted that in Windows 10 I also had a black screen issue with the 4K monitor, so I figure this issue is not specific to macOS. Though it is definitely worse in macOS - on Windows 10, I've had the black screen on the 4K monitor a couple of times, and it's always gone away after I've replugged the monitor once. In macOS, I get it every single boot, and it usually takes 2, 3 or even 4 attempts before I get a picture on it.

I have working h264 and h265 HW accel with iMacPro 1.1, except I have been getting regular crashes when I try to use it. In particular, trying to export a Premiere Pro project with HW h264 encoding will seem to run fine for 4-5 minutes, then suddenly all displays stop updating, and if I SSH in and check the logs, I will see many instances of:

kernel[0:472] (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt

As well as errors about GPU hang.

However, I also had a crash a couple of times in Windows 10 when trying some HW h264 encoding, so I am not 100% sure if this is specific to macOS. But, again, if it's not specific, it's certainly worse.

I have been playing around in the BIOS with my OC settings to see if the issue will go away with different settings. But I don't hold out a huge amount of hope.

I tried the shikivga line with MP 5.1 SMBIOS, and this gave me h264 encode/decode, but not h265. And as mentioned, I could never get my 4K monitor to reliably connect at 4K@60, so this made MP 5.1 a bit of a non-starter for me. So I never bothered testing if HW h264 encode would crash in MP5.1 like it currently is with iMacPro 1.1.

So right now it's all a bit of a mess, but I do have a somewhat usable system. Yesterday I actually got a whole day of useful work out of it! Although there are still weird issues; like yesterday I did a lot of After Effects work, and video previews felt really slow, slower than I remember from using my R9 280X. Then at the end of the night I ran a Geekbench 4 test, and only got 150k on the Metal Compute test, where it should be 200k.

I'd be very interested if Slice had anything to say about the errors.

Edited September 21, 2019 by TheBloke

dragonmel · September 21, 2019

1 hour ago, TheBloke said:

@dragonmel I've just realised you must be the duece guy I've been speaking to on mald0n's forum?

When you mentioned your mobo and CPU I thought it was a coincidence, especially when you say here you'd love to use iMacPro SMBIOS, and over there you said you wanted MP 5.1. But you've just written about the speed x12 multiplier on both forums, so, must be you.

Anyway, yes I have the same problem with iMacPro 1.1 and x12 multiplier. But it is resolved by sleep & wake.   Given I have to sleep & wake anyway to use more than one port, this is not the end of the world.

So right now I am still in iMacPro 1.1. From boot I get a picture on DP1, then I put the system to sleep, plug in the other monitors, then wake. At that point I will have all monitors detected, and proper power management, with x12 - x24 multipliers and normal benchmark scores.

However, my 4K @ 60 monitor often will show no picture. It's detected at 4K@60, and shows in Displays, but it's just black.   I have to unplug/replug it a couple of times, and/or power it on and off, before eventually it shows a picture.   This is a big improvement on the MacPro 5.1 SMBIOS, where it normally only connects at 2084x1080, or sometimes at 4K@30, but almost never at 4K@60.

I noted that in Windows 10 I also had a black screen issue with the 4K monitor, so I figure this issue is not specific to macOS. Though it is definitely worse in macOS - on Windows 10, I've had the black screen on the 4K monitor a couple of times, and it's always gone away after I've replugged the monitor once.   In macOS, I get it every single boot, and it usually takes 2, 3 or even 4 attempts before I get a picture on it.

I have working h264 and h265 HW accel with iMacPro 1.1, except I have been getting regular crashes when I try to use it.   In particular, trying to export a Premiere Pro project with HW h264 encoding will seem to run fine for 4-5 minutes, then suddenly all displays stop updating, and if I SSH in and check the logs, I will see many instances of:

kernel[0:472] (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt

As well as errors about GPU hang.

However, I also had a crash a couple of times in Windows 10 when trying some HW h264 encoding, so I am not 100% sure if this is specific to macOS. But, again, if it's not specific, it's certainly worse.

I have been playing around in the BIOS with my OC settings to see if the issue will go away with different settings. But I don't hold out a huge amount of hope.

I tried the shikivga line with MP 5.1 SMBIOS, and this gave me h264 encode/decode, but not h265. And as mentioned, I could never get my 4K monitor to reliably connect at 4K@60, so this made MP 5.1 a bit of a non-starter for me. So I never bothered testing if HW h264 encode would crash in MP5.1 like it currently is with iMacPro 1.1.

So right now it's all a bit of a mess, but I do have a somewhat usable system. Yesterday I actually got a whole day of useful work out of it! Although there are still weird issues; like yesterday I did a lot of After Effects work, and video previews felt really slow, slower than I remember from using my R9 280X. Then at the end of the night I ran a Geekbench 4 test, and only got 150k on the Metal Compute test, where it should be 200k.

I'd be very interested if Slice had anything to say about the errors.

yeah.. its me

the iMac love/hate... would love h265 hardware encode.... can't get that with Mac Pro 5,1 yet.. and likely never will.. apples marketing department I am sure told engineering to disable it to make the iMac pro and forthcoming new cheesgrater $36,000 new mac pro a must by for video production folks.. ie run people out of perfectly good hardware to make a buck

I feel that the iMac hangs and freezes are likely a combination of wrong chipset/cpu power management since using that SMBIOS would load the wrong kexts for your hardware.. AND the video encoder frameworks for the iMac Pro don't just use the VEGA but use the t2 chip as well...

and unlike the WEG solutions for running 'headless' intel by shutting off internal GPU, there is no way (and no reason) in a real iMac pro to disable the T2 so.. .that coupled with my other post on the fact that our boards have not gotten the most recent microcode patches for the last 6 years.. all ends up being messy

left the system sit 12 hours last night and woke fine

inject ATI was a mess with my SMBIOS.. detailed on the other site..

TheBloke · September 22, 2019

Some updates:

I have got my Vega 64 almost fully working in my second Hack, the H77. From boot, I get a picture on all ports, both DP and HDMI, with no need to sleep.

The solution? So simple: Set CSM to Disabled in BIOS. So @bobpedro if your UEFI motherboard has the option to set CSM to Disable, try this; you may find it solves your problems.

This also confirms my earlier belief that these problems are UEFI related, and tells me that there is very likely no solution for legacy motherboards, or UEFI boards without CSM control.

Unfortunately, my Vega 64 is still not fully functional in the H77. Using HW h264 encode will lead to GPU hangs/display freezes. Exact timing is variable, but in certain tests it is 100% repeatable.

I was able to export a 10 minute Premiere Pro project, which took about 25 minutes to export (it has lots of After Effects sections) and exported successfully; something I was never able to do before using HW accelerated encode. However I then tried using a screen recording app (Screenflick) to record 1920x1080 at 60 FPS, and the GPU would always hang after 2 - 5 minutes of recording. I've tested in Windows 10 on the same system, and there are no hangs there, so I do not believe there is any hardware issue with the card.

I've tested HW h264 encode both using the iMacPro 1.1 SMBIOS, and also using my standard iMac 14.2 SMBIOS with "shikigva=96 shiki-id=Mac-7BA5B2D9E42DDD94" boot arguments, to simulate the iMacPro. The former gives me h264 and h265 encode and decode. The latter gives me h264 encode/decode, but no h265 encode (might have h265 decode, not checked that.) But both methods result in GPU hangs/freezes after some time using the h264 encode, which is a big shame.

When the GPU hangs during h264 encode, I get the following series of errors, and all displays freeze. I can still SSH in and run commands, but shutdown -r now does not work; the only solution is a hardware reset or shutdown.

kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
kernel[0]: (IOAcceleratorFamily2) AMDRadeonAccelerator: IOAccelDisplayPipeTransaction time out after 300ms. framebufferIndex=5
kernel[0]: (IOAcceleratorFamily2)     framebufferIndex=5, wsaa=17
kernel[0]: (IOAcceleratorFamily2)     eventInterruptEnabled=0, transactionInterruptEnabled=1, vblInterruptEnabled=0
kernel[0]: (IOAcceleratorFamily2)     powerOff=0, pipeTerminated=0, acceleratorEnabled=1, fWSAA=17
kernel[0]: (IOAcceleratorFamily2)     lastIOGraphicsMessageEvent=93, fbIndex=5
kernel[0]: (IOAcceleratorFamily2)     transactionQueueReadCount=12055, transactionQueueWriteCount=12056
kernel[0]: (IOAcceleratorFamily2) lastCompletedTransaction: ID=12054 dirtyBits=0x1 options=0x1
kernel[0]: (IOAcceleratorFamily2)     pendingTransaction ID=12055, isTransactionComplete()=0, dirtyBits=0x1, options=0x1, errorCode=0x0, submittedReturn=0xe0014042
kernel[0]: (IOAcceleratorFamily2)     readTransaction ID=12056, dirtyBits=0x1, options=0x1, errorCode=0x0, submittedReturn=0xe0014042, event hasn't finished
kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
<last message repeated many times>
kernel[0]: (IOAcceleratorFamily2) virtual IOReturn IOAccelEventMachine2::waitForStamp(int32_t, stamp_t, stamp_t *): initial wait for 1 second expired. Continue wait for 4 seconds. stamp 15313 (gpu_stamp=15312)
kernel[0]: (IOAcceleratorFamily2) virtual IOReturn IOAccelEventMachine2::waitForStamp(int32_t, stamp_t, stamp_t *): initial wait for 1 second expired. Continue wait for 4 seconds. stamp 9557 (gpu_stamp=9556)
kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
<last message repeated many times>
kernel[0]: (IOAcceleratorFamily2) virtual IOReturn IOAccelEventMachine2::waitForStamp(int32_t, stamp_t, stamp_t *): timeout waiting for AMDRadeonAccelerator stamp 15313 (gpu_stamp=15312)
kernel[0]: (IOAcceleratorFamily2) virtual IOReturn IOAccelEventMachine2::waitForStamp(int32_t, stamp_t, stamp_t *): timeout waiting for AMDRadeonAccelerator stamp 9557 (gpu_stamp=9556)
kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt

<last message repeated many times>

kernel[0]: (IOAcceleratorFamily2) virtual void IOAccelEventMachineFast2::checkGPUProgress() - Signaling hardware error on channel 0..
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartEnqueued stampIdx=0 type=2
kernel[0]: (IOAcceleratorFamily2) virtual void IOAccelEventMachineFast2::checkGPUProgress() - Signaling hardware error on channel 2..
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartSignaled stampIdx=2 type=2 prevType=0 numStamps=22
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartEnqueued stampIdx=2 type=2
kernel[0]: (IOAcceleratorFamily2) virtual void IOAccelEventMachineFast2::checkGPUProgress() - Signaling hardware error on channel 5..
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartSignaled stampIdx=5 type=2 prevType=0 numStamps=22
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartEnqueued stampIdx=5 type=2
kernel[0]: (IOAcceleratorFamily2) virtual void IOAccelEventMachineFast2::checkGPUProgress() - Signaling hardware error on channel 6..
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartSignaled stampIdx=6 type=2 prevType=0 numStamps=22
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartEnqueued stampIdx=6 type=2
kernel[0]: (IOAcceleratorFamily2) virtual void IOAccelEventMachineFast2::checkGPUProgress() - Signaling hardware error on channel 12..
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartSignaled stampIdx=12 type=2 prevType=0 numStamps=22
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartEnqueued stampIdx=12 type=2
kernel[0]: (IOAcceleratorFamily2) virtual void IOAccelEventMachineFast2::checkGPUProgress() - Signaling hardware error on channel 17..
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartSignaled stampIdx=17 type=2 prevType=0 numStamps=22
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartEnqueued stampIdx=17 type=2
kernel[0]: (IOAcceleratorFamily2) virtual void IOAccelEventMachineFast2::checkGPUProgress() - Signaling hardware error on channel 18..
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartSignaled stampIdx=18 type=2 prevType=0 numStamps=22
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartEnqueued stampIdx=18 type=2
kernel[0]: (IOAcceleratorFamily2) virtual void IOAccelEventMachineFast2::checkGPUProgress() - Signaling hardware error on channel 19..
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartSignaled stampIdx=19 type=2 prevType=0 numStamps=22
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::signalHardwareError(eRestartRequest, int32_t): GPURestartEnqueued stampIdx=19 type=2
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::hardwareErrorEvent(): setting restart type to 2 (channel 0)
kernel[0]: (IOAcceleratorFamily2) void IOAccelEventMachine2::hardwareErrorEvent(): GPURestartDequeued stampIdx=0 type=2
kernel[0]: (AMDRadeonX5000) [3:0:0]: channel 0 event timeout
kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
kernel[0]: (IOAcceleratorFamily2) virtual IOReturn IOAccelEventMachine2::waitForStamp(int32_t, stamp_t, stamp_t *): a graphics error occurred, exitting..
kernel[0]: (IOAcceleratorFamily2) virtual IOReturn IOAccelEventMachine2::waitForStamp(int32_t, stamp_t, stamp_t *): a graphics error occurred, exitting..
kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
kernel[0]: (AMDRadeonX5000) [3:0:0]: channel 0 GFX is hung! (lastReadTimestamp=0x0001212a) channelResetMask 0x00000000
kernel[0]: (AMDRadeonX5000HWLibs) AMD Cail: [3:0:0] GPU HangState 0x00000040, HangFlags 0x00000005: IndividualEngineHang 1, NonEngineBlockHang 0, FenceNotRetired 1, PerEngineReset 1, FullAsicReset 0
kernel[0]: (AMDRadeonX5000HWLibs) [3:0:0] GPU HangState 0x00000040, HangFlags 0x00000005: IndividualEngineHang 1, NonEngineBlockHang 0, FenceNotRetired 1, PerEngineReset 1, FullAsicReset 0
kernel[0]: (IOAcceleratorFamily2) virtual sIOAccelEvent *IOAccelFIFOChannel2::getFirstPendingEvent(): All are finished
kernel[0]: (IOAcceleratorFamily2) virtual sIOAccelEvent *IOAccelFIFOChannel2::getFirstPendingEvent(): All are finished
kernel[0]: (IOAcceleratorFamily2) virtual sIOAccelEvent *IOAccelFIFOChannel2::getFirstPendingEvent(): All are finished
kernel[0]: (IOAcceleratorFamily2) virtual sIOAccelEvent *IOAccelFIFOChannel2::getFirstPendingEvent(): All are finished
kernel[0]: (IOAcceleratorFamily2) virtual sIOAccelEvent *IOAccelFIFOChannel2::getFirstPendingEvent(): All are finished
kernel[0]: (IOAcceleratorFamily2) virtual sIOAccelEvent *IOAccelFIFOChannel2::getFirstPendingEvent(): All are finished
kernel[0]: (IOAcceleratorFamily2) virtual sIOAccelEvent *IOAccelFIFOChannel2::getFirstPendingEvent(): All are finished
kernel[0]: (IOAcceleratorFamily2) virtual sIOAccelEvent *IOAccelFIFOChannel2::getFirstPendingEvent(): All are finished
kernel[0]: (IOAcceleratorFamily2) virtual sIOAccelEvent *IOAccelFIFOChannel2::getFirstPendingEvent(): All are finished
kernel[0]: (IOAcceleratorFamily2) void IOAccelFenceMachine::fence_timeout(IOTimerEventSource *): AMDRadeonAccelerator prodding blockFenceInterrupt
kernel[0]: (AMDRadeonX5000) [3:0:0] ** AMDRadeonX5000_AMDVega10GraphicsAccelerator Device in slot: SLOT--1 **
kernel[0]: (IOAcceleratorFamily2) Trying to restart GPU (Radeon RX Vega 64)...
kernel[0]: (AMDSupport) AMD Recovery Display.
kernel[0]: (Sandbox) Sandbox: tailspin(1751) deny(1) sysctl-write kern.procname

I note that this first error, "AMDRadeonAccelerator: IOAccelDisplayPipeTransaction" is the same error I first posted about here, on my X58 Hack with the R9 280X getting black screens on boot. I assume this is because it's a generic message when the GPU experiences problems, rather than they have related causes (as the issue on the X58 is surely related to no UEFI). But I don't know for sure of course.

So all in all, very frustrating. I keep making small amounts of progress, then spending a huge amount of time on testing, only to find that in no scenario can I get a fully working system. One of the reasons I bought the new GPU was to help with h264 encode, but now I am booting my H77 into Windows 10 to do it there instead.

Oh well, at least I have learnt a little bit more in the process..

Edited September 22, 2019 by TheBloke

exquirentibus · September 25, 2019

On 9/2/2019 at 3:06 PM, bobpedro said:

@HighSierraWannabe I ordered this adapter from Club3D as it is one of the only adapters available which is capable of a full 2560x1600 resolution. It arrived today, and after some configuration I was able to get the full resolution to work. The only issue I noticed is some blue flickering on black screens (noticeable on boot and youtube)... I think I will give the Apple (MB571Z/A) Adapter another try (I can lend one from work) as it might be superior to the Club3D adapter... but then an additional miniDP -> DP adapter is necessary - not sure if its a good idea...

I am on Mojave 10.14.5 now, but I want to perform the update in the next few weeks.

I think the problem exists for all RX cards, as @jsl2000 has it with the RX560, I have it with RX570 and @TheBloke with a RX580.

EDIT1: After uninstalling SwitchResX the blue flickering is gone, might have been related to this. Not sure if its worth trying the Apple (MB571Z/A) Adapter then...

EDIT2: I noticed some small red dot flickering in terminal window, verified by making it fullscreen. Seems like the adapter has some issues with black or the disorder is only noticeable there...

EDIT3: Today I tried the Apple (MB571Z/A) adapter at from with a miniDP -> DP adapter and it works perfectly. No more flickering and a lot more supported resolutions to choose from. The Club3D Adapter goes back, MB571Z/A ordered on eBay.

I tried a Saphire Radeon RX 570 with a single DVI (plus hdmi, display port) and also an Asus Radeon 570 with two DVI ports.

All of the DVI ports worked for me.

I'm now running 10.14.6 with (3) 2660 X 1440 Dell monitors using two DVI ports and one Display Port -> DVI (with the Club3D adapter you mentioned). The DVI ports gave me zero trouble! Thanks for the Cllub3d recommendation!!

Edited September 25, 2019 by exquirentibus

WhatEverGreen Support Topic

Recommended Posts

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites

Link to comment

Share on other sites