Asus X299 - Support

[djlild7hina]

Welp Windows is crashing now when I play games. I’m returning the Encore and going with the Sage II. Hope it works out better...

 

[djbuddha]

I have a very odd issue suddenly (I don't think it's related to RadeonBoost, but it could be- this is hard to diagnose).

So as I wrote earlier in the thread, I am using this machine for day-to-day operations. And for editing/exporting in Final Cut X and encoding MP4's w/ Handbrake or Sorenson Squeeze it's flawless.

I run iStat Menus to keep an eye on things, and the GPU fires up when it needs to in FCPX and the cores light up exactly as they need to in all the programs mentioned.

My achilles-heel is OBS.

I've had this specific issue twice now, and can't solve it. On NYE, I was trying to stream a broadcast to my cousins in the US (I'm in Canada) because the radio networks I DJ for have a region lock on their platforms. I'm allowed to do this privately- so I attempted it w/ OBS. I had this issue last week but thought it was a random anomaly, but then it did it again. I'm not actually DJ'ing at this time- it's running a pre-recorded stream.

Here's the issue. I run OBS and 80% of the time, the stream is flawless (not forgoing any internet connection issues, but that's something my ISP needs to deal with). In this particular case, I just want to leave it running- and most of the time it works, but the computer will 'seemingly' crash. Now I say this because I came back to the machine to check on it, all the screens are off, the fans are SCREAMING, but the machine is dead-cold to the touch. I hold power for 5 seconds and reboot. In OBS, I have hardware encoding enabled for my GPU (Vega 64). There's another option called 'AMD Software Encoding' which seems to use the GPU because I get around the same GPU / CPU usage using that vs just regular software encoding (which obviously just only uses my CPU)- for my scenes, I'd like to use GPU strictly. This 'AMD Software Encoding' option seems to trigger the issue more often than 'AMD Hardware Encoding' but on NYE, it happened with the 'AMD Hardware Encoding' option.

Everything comes back up, but there's no kernel panic log- so I don't think it's a KP. Where can I identify a log of what exactly happened? I MAY be able to re-produce it over the next few days. But when I do my weekly stream on Monday evenings, I don't want to be walking on egg-shells LOL.

Anybody know where I can identify this? I have to work with this machine today for my day-job, so I'm sure it'll be fine, but I may test later today if anybody can give me some insight as to where to look for logs, or what may be happening.

 

[izo1]

I have a very odd issue suddenly (I don't think it's related to RadeonBoost, but it could be- this is hard to diagnose).

So as I wrote earlier in the thread, I am using this machine for day-to-day operations. And for editing/exporting in Final Cut X and encoding MP4's w/ Handbrake or Sorenson Squeeze it's flawless.

I run iStat Menus to keep an eye on things, and the GPU fires up when it needs to in FCPX and the cores light up exactly as they need to in all the programs mentioned.

My achilles-heel is OBS.

I've had this specific issue twice now, and can't solve it. On NYE, I was trying to stream a broadcast to my cousins in the US (I'm in Canada) because the radio networks I DJ for have a region lock on their platforms. I'm allowed to do this privately- so I attempted it w/ OBS. I had this issue last week but thought it was a random anomaly, but then it did it again. I'm not actually DJ'ing at this time- it's running a pre-recorded stream.

Here's the issue. I run OBS and 80% of the time, the stream is flawless (not forgoing any internet connection issues, but that's something my ISP needs to deal with). In this particular case, I just want to leave it running- and most of the time it works, but the computer will 'seemingly' crash. Now I say this because I came back to the machine to check on it, all the screens are off, the fans are SCREAMING, but the machine is dead-cold to the touch. I hold power for 5 seconds and reboot. In OBS, I have hardware encoding enabled for my GPU (Vega 64). There's another option called 'AMD Software Encoding' which seems to use the GPU because I get around the same GPU / CPU usage using that vs just regular software encoding (which obviously just only uses my CPU)- for my scenes, I'd like to use GPU strictly. This 'AMD Software Encoding' option seems to trigger the issue more often than 'AMD Hardware Encoding' but on NYE, it happened with the 'AMD Hardware Encoding' option.

Everything comes back up, but there's no kernel panic log- so I don't think it's a KP. Where can I identify a log of what exactly happened? I MAY be able to re-produce it over the next few days. But when I do my weekly stream on Monday evenings, I don't want to be walking on egg-shells LOL.

Anybody know where I can identify this? I have to work with this machine today for my day-job, so I'm sure it'll be fine, but I may test later today if anybody can give me some insight as to where to look for logs, or what may be happening.

If you are on Big Sur you don't need RadeonBoost anymore. It seems that Apple has fixed that issue in their AMD drivers.

For CPU fans, set custom curves and cap them so they never go crazy (this can be done in the BIOS). For GPU fans you can set custom power tables.

But I think this is software related, and it is an odd issue. I haven't used OBS in ages, but is it possible to hit up the developers and talk to them about it? They might patch it for you if you give them a breakdown of what happens.

 

[djbuddha]

If you are on Big Sur you don't need RadeonBoost anymore. It seems that Apple has fixed that issue in their AMD drivers.

For CPU fans, set custom curves and cap them so they never go crazy (this can be done in the BIOS). For GPU fans you can set custom power tables.

But I think this is software related, and it is an odd issue. I haven't used OBS in ages, but is it possible to hit up the developers and talk to them about it? They might patch it for you if you give them a breakdown of what happens.

Stuck on Mojave because of some 32-bit apps I need. Is there anywhere I can find a system log of what happened that's not necessarily a KP log? Like if I run Console, will it keep the info after the apparent crash?

EDIT: Ok I went fishing in Console and found: Kernel_2020-12-28-150645_DJ-Buddhas-iMac-Pro.gpuRestart which co-incides with is the error that happened a few days ago. The one from NYE is: Kernel_2020-12-31-220840_DJ-Buddhas-iMac-Pro.gpuRestart

So clearly it's a GPU issue. I'm going to try to replicate this as I did before, and then try to remove RadeonBoost and see if it still happens.

It's a shame if it is RadeonBoost causing this, I'm getting an easy 30% consistent uplift in performance from my GPU-bound applications. Hey @CaseySJ ... I spoke too soon LOL

Event:               GPU Reset
Date/Time:           Mon Dec 28 15:06:45 2020
Application:      
Path:            
Tailspin:            /Library/Logs/DiagnosticReports/gpuRestart2020-12-28-150645.tailspin
GPUSubmission Trace ID: 0
OS Version:          Mac OS X Version 10.14.6 (Build 18G95)
Graphics Hardware:   Radeon RX Vega 64
Signature:           1

Report Data:

GPU Log Version: 1

Restart Channel: 6 SDMA1 GFX

---THE STATE OF THE DRIVER---

AMDRadeonX5000_AMDVega10GraphicsAccelerator PCIe Device: [167:0:0] State: ENABLED
TotalVideoRAMBytes: 0x00000001ff000000 (8573157376)

[00] Channel: GFX
  CompletedTS = 0x0126f972, SubmittedTS = 0x0126f9c1
  PendingEvent: YES
    NumberOfPendingCB = 79, FirstPendingTS = 0x0126f973, LastPendingTS = 0x0126f9c1
    FirstPendingCB:
      ProcessID = 10703, ProcessName = obs, SubmitContext = GLCtx (0xffffff811eb3a800)
      NumberOfWaits = 2
        1 - Channel 19 GFX_HP, TS = 0x0029df0f (passed)
        2 - Channel 18 VMPT, TS = 0x007c2daf (passed)
      GPUAddress = 0x0000000402060000, Size = 0x00000aa0, VMID = 6
      ContentValidation = SKIP

[01] Channel: C0
  CompletedTS = 0x00000000, SubmittedTS = 0x00000000
  PendingEvent: NO
    NumberOfPendingCB = 0

[02] Channel: C1
  CompletedTS = 0x0005ab8c, SubmittedTS = 0x0005ab8e
  PendingEvent: YES
    NumberOfPendingCB = 2, FirstPendingTS = 0x0005ab8d, LastPendingTS = 0x0005ab8e
    FirstPendingCB:
      ProcessID = 10646, ProcessName = obs, SubmitContext = MetalCtx (0xffffff810e42d000)
      NumberOfWaits = 1
        1 - Channel 5 SDMA0 GFX, TS = 0x00ce9e04
      GPUAddress = 0x0000000400040000, Size = 0x00000040, VMID = 13
      ContentValidation = PASS
      Execution point 0x40
      Buffer range 0x0 .. 0x3f:

Then there's this at the bottom before all the "0x0000ffff" GPU info stuff:

VM Protection Fault (GFX): YES
    Page GPUAddress = 0x0000fffffffff000, VMID = 15
    Failing Protection = VALID, READ, WRITE, EXECUTE, DUMMY PAGE
    Memory Client ID = 511 (UNKNOWN)
    Memory Client R/W = WRITE
    Page table: 0x0000ffffffffe000 .. 0x0000fffffffff000
Requested range (start 0xffffffffe000, size 0x1001) is invalid.

VM Protection Fault (MM): YES
    Page GPUAddress = 0x0000fffffffff000, VMID = 15
    Failing Protection = VALID, READ, WRITE, EXECUTE, DUMMY PAGE
    Memory Client ID = 511 (UNKNOWN)
    Memory Client R/W = WRITE
    Page table: 0x0000ffffffffe000 .. 0x0000fffffffff000
Requested range (start 0xffffffffe000, size 0x1001) is invalid.

 

[izo1]

Stuck on Mojave because of some 32-bit apps I need. Is there anywhere I can find a system log of what happened that's not necessarily a KP log? Like if I run Console, will it keep the info after the apparent crash?

EDIT: Ok I went fishing in Console and found: Kernel_2020-12-28-150645_DJ-Buddhas-iMac-Pro.gpuRestart which co-incides with is the error that happened a few days ago. The one from NYE is: Kernel_2020-12-31-220840_DJ-Buddhas-iMac-Pro.gpuRestart

So clearly it's a GPU issue. I'm going to try to replicate this as I did before, and then try to remove RadeonBoost and see if it still happens.

It's a shame if it is RadeonBoost causing this, I'm getting an easy 30% consistent uplift in performance from my GPU-bound applications. Hey @CaseySJ ... I spoke too soon LOL

Event:               GPU Reset
Date/Time:           Mon Dec 28 15:06:45 2020
Application:    
Path:          
Tailspin:            /Library/Logs/DiagnosticReports/gpuRestart2020-12-28-150645.tailspin
GPUSubmission Trace ID: 0
OS Version:          Mac OS X Version 10.14.6 (Build 18G95)
Graphics Hardware:   Radeon RX Vega 64
Signature:           1

Report Data:

GPU Log Version: 1

Restart Channel: 6 SDMA1 GFX

---THE STATE OF THE DRIVER---

AMDRadeonX5000_AMDVega10GraphicsAccelerator PCIe Device: [167:0:0] State: ENABLED
TotalVideoRAMBytes: 0x00000001ff000000 (8573157376)

[00] Channel: GFX
  CompletedTS = 0x0126f972, SubmittedTS = 0x0126f9c1
  PendingEvent: YES
    NumberOfPendingCB = 79, FirstPendingTS = 0x0126f973, LastPendingTS = 0x0126f9c1
    FirstPendingCB:
      ProcessID = 10703, ProcessName = obs, SubmitContext = GLCtx (0xffffff811eb3a800)
      NumberOfWaits = 2
        1 - Channel 19 GFX_HP, TS = 0x0029df0f (passed)
        2 - Channel 18 VMPT, TS = 0x007c2daf (passed)
      GPUAddress = 0x0000000402060000, Size = 0x00000aa0, VMID = 6
      ContentValidation = SKIP

[01] Channel: C0
  CompletedTS = 0x00000000, SubmittedTS = 0x00000000
  PendingEvent: NO
    NumberOfPendingCB = 0

[02] Channel: C1
  CompletedTS = 0x0005ab8c, SubmittedTS = 0x0005ab8e
  PendingEvent: YES
    NumberOfPendingCB = 2, FirstPendingTS = 0x0005ab8d, LastPendingTS = 0x0005ab8e
    FirstPendingCB:
      ProcessID = 10646, ProcessName = obs, SubmitContext = MetalCtx (0xffffff810e42d000)
      NumberOfWaits = 1
        1 - Channel 5 SDMA0 GFX, TS = 0x00ce9e04
      GPUAddress = 0x0000000400040000, Size = 0x00000040, VMID = 13
      ContentValidation = PASS
      Execution point 0x40
      Buffer range 0x0 .. 0x3f:

Then there's this at the bottom before all the "0x0000ffff" GPU info stuff:

VM Protection Fault (GFX): YES
    Page GPUAddress = 0x0000fffffffff000, VMID = 15
    Failing Protection = VALID, READ, WRITE, EXECUTE, DUMMY PAGE
    Memory Client ID = 511 (UNKNOWN)
    Memory Client R/W = WRITE
    Page table: 0x0000ffffffffe000 .. 0x0000fffffffff000
Requested range (start 0xffffffffe000, size 0x1001) is invalid.

VM Protection Fault (MM): YES
    Page GPUAddress = 0x0000fffffffff000, VMID = 15
    Failing Protection = VALID, READ, WRITE, EXECUTE, DUMMY PAGE
    Memory Client ID = 511 (UNKNOWN)
    Memory Client R/W = WRITE
    Page table: 0x0000ffffffffe000 .. 0x0000fffffffff000
Requested range (start 0xffffffffe000, size 0x1001) is invalid.

Looks GPU driver related. If you can move to at least Catalina that would be nice. Or Big Sur. Apple kept improving AMD drivers overtime.

Can you install Big Sur on another cheap SSD to test stuff out to see if you can find alternatives to 32bit apps you need to use? and to see if OBS works better? If I remember correctly Mojave was the first OS to support Vega drivers and usually first iteration drivers from Apple aren't that great.

 

[djbuddha]

Looks GPU driver related. If you can move to at least Catalina that would be nice. Or Big Sur. Apple kept improving AMD drivers overtime.

Can you install Big Sur on another cheap SSD to test stuff out to see if you can find alternatives to 32bit apps you need to use? and to see if OBS works better? If I remember correctly Mojave was the first OS to support Vega drivers and usually first iteration drivers from Apple aren't that great.

Unfortunately I can't move over. I have Catalina on my 2019 Macbook Pro and there's no alternatives as I've been searching for months to have alternatives on that machine.

I'll attempt in a few hours by first re-creating the issue and finding a way to constantly replicate it. Then I'll remove RadeonBoost and try again. I'm running 1.0 of RadeonBoost and in the version right after mine, they immediately removed Vega 64 support because of broken fan control in the Vega10 drivers from 10.14.5. That is probably why the fans are going crazy when the machine crashes (in theory).

Worst case, I just reboot w/ RadeonBoost disabled when I do OBS streams. IF that's the issue.

 

[izo1]

Unfortunately I can't move over. I have Catalina on my 2019 Macbook Pro and there's no alternatives as I've been searching for months to have alternatives on that machine.

I'll attempt in a few hours by first re-creating the issue and finding a way to constantly replicate it. Then I'll remove RadeonBoost and try again. I'm running 1.0 of RadeonBoost and in the version right after mine, they immediately removed Vega 64 support because of broken fan control in the Vega10 drivers from 10.14.5. That is probably why the fans are going crazy when the machine crashes (in theory).

Worst case, I just reboot w/ RadeonBoost disabled when I do OBS streams. IF that's the issue.

Maybe you can convert your Vega64 to liquid cooled. Dunno what other solution there is to be honest.

 

[djbuddha]

Maybe you can convert your Vega64 to liquid cooled. Dunno what other solution there is to be honest.

That is a very likely option down the road. I'll play around with it soon and see what happens. If it still happens when I remove RadeonBoost, I'll be super confused. But I'll put in a support ticket w/ the OBS guys and see if they can fix whatever is causing it.

Ideally, I'd love to just run RadeonBoost. When I run long workloads with it, and my GPU is running at max for like an hour it works perfectly fine and doesn't climb above 51 degrees. It's just OBS that causes issues in my case.

 

[Loloflatsix]

I made great progress yesterday and today: phew...
After three weeks of issues and failures : it was crazy for me after spending so much money and time going up the wall ...

But yesterday after the message from @djlild7hina about issue on Windows too, I planed to resolve the problems or return the Mobo too.

After doing more tests, no problem on Windows but High Sierra was unstable.

There was a fact that my bios clock was always frozen : whatever the real cause: It was sure to be a First explanation to my problem : yeah : when i removed the battery, weeks ago, to clear CMOS and after putting it back for some reason, it was dead ( amazing on a brand new mobo ...).
So a new CR2032 and lets go...

Second explanation : after clear CMOS a first time I used the Bios Flash Back Button : classic procedure but I remember I used the Bios renamer [EG R6EE.CAP] instead doing that manually...
This time I reflashed the two bioses but i renamed the bios files manually.

After adjusting the bios settings, tried with macOS Catalina.

And the light has come : I renewed my EFI and my USBmap for iMacPro 1,1 : and at the first attempt , it was done.

Pcie allocation BIOS config: 16x/8x/4x

Slot 1 - RX VEGA 56
Slot 2 - N/A
Slot 3 - RX VEGA 56
Slot 4 - N/A

NVMe location :

M 2_1 socket the lower one : Samsung 970 EVO ( Windows)
M 2_2 socket the top one : WD Black SN750 ( Catalina ---->CC copy from DIMM 2_2)

DIMM 2_1 left : Samsung 970 EVO ( 1TO Data)
DIMM 2_2 right : PNY XLR8 CS3030 ( 250 GB Catalina)

Yes really and no crash no freeze...phew...

Even Aquantia AQC107 10G works.

I will do more test tomorrow. Before calling for RMA.

Happy New Year and all the best for 2021.

 

[djlild7hina]

I made great progress yesterday and today: phew...
After three weeks of issues and failures : it was crazy for me after spending so much money and time going up the wall ...

But yesterday after the message from @djlild7hina about issue on Windows too, I planed to resolve the problems or return the Mobo too.

After doing more tests, no problem on Windows but High Sierra was unstable.

There was a fact that my bios clock was always frozen : whatever the real cause: It was sure to be a First explanation to my problem : yeah : when i removed the battery, weeks ago, to clear CMOS and after putting it back for some reason, it was dead ( amazing on a brand new mobo ...).
So a new CR2032 and lets go...

Second explanation : after clear CMOS a first time I used the Bios Flash Back Button : classic procedure but I remember to use the Bios renamer [EG R6EE.CAP] instead doing that manually...
This time I reflashed the two bioses but i renamed the bios files manually.

After adjusting the bios settings, tried with macOS Catalina.

And the light has come : I renewed my EFI and my USBmap for iMacPro 1,1 : and at the first attempt , it was done.

Pcie allocation BIOS config: 16x/8x/4x

Slot 1 - RX VEGA 56
Slot 2 - N/A
Slot 3 - RX VEGA 56
Slot 4 - N/A

NVMe location :

M 2_1 socket the lower one : Samsung 970 EVO ( Windows)
M 2_2 socket the top one : WD Black SN750 ( Catalina ---->CC copy from DIMM 2_2)

DIMM 2_1 left : Samsung 970 EVO ( 1TO Data)
DIMM 2_2 right : PNY XLR8 CS3030 ( 250 GB Catalina)

Yes really and no crash no freeze...phew...

Even Aquantia AQC107 10G works.

I will do more test tomorrow. Before calling for RMA.

Happy New Year and all the best for 2021.

View attachment 503602

Yay! You made progress! I did too lol. I took out my 3090 and put it in my z390 build for windows and it also randomly rebooted when playing breath of the wild on Cemu emulator. So it wasn’t a motherboard issue but not sure if my issue is caused by the 3090 or some windows thing. I changed some settings and it has been solid today *knock on wood*

So I’m basically back to where I was before this whole thing happened

Sleep still isn’t working though.. seems to be hit or miss and I think it’s waking up from handoff?