MP 7,1 Dual MPX W6800X Duos Crashing System

[awkward_eagle]

Getting system crashes, freezing, and random restarts with 2x W6800X Duos even when doing nothing. Same system worked fine with 2x Radeon Pro Vega IIs and the Duos worked fine in another system. Seems to only be a problem when monitors plugged into the top most GPU but haven't tested it enough to be definitive. Reformatted the system just to rule it out but had no effect. Expansion Slot Utility not throwing any warnings. Pool A at 100% and Pool B at 94%.

System Specs :
16c Xeon
2x W6800X Duos
192GB Ram
Ventura 13.2.1
XDR Display

All slots filled : order is top to bottom

Sonnet USBC card
Blackmagic Intensity Pro 4K
Afterburner
W6800X Duo
W6800X Duo

Trying figure out if this is a software or hardware thing, and if so, is it the motherboard or the card(s). Any help would be appreciated.

Edit : Forgot to mention I've tried swapping the card placement and its always the top slot. On both systems having the monitor plugged into the bottom card causes Photoshop to crash.

 

[nisse32]

I also have system crashes, freezing, and random restarts with a 6800XT...

System Specs :
24c Xeon
RX 6800XT
96GB Ram
Ventura 13.2.1
LG 4K TV
Boot drive: Samsung SSD 980 Pro on PCI-card

See logfile below, I dont know how to read the logs, but maybe it indicates some kind of graphics problem? This computer is newly installed so I cannot connect the problems to a specific upgrade...
-----------------------------------------------------------
Crashed Thread: 0 ws_main_thread Dispatch queue: com.apple.tcc.preflight.kTCCServiceScreenCapture

Exception Type: EXC_CRASH (SIGKILL)
Exception Codes: 0x0000000000000000, 0x0000000000000000

Termination Reason: Namespace WATCHDOG, Code 2
checkin with service: WindowServer returned not alive with context:
unresponsive work processor(s): WindowServer main thread
40 seconds since last successful checkin, 130 total successful checkins since 1360 seconds ago (0 induced crashes)
checkin with service: WindowServer returned not alive with context:
unresponsive work processor(s): WindowServer main thread
40 seconds since last successful checkin, 130 total successful checkins since 1360 seconds ago (0 induced crashes)

 

[awkward_eagle]

I also have system crashes, freezing, and random restarts with a 6800XT...

System Specs :
24c Xeon
RX 6800XT
96GB Ram
Ventura 13.2.1
LG 4K TV
Boot drive: Samsung SSD 980 Pro on PCI-card

See logfile below, I dont know how to read the logs, but maybe it indicates some kind of graphics problem? This computer is newly installed so I cannot connect the problems to a specific upgrade...
-----------------------------------------------------------
Crashed Thread: 0 ws_main_thread Dispatch queue: com.apple.tcc.preflight.kTCCServiceScreenCapture

Exception Type: EXC_CRASH (SIGKILL)
Exception Codes: 0x0000000000000000, 0x0000000000000000

Termination Reason: Namespace WATCHDOG, Code 2
checkin with service: WindowServer returned not alive with context:
unresponsive work processor(s): WindowServer main thread
40 seconds since last successful checkin, 130 total successful checkins since 1360 seconds ago (0 induced crashes)
checkin with service: WindowServer returned not alive with context:
unresponsive work processor(s): WindowServer main thread
40 seconds since last successful checkin, 130 total successful checkins since 1360 seconds ago (0 induced crashes)

I should have thought to add the log but forgot. But having read through it before submitting it read the same. Still no idea what it means.

 

[prefuse07]

Doubt this will make any difference, but have you tried a PRAM and SMC reset?

 

[awkward_eagle]

Doubt this will make any difference, but have you tried a PRAM and SMC reset?

PRAM and SMC reset didn't work.

As a test I plugged the monitor into the bottom card and left it to go overnight without a problem. Before responding to this test I did both resets. After restart I plugged the monitor into the top card and instantly got a freeze. Mouse could move but not click on anything.

Ran Geekbench Metal on each card and they all worked fine.

Monitors back in the bottom card for now.

 

[smckenzie]

I had the same photoshop issue with my Dual Duo's, drove me crazy until like you I figured out plugging the display into the lower Duo worked. However, I started getting issues where when the Mac went to sleep or was left for a while it would hard shut down. Can't remember what the error was now but it has caused problems before on prior OS's.

In the end I reverted back to Monterey from Ventura and did a SMC reset. Since then, no issues. Photoshop now doesn't care what port I use or how many displays.

YMMV.

 

[awkward_eagle]

I had the same photoshop issue with my Dual Duo's, drove me crazy until like you I figured out plugging the display into the lower Duo worked. However, I started getting issues where when the Mac went to sleep or was left for a while it would hard shut down. Can't remember what the error was now but it has caused problems before on prior OS's.

In the end I reverted back to Monterey from Ventura and did a SMC reset. Since then, no issues. Photoshop now doesn't care what port I use or how many displays.

YMMV.

It's reassuring to hear it's a likely a software problem. Dragging this system to the Apple Store isn't something I was looking forward to. These systems never sleep so fingers crossed it holds overnight. I was having the Photoshop problem in Monterey too so that seems more like an adobe specific bug.

 

[SDAVE]

Try using another thunderbolt port that's not on any GPU
macOS should route the signal from your MPX module(s) to another TB port

Also a note on using the latest versions of any OS in production machines. It's always a rule of thumb to stay a year behind, that's why I am still on Monterey 12.6.3 on the 7,1

 

[tommy chen]

ventura has huge problems, and since 13.2 they are even bigger and most not present in 13.1.
nevertheless, monterey is much more stable.

 

[awkward_eagle]

Never thought of using a non gpu port for the monitor. Those are filled with thunderbolt raid drives but I expect those work fine on the MPX card ports.

Haven’t used the 13.2 beta. What kind of problems have been found that are worse than 13.1?

Edit : Used non GPU port and it crashed the Lock Screen but recovered. Seems normal now. Looking at the crash report , Expansion Slot Utility, and the System Profiler it tried to use the bottom slot ( slot 1 ) but routed to the wrong GPU before correcting itself to the same slot but a different GPU on the same module. So it appears macOS prefers the bottom slot for some reason. Thankfully it's the more stable of the two.

Edit 2 : missed the Ventura versions. Read it as Beta 13.3

 

[SDAVE]

Never thought of using a non gpu port for the monitor. Those are filled with thunderbolt raid drives but I expect those work fine on the MPX card ports.

Haven’t used the 13.2 beta. What kind of problems have been found that are worse than 13.1?

Edit : Used non GPU port and it crashed the Lock Screen but recovered. Seems normal now. Looking at the crash report , Expansion Slot Utility, and the System Profiler it tried to use the bottom slot ( slot 1 ) but routed to the wrong GPU before correcting itself to the same slot but a different GPU on the same module. So it appears macOS prefers the bottom slot for some reason. Thankfully it's the more stable of the two.

Definitely looks like a Ventura issue. Try filing a report with Apple using the Feedback app and explaining your situation.

In the meantime, since you probably don't want to downgrade to Monterey, I would suggest sticking with what's working for now until Apple fixes this issue with Ventura.

FYI they do read feedback reports, so definitely worth submitting a ticket. The Mac Pro is on a low priority list for them, so your submission will probably stick out.

 

[awkward_eagle]

Definitely looks like a Ventura issue. Try filing a report with Apple using the Feedback app and explaining your situation.

In the meantime, since you probably don't want to downgrade to Monterey, I would suggest sticking with what's working for now until Apple fixes this issue with Ventura.

FYI they do read feedback reports, so definitely worth submitting a ticket. The Mac Pro is on a low priority list for them, so your submission will probably stick out.

Using the Feedback app looks like it only works with the Beta Software and App Developer Programs. I'm not part of either. Is there any other place to submit this not involving these programs?

Edit : Forgot about feedback.apple.com

 

[SDAVE]

Using the Feedback app looks like it only works with the Beta Software and App Developer Programs. I'm not part of either. Is there any other place to submit this not involving these programs?

Edit : Forgot about feedback.apple.com

No it works in the public release of macOS too. Just open Feedback Assistant in Applications and login and submit a detailed report with screenshots and logs if possible.