PC Graphics

[Ollonii]

I have two Radeon Pro W6600s in a Dell T7920 Workstation. Each W6600 has two HP U28 4k HDR monitors attached via DisplayPort. Most of the time, everything works very well but occasionally (about 25% of the time), when the system wakes from sleep, one of the four monitors doesn't wake and remains undetected by the system. The only way to get the "missing" monitor back is to remove and restore power to the monitor (shutting it off is not sufficient). The monitor that doesn't wake in these cases varies, seemingly randomly, and might be on either of the cards.

For reference, the O/S is Windows 11 Enterprise Build 22000.493 and I've tried both sets of Radeon Pro drivers (Radeon™ Pro Software for Enterprise 21.Q4 and Adrenalin 22.2.1). The system has two Intel 4210 CPUs and 192 GB memory and both video cards are on CPU1. The monitors (and cables) are certified for DP 1.2 whereas the W6600 is DP1.4 (which shouldn't matter).

Has anyone else experienced this behavior? Results can be unpredictable when a monitor shuts off during a design session so this is more than inconvenient. Thanks for any help.

[fsadough]

1. On which PCIE slots are the W6600 GPUs installed?
2. May I ask why you need 2x W6600?
3. What are the resolution, refresh rate and bpc you are running on the monitors?
4. What is the DP version set to on your monitor?
5. Can you provide an AMDZ Report?

- Please extract the amdz-v295.zip available from https://we.tl/t-lLPEZDqLEk
- Run amdz.exe file as an Administrator
- Select “Save All“ and “TXT“ as the output format
- Click on the blue button to save the report
- The .txt file will be saved in the same folder where you extracted the zipped file

[Ollonii]

Thanks very much for your reply and help.

The W6600s were, inexplicably, in slots 3 and 5. I have moved them to slots 2  and 4. The monitors are all DP 1.2 and are set to 3840 x 2160 @ 60 Hz and color is 10 bpc. The 2xw6600 was based on my previous experience with Tesla adapters which performed better when rendering to two (or more) screens at a time (that is, two or more render processes on unrelated tasks) when there were two (or more) adapters. Is that not the case with these GPUs? It would be great to consolidate to one W6800, instead.

I've run the AMDZ report, which is quite detailed and large. Is there a preferred way to send it to you?

Thank you again for the help.

[fsadough]

You can use WeTransfer to send the report back to me. On multi-gpu for rendering purposes you have to make sure that your application supports it.

[Ollonii]

Please see the output file here: https://1drv.ms/t/s!AmxiUwTHUl_bnmX2Zr-2OIfQBDsV?e=QmAWab.

As for the GPUs, it's actually two different programs running at the same time and rendering to different monitors.

[fsadough]

Please try to flash the Video BIOS of your W6600 GPUs. See instructions, VBIOS and AMDVBFLASH tool from the link below:

Download link
https://we.tl/t-czDqVM6QTW

[Ollonii]

Thank you very much. Both adapters have been updated to 113-D5330100-101. It will take a little time to verify that the issue has gone away, but I am hopeful! I certainly appreciate the effort.

By the way, running the installed amdvbflashwin.exe produces the error "Critical file missing. Please download and run video BIOS flash files again." and the program exits. I thought you might want to know in case you aue the package again.

One other question, if I may, please. You suggested earlier that I might do better with a single adapter with better muscle. Is this still true given the (independent) multi-process, multi-render I am doing? Would I be better off returning the two W6600s and getting a single W6800?

Thank you, again, for all your help.

[Ollonii]

Thank you for the reply.

I updated the VBIOS on both GPUs to 113-D5330100-101 yesterday. Everything seemed to work well during the day but, this morning, the same problem occurred. I can't tell yet whether this will move randomly amongst the four monitors (as it did before) but the issue appears to be the same.

I also tried to post a reply yesterday, but it seems to have been lost. I was wondering if I'd be better off returning the two W6600s and getting a W6800? This might solve my problem, but it doesn't help you with the W6600!

[Ollon2]

For some reason, the forum will not allow me to respond so I've created another ID.

Thank you, yet again, for your help. 

I updated the VBIOS on both GPUs to 113-D5330100-101 yesterday. Everything seemed to work well during the day but, this morning, the same problem occurred. I can't tell yet whether this will move randomly amongst the four monitors (as it did before) but the issue appears to be the same.

I also tried to post a reply yesterday, but it seems to have been lost or rejected. I was wondering if I'd be better off returning the two W6600s and getting a W6800? This might solve my problem, but it doesn't help you with the W6600!

[Ollonii]

For some reason, the forum will not allow me to respond to the original post. I've tried several times over two days and the replies keep getting discarded. I have no idea what I'm doing incorrectly.
 
Thank you, yet again, for your help. 
 
I updated the VBIOS on both GPUs to 113-D5330100-101 yesterday. Everything seemed to work well during the day but, this morning, the same problem occurred. I can't tell yet whether this will move randomly amongst the four monitors (as it did before) but the issue appears to be the same. (02-22-22: this evening, one monitor on each card refused to wake from sleep. This is the first time that has happened.)
 
I also tried to post a reply yesterday, but it seems to have been lost or rejected. I was wondering if I'd be better off returning the two W6600s and getting a W6800? This might solve my problem, but it doesn't help you with the W6600!

[Ollonii]

For some reason, the forum will not allow me to respond to the original post. I've tried several times since Saturday and the replies keep getting discarded. I have no idea what I'm doing incorrectly.
 
Thank you, yet again, for your help. 
 
I updated the VBIOS on both GPUs to 113-D5330100-101 yesterday. Everything seemed to work well during the day but, this morning, the same problem occurred. I can't tell yet whether this will move randomly amongst the four monitors (as it did before) but the issue appears to be the same. (02-22-22: this evening, one monitor on each card refused to wake from sleep. This is the first time that has happened. 02-23-22: today, one card failed completely on two separate occasions.)
 
I've given up on running two cards and now have only one W6600 in slot 4 running all 4 monitors. Rendering seems slower but, so far, no issues with waking from sleep.

[Ollonii]

Please accept my apologies for all of the preceding posts. For some reason, the site was saving them up but not displaying them. I tried several times because I didn't want you to feel I was ignoring your suggestions. Suddenly, several days of posts appeared!

It's been a day or two since I removed one of the W6600s. I just came back from lunch and one of the monitors did not recover from sleep. I was able to do some testing and it appears the monitor did not recognize the timing of the signal from the W6600. First, it could not auto-detect the resolution and second, it could not detect the v-sync properly. It appeared to be a non-standard frequency. One thought I had is that the levels out of the card may be too close to the switching levels on the monitor but I couldn't test this. I also don't know how much hysteresis both devices support.

I have no idea if this helps, but thought I'd mention it.

[fsadough]

Can be your cable or monitor