2 cards, both locked in compute only mode.
Why put 4 dp ports on the cards then lock them into compute only mode, and as far as I can tell the only way to change it (smi doesn’t work) is to use a mode change tool that nvidia lock away for devs only.
Hi and welcome. Can you give me more context here? You purchased two RTX 6000 Pro Server cards? Where did you buy them from?
I got them from a company call scan in the uk
I emailed PNY, to their credit they sent me the device mode selector tool, I changed them to 8gb bar setting, rebooted, then ran nvidia-smi.exe -fdm 0 and would you believe it, works just fine now. (I did also have to disable the CSM module in bios/boot as they were giving a code 12 error)
Its pretty poor showing with the cost of these cards that they are locked in compute mode, and no clear path from NVidia on how to resolve it. I’ve wasted weeks and a lot of money on upgrading hardware trying to figure out what the problem was)
It looks like I’m not the only person who has had this issue on the forum, and people have even RMA what were actually functional cards. So I doubt I am the only person, so maybe this post saves someone the headaches this has given me
Interesting. This is the first I am hearing about it. Since the cards are so specialized and expensive, could you have not request that they be shipped in Graphics Mode first ?
Turns out nvidia is happy to sell enterprise level cards with the price tag attached . But refuses to support you unless you are a corporation or university.
For the record, I did call the distributor to inform them of my issues and the solution, so hopefully they can help others.
Pny also informed me that nvidia dictate that since ada 6000 they are locked into compute mode.
This is fine but at least inform the buyer. And don’t lock the solution away.
I still haven’t received the device select mode tool from nvidia. If it wasn’t for pny my cards would have been shipped back.
I have a use case I can’t discuss, but I need the graphic output. This has cost me weeks of my time and staff time, and massive upset from my customer, I’m now potentially 16k out of pocket because they are so upset with me that I’m now personally the owner of 2 expensive cards I don’t have a use for.
Nvidia need to get there sh#t together and realise that not everyone who needs these cards are already a huge enterprise. Especially when doing things under nda so I’m not explaining my use case.
Again I will say pny were the hero of the day and gave me the tool to resolve the issue.
Nvidia needs to support their products. Especially the bleeding edge consumers who do your beta testing for you.
Thanks
Col
Please explain why my enquiry is timed out after 30 days when I have only just this week requested direct support from nvidia?
I am sorry for the bad purchase experience with PNY. I will try to pass this along to someone here, but this is really a forum for OMNIVERSE. Are you an OMNIVERSE user? If so, are you an Enterprise Omniverse user?
Can you give me the precise card model number?
No PNY where excellent, its Nvidia who need to learn from them how to help customers.
Ok. I am trying to help, as best I can, even though it is not Omniverse and not part of my scope. But in order to do so, I would so specifics.
I will close this thread then.