>>106950652
>selling pallets of 5090s to a foreign adversary
If you actually understood the scale of AI training & inference, you would know those toys mean fuck all because they don't support NVLink.
They're not training missile guidance systems with a pallet of 5090s.
4chan Search
3 results for "3a45934b4a35749eb7e545f3ac75d0b1"
>>105769966
You might want to check the spec sheets, I don't believe the RTX pro 6000 blackwell nor its server edition supports NVLink.
https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/quadro-product-literature/workstation-datasheet-blackwell-rtx-pro6000-x-nvidia-us-3519208-web.pdf
They stopped that after RTX 30/Ampere.
L40S mentioned in the pic is last-gen's equivalent to the 6000 Blackwell Server card.
You might want to check the spec sheets, I don't believe the RTX pro 6000 blackwell nor its server edition supports NVLink.
https://www.nvidia.com/content/dam/en-zz/Solutions/design-visualization/quadro-product-literature/workstation-datasheet-blackwell-rtx-pro6000-x-nvidia-us-3519208-web.pdf
They stopped that after RTX 30/Ampere.
L40S mentioned in the pic is last-gen's equivalent to the 6000 Blackwell Server card.
>>105682476
>Of course they shipped more when 5000 series launched. 3/4 of it going to AI servers.
Serious AI projects don't want those GeForce cards because NVIDIA deliberately dropped NVLink support.
They don't even want the repackaged variants of these GPUs like the L40/S.
These cards are hardly worth it for even AI startups if they can rent some more performant & cost-efficient servers.
>Of course they shipped more when 5000 series launched. 3/4 of it going to AI servers.
Serious AI projects don't want those GeForce cards because NVIDIA deliberately dropped NVLink support.
They don't even want the repackaged variants of these GPUs like the L40/S.
These cards are hardly worth it for even AI startups if they can rent some more performant & cost-efficient servers.