While the GPU’s created aren’t used for gaming, the wafers that the dies are made with could’ve been allocated to produce the dies for consumer level graphics cards right?
So with datacenter GPUs (excellerators is the more accurate term, honestly), historically they were the exact same architecture as nVidia’s gaming GPUs (usually about half to a full generation behind. But in the last 5 years or so they’ve moved to their own dedicated architectures.
But more to your question, the actual silicon that got etched and burned into these datacenter GPUs could’ve been used for anything. Could’ve become cellular modems, networking ASICs, SDR controllers, mobile SOCs, etc. etc. but more importantly these high dollar data center GPUs are usually produced on the newest, most expensive process nodes so the only hardware that would be produced would be similarly high dollar, and not like basic logic controllers used in dollar store junk
While the GPU’s created aren’t used for gaming, the wafers that the dies are made with could’ve been allocated to produce the dies for consumer level graphics cards right?
So with datacenter GPUs (excellerators is the more accurate term, honestly), historically they were the exact same architecture as nVidia’s gaming GPUs (usually about half to a full generation behind. But in the last 5 years or so they’ve moved to their own dedicated architectures.
But more to your question, the actual silicon that got etched and burned into these datacenter GPUs could’ve been used for anything. Could’ve become cellular modems, networking ASICs, SDR controllers, mobile SOCs, etc. etc. but more importantly these high dollar data center GPUs are usually produced on the newest, most expensive process nodes so the only hardware that would be produced would be similarly high dollar, and not like basic logic controllers used in dollar store junk
Absolutely.