If it was overspecced before, then that means it was using parts more expensive than it needed to. Nobody makes RAM that is slower and also more expensive for the same capacity. Logically, this should translate to lowered prices for the GPUs using the cheaper parts.
Honestly NVIDIA shareholders don’t give a shit about the discrete GPU market as long as NVIDIA is able to overcharge the datacenters and reek of insane profits.
Unfortunately, the crypto boom normalised those prices and now there is no turning back.
For all we know, they used overspecced RAM because it was what was available in the quantities needed, or they got a good price from the supplier - which is something that has specifically happened with hardware I’ve worked on before. Again, we don’t actually know the specific pricing details. Higher speed does not inherently mean higher cost.
If it was overspecced before, then that means it was using parts more expensive than it needed to. Nobody makes RAM that is slower and also more expensive for the same capacity. Logically, this should translate to lowered prices for the GPUs using the cheaper parts.
But think of Nvidia’s shareholders! /s
Honestly NVIDIA shareholders don’t give a shit about the discrete GPU market as long as NVIDIA is able to overcharge the datacenters and reek of insane profits.
Unfortunately, the crypto boom normalised those prices and now there is no turning back.
For all we know, they used overspecced RAM because it was what was available in the quantities needed, or they got a good price from the supplier - which is something that has specifically happened with hardware I’ve worked on before. Again, we don’t actually know the specific pricing details. Higher speed does not inherently mean higher cost.