Quote from: Fabio on Today at 00:09:40Well that's expected. Why launch new card on gamer market... When they are doing a killing selling to companies. Bet they have the card ready for months now, but 0 interest in launching it.
Any chip that went for that product line would be a chip they couldn't sell for 10 times the markup if for a company/datacenter
A GB202 chip based RTX PRO 6000 (96 GB VRAM) workstation GPU costs like 15-20,000 $ right now, while the RTX 5090 (32 GB VRAM) is 5,000. Both doubled in price since release. So indeed, the same GB202-based workstation GPU can be sold for a multiple times higher price.
Same goes for GB203-based RTX 5070 Ti and RTX 5080 vs RTX PRO 4000 and RTX PRO 4500.
And since AI fuels the demand for these high VRAM per volume or per Watt TDP workstation GPUs, it is clear where NVIDIA wants these GB202, GB202 chips to go.
Or NGREEDIA could release the Super refresh in a 0.25 to 1 years and RTX 60 in another 1-2 years. But since all the AI folks are waiting for 5070 Ti 24 GB VRAM GPUs, NGREEDIA is, of course, going to wait as long as they can (at latest until AMDone catches up to at least using GDDR7).
The lowest common denominator for consumers is to wait until every datacenter has been built / saturated and the demand stops or decreases (silicon wafers are still being consumed, it is just that a lot of made chips are going into datacenters, so, let's hope it saturates at some point).
Another hope is that China releases a 4070 / 5070 / 5070 Ti equivalent in performance (even if they are less energy efficient by 1 full node shrink (25%)) and uses 3 GB GDDR7 chips (=18 GB VRAM and 24 GB VRAM, just like a 5070 Super and 5070 Ti Super would be).
English
Español
Français
Italiano
Nederlands
Polski
Português
Русский
Türkçe
Svenska
Chinese
Magyar