thats what the article assuming Nvidia could do. Fusing off 4090 defeat the purpose of sending AD102 to China, if they do that they might as well send only AD103.
nothing stop nvidia underclocking 4090 by massive number & sell a clock speed limit 4090. I think Nvidia is going to do the unclocking path, letting the Chinese to OC it.
feels like only Qualcomm & Huawei among the 4 that is well ahead of everyone else.
At this point he might be moving even chip design out of US to avoid getting caught in geopolitical mess.
yup his GPU is actually holding back the i5 3470 lol
Something like 1650 will fix this problem or a Radeon RX560/RX460.
I am typing this post on my 2500K, and that is still my "fastest" desktop PC. :o
Nvidia and AMD should have allow AIB freedom to put whatever amount of Vram they like.
It use to be like this. We had 4GB GTX960, R9-285, gtx680/770, 8GB 290X.
Laptop use to have double of vram back in old days. Nowadays they have less than Desktop ones... zzz
I refuse to buy a 4060/4070 laptop because of 8GB vram, Laptop screen has since moved to 2560x1600p. There is no way 8GB is enough for a 1600p screen.
these companies they pre-order them, they arent like your average miners that go to AIB order a bulk like black market.
5950X3D
useless if it is only 1 chiplet, I dont think I want to deal with OS to decide which core I should use. 5950XD need to have 2 chiplet with 3d cache for scheduling simplicity.
It is kinda shame the ethernet for transferring digital AV signal never really take off.
We got HDMI, display port instead. It would be so much more simple if we can run internet & transfer AV in 1 single cable.
It is kinda shame the ethernet for transferring digital AV signal never really take off.
We got HDMI, display port instead. It would be so much more simple if we can run internet & transfer AV in 1 single cable.
ELI5,
why the cache/memory always seems to be on top of the processing chip, not bottom?
There is no point fusing off 4090 permanently down to 100CU, that is just wastage of silicon. At that point Nvidia might as well save the trouble, just ship a fully enabled super clock AD103 which is at 84CU, the performance is already very close.