Means the margins are obscene because the volume is tiny.
No it means the volume is tiny.
You can't just scale hybrid bonding capacity 10 times, that's a ton of equipment.
no.
You really need to talk to some actual real SIMD people.
Well neither will ever have any SIMD code running on them.
no it didn't, FMA-wise we went from 1 to 2 where it matters and that's it.
Even Knights* thingamajigs had 2 FMA units for very obvious reasons.
Yeah lol they do it all the time later in the product lifecycle.
You won't believe the places you can find AD103 in.
Some NV SKUs become mysterymeat in truest sense of that word.
GB206 is a 128b die so no.
Efficiency is not the design point for anything RDNA4.
You can just like disable IMC segments you know.
Tonga is like the most notable example of that in AMD lands.
The full die was 384b but what shipped outside of Apple lands was 256b. Too bad!
Uh no brother that's not how that works.
It adds 4 cycles and half speed caches are just not happening, that's terribly slow under all circumstances.
Active cache power just isn't all that high.
You want less NUCA domains, not more.
Not enough SoIC wafer starts in the world for that.
Not really you just want the least amount of NUCA domains possible, especially for cloud parts.
This site uses cookies to help personalise content, tailor your experience and to keep you logged in if you register.
By continuing to use this site, you are consenting to our use of cookies.