Caporegime
![]()
Nvidia RTX 50 GB102 Blackwell early internal performance gains only in range of Pascal to Turing and not 2.6x from AD102, increased SM counts on the anvil
Nvidia's next gen RTX 50 Blackwell with purported 144 SMs isn't reaching ambitious performance gains in internal testing, according to latest rumors. Apparently, Blackwell's performance gains vis-à-vis AD102 are in the range of what we've seen from Pascal to Turing, but Nvidia wants more. This...www.notebookcheck.net
If it stays that way and perf doesn't hit the rumoured targets, then don't expect higher end 40xx prices to drop.
While people were ripping into AMD for their shallow RDNA3 gains by going chiplet, it was pointed out by a few tuber nerds that nvidia are going to hit a wall soon. The leap the 4090 did was decent, but the above would make sense; although its quite early to be believing anything tbh.
Nvidia is already using an improved TSMC 5NM,ie,4N so moving to TSMC 3NM is going to yield less improvements. Ada Lovelace saw a two node jump being on TSMC 5NM 4N from Samsung 8NM,which was more a 10NM class process node.
This generation AMD,no doubt decided to make sure they can get over the teething problems of chiplets.One of them is power and extra die area. You saw that when AMD actually started using MCM Zen/Zen+ commercial designs and it was only with Zen2/Zen3 they managed to finally get over the power problems.
Last edited: