Insider Greymon55 has claimed in a new tweet that AMD’s flagship RDNA 3 GPU, the Navi 31, boasts up to 92 TFLOPS of FP32 performance. This is 17 TFLOPs higher than his previous estimation and four times greater than the Navi 21-based Radeon RX 6900 XT’s level of peak single precision compute performance, which is listed as 23.04 TFLOPs on red team’s official website. The implication is that AMD’s next flagship graphics card, the Radeon RX 7900 XT, could reach clocks of up to 3 GHz, a significant upgrade over the current flagship’s 2.25 GHz boost frequency.
AMD Navi31 flagship RDNA3 GPU could hit 92 TFLOPS of FP32 performance, four times more than Navi21 (VideoCardz)
Greymon55 earlier mentioned that there is a major change to the specs that were shared for more than a year, but has not explained what those changes are. However, should AMD keep 60 WorkGroup Processor (WGP) and 15360 Stream Processor configuration, such single-precision compute performance would require 2995 MHz GPU clock. However, it is very likely that similarly to NVIDIA and their Ada architecture, AMD will be introducing a matrix of different 32-bit instructions that result in higher floating point performance.
NVIDIA’s next generation of flagship graphics card may boast an even greater level of FP32 performance, however. Further speculation from Greymon55 and kopite7kimi have opened up the possibility that the AD102, green team’s Ada GPU for the GeForce RTX 4090 and other flagship models, could enable up to 100 TFLOPs of FP32 performance. The GeForce RTX 3090 Ti (GA102) delivers up to 40 TFLOPs of performance in comparison.
To be honest, I don't have much information about AMD. Maybe Lisa and Jensen's competition will give us a 100TFLOPS gaming war in a few months.
— kopite7kimi (@kopite7kimi) April 29, 2022
Maybe it's 90t+ vs 100t+🤔
— Greymon55 (@greymon55) April 30, 2022