According to a new rumor, AMD’s RX-7000 flagship should reach 92 TFLOPS in FP32 tasks. That would be almost exactly four times the value of the RX 6900 XT.
With the new Radeon and Geforce graphics cards of the RX 7000 and RTX 4000 types, there will definitely be a leap in performance and efficiency by the end of the year. How big this will really be in the end is not yet entirely clear. According to the well-known leaker @Greymon55, there could be an immense performance boost at least for AMD’s flagship.
Even faster than expected?
Last November, Greymon55 published some information about Navi 31, AMD’s largest GPU. At that time there was talk of 15,360 shaders running at 2.4 to 2.5 GHz and thus achieving around 75 TFLOPS of computing power (FP32). This last value has now been corrected. In a new tweet, Greymon55 speaks of 92 TFLOPS.
For comparison: AMD’s current flagship, the RX 6900 XT, has 23 TFLOPS ex works. Purely based on the computing power, a potential Radeon RX 7900 XT would be almost exactly four times faster than the previous flagship. With the performance value mentioned in November, however, it would have been “only” a factor of 3.26.
Also exciting: AMD Radeon RX 7000: rumors about a release at the turn of the year and the state of the speculated technology [Update]
If the new performance specification is correct, it is unclear how AMD intends to achieve the said 92 TFLOPS. In this context, the Videocardz website brings a modified set of instructions into play, which is intended to increase performance with the same number of shaders. Such a change should actually have been known to AMD in November. However, a more short-term change in performance would be possible via the GPU clock. This would now have to be around 3.0 GHz.
Source: via Videocardz
The post Leaker speaks of four times the computing power appeared first on Gamingsym.