Radeon RX 7900, AMD Radeon RX 7900 GPUs would be the overclock beasts, as they would easily reach up to 3.0 GHz,

AMD Radeon RX 7900 GPUs would be the overclock beasts, as they would easily reach up to 3.0 GHz

Rumors of higher GPU speeds in the Radeon RX 7900 were wrong, but is there a chance we’ll see very high GPU speeds in custom RX 7900 graphics cards. According to Techpowerup, which was invited to AMD’s launch event, the company said it designed the Navi 31 GPU to easily hit 3.0GHz, which is well above official RX 7900 series speeds.

The AMD RX 7900 XTX has a GPU speed of only 2.5 GHz, while the RX 7900 XT only reaches 2.4 GHz. This is, of course, contrary to the rumors about 3.0 GHz clocks out of the box.

AMD Radeon RX 7000 Event: Live Interviews & Analysis

and now also YouTuber coreteks claims to have information on the subject. All this thanks to preliminary drivers, AMD’s partners see an overclocking potential of 3% with the custom Radeon RX 7900 series. That’s actually a bit low, but with the final drivers, surely that % is much higher.

Some people asked during the stream so I asked AIBs and they are saying about only 3% OC over the XTX with preliminary drivers (that have been used mostly for thermal testing, to be fair). If true then it looks like the XTX is already close to maxed out 😢 pic.twitter.com/X9E7E7EbSwIQ

– coreteks (@coreteks) November 4, 2022

Unfortunately, no AMD partners have presented their custom cards showing off their features. We only got to see an ASUS TUF, but it has no features yet.

And its arrival is very close, so AMD should clarify for example the number of Stream Processors for the Radeon 7900 series. Both AMD and partners now show 6144 cores for the top-of-the-line, while almost all media claim 12288.

The AMD RDNA3 architecture has a Dual Issue design, which can now execute not one but two FP32 arithmetic commands at the same time. What this means is that each CU can now do 128 FP32 calculations instead of 64 (RDNA2). To reach the advertised 61 TFLOP, you would have to multiply 6144 SP × 4 × 2.5 GHz ≅ 61 TFLOP or use the same method we use for all modern GPUs: 12288 SP × 2 × 2.5 GHz. Obviously, the second option should be more readable for the users.