There’s a lot we’ve heard about AMD’s upcoming Big Navi graphics card: The number of compute units, ray-tracing capabilities, memory size as well as the TDP. However, the operating clocks have been all but uncertain up until now.
As per Patrick Schur, the Navi 21 XT GPU will run at a Game Clock of 2.4GHz, with a TGP of just 255W. Keep in mind that Game Clock is different from the Boost Clock which is a bit higher (likely 2.55GHz here). It’s the average clock speed sustained by the GPU during prolonged gaming sessions while the boost clock refers to the maximum achievable GPU core clock under load.
Update: As per certain sources, Navi 21 XT will feature a lower game clock, somewhere in the 2-2.1GHz range, with the boost clock being up to 2.4GHz. As you can see, the deltas between the base and boost clocks have grown significantly compared to Navi 10. This means that the vanilla models’ performance may vary quite notably from that of the higher-end factory OC variants. This will be a cause of concern among both reviewers as well as enthusiasts as you’re basically getting different performance tiers with the same GPU.
The TGP is also quite impressive. Considering that the Radeon RX 5900XT will be more or less on par with the RTX 3080 which draws as much as 400W under load, we’re looking at a much more power-efficient part drawing just under 300W.
At the same time, at higher resolutions, the memory bandwidth is more important than the core clocks which is why there’s a good chance that the RTX 3080 might be slightly faster at 4K. And then there’s the matter of ray-tracing. Some rumors state that part of the CUs will be reserved for the BVH acceleration, with ray-intersection units being limited to ray-triangle and ray-box intersection testing. If this is indeed true, then NVIDIA’s RTX 30 series GPUs will likely be notably ahead on ray-traced titles.
Last, but not the least, take this entire thing with a grain of salt. Just another week or so before we know what’s true and what’s made up.