Earlier quoted context omitted.
The fact that the entry-level card can run its clock 30% faster than the same line's top-end offering supports the story pretty solidly, IMO. Not supporting 8 GB of vram can't have given them enough breathing room to casually run the clock that fast, so this has to have been decided way up front. Like the article says, it remains to be seen whether this will also make the card bad at gaming, but I think that AMD at l…
Isn't it a power budget and heat dissipation thing? For example the Xbox Series X and Playstation 5 GPUs have pretty similar architectures, but the PS5 has fewer compute units at a higher frequency. Surely Microsoft would have loved to clock theirs as high as Sony. PS5: 10.28 Teraflops, 36 Compute Units running at 2.23GHz (variable frequency) XSX: 12.11 Teraflops, 52 Compute Units running at 1.825GHz (fixed)
Yes, this seems to be what's happening here. Less power used on memory bandwidth, means more is available for other things. This card is not so much "bad at mining" and more like trying to design for a different niche that might be underserved by all the "good at mining" GPU's.