If what Cerny says can be believed, then it just means they spent a substantial amount more money on cooling to maintain silence and performance.
It could be a disaster, since we have an idea of what RDNA's thermals are like at high clocks. But since it's RDNA2 maybe it's not that absurd with the improved efficiency.
Good point. But there's something else going on here as well. The Github leak for MS showed 56 CU's with only 52 active I believe, which is what people tend to expect- CU's being disabled is standard practice for chip yield reasons. But the MS RELEASE shows 52CU's as the FULL CU'S on the chip! Same for Sony..Github showed 36 active CU's and people assumed that meant at least 40 CU's on chip with 4 disabled.
So has AMD solved yield issues to the extent that manufacturers can have no "wasted" silicon in the form of disabled CU's? If so, that's pretty HUGE, isn't it? Again, unless my recall is failing me here. That could mean BIG things for manufacturers in the future.
I'm still wondering what MS split memory pool with a good chuck of "slow" memory will mean for the overall performance of the box. Again, I think that's one reason that "sources" all felt the two would have similar TF numbers. Real world performance MAY WELL IN FACT BE almost indistinguishable between them. Depending on these factors.
Bottom line, we need to see the games. But I'm confident we will all be happy with either system overall. I just think Sony went for smaller form factor as well as a different philosophy to get performance. Also expect they were continuing to target the $399 area where MS will likely have to be at at least $499 unless they take a large loss per unit. We'll see!