I igor_kavinski Quasi-regular Joined Aug 15, 2025 Messages 412 Reaction score 227 Tuesday at 12:17 PM #1 https://www.overclock.net/posts/29548227/
Grimlakin Forum Posting Supreme Joined Jun 24, 2019 Messages 10,174 Reaction score 6,554 Tuesday at 7:10 PM #2 That could be ininteresting. I wonder what the science behind that is or is it just a messed up measurement not dividing the mhz by more nodes.
That could be ininteresting. I wonder what the science behind that is or is it just a messed up measurement not dividing the mhz by more nodes.
I igor_kavinski Quasi-regular Joined Aug 15, 2025 Messages 412 Reaction score 227 Tuesday at 11:48 PM #3 Grimlakin said: I wonder what the science behind that is Click to expand... The single E-core in the cluster gets to party with the cache all by itself, allowing it to get its highest throughput without running into cache bandwidth contention issue. May also have something to do with fewer ring stops to deal with.
Grimlakin said: I wonder what the science behind that is Click to expand... The single E-core in the cluster gets to party with the cache all by itself, allowing it to get its highest throughput without running into cache bandwidth contention issue. May also have something to do with fewer ring stops to deal with.