Here guy's but doesn't this also mean that the benchs weve been getting on Epyc 32 core Windows Server are all severely hampered by the same bad scheduler ?
I mean it's not a small discrepancy, a lot of the workloads are 2.5x faster in Linux.
I can't wait to see the actual numbers for a 32 Core Epyc Windows Server against Xeon when this is fixed...
Expect the gap to increase a lot... well 2.5x in certain workloads...
If this makes all the previous benchmark results wrong on Threadripper and Epyc...well lol This is huge.
I wonder what the 64 Core is going to look like, ridiculous I'd say