It was smart of Intel to release the compiler well in advance of the processor. I wonder if Intel has released chips to important software vendors too.
With SIMD instructions with low latencies, you can't really get much of a performance improvement by speeding up that instruction. There's no way that you can catch up to a useful instruction that replaces half-a-page of code with IPC improvements - you need that new instruction. Intel can thank IBM for the assist. Or maybe Apple.
Market Data 
Markets 
AI
