2 ms·
I don't think we're talking about the same thing. I'm probably not helping by being sloppy distinguishing between opcodes and µops. But I'm referring to the h
by nkurz 12y ago
I don't think we're talking about the same thing. I'm probably not helping by being sloppy distinguishing between opcodes and µops. But I'm referring to the hardware decoded µop cache that mostly replaced the "loop stream detector" in Intel processors post-Nehalem. No context switching, no OS support, rather hardware optimizations targeted at the interface between the outward facing x64 (CISC) instruction set and the post-encoding (RISC) µop driven monster within.
It would be a box along the same path as the one labelled "uOP Cache Hit Logic":
http://pc.watch.impress.co.jp/video/pcw/docs/601/851/p16.pdf http://pc.watch.impress.co.jp/video/pcw/docs/601/851/p16.pdf
- vardump 12y agoI think that's what the SIMD instructions largely do. Like AVX-512 when talking about Skylake. A more direct interface to the "monster within". The monster got bigger, so there's a need to feed it with bigger chunks of work. µops are just an intermediary translation format. The point in JITting is to have the CPU translating the intent better to actual µops.