Commits on May 6, 2024

1. [LoopUnroll] Clamp PartialThreshold for large LoopMicroOpBufferSize …

   The znver3/znver4 scheduler modules are outliers, specifying very
   large LoopMicroOpBufferSizes at 512, while typical values for
   other subtargets are on the order of ~50. Even if this information
   is micro-architecturally correct (*), this does not mean that we
   want to runtime unroll all loops to a size that completely fills
   the loop buffer. Unless this is the single hot loop in the entire
   application, the massive code size increase will bust the micro-op
   and instruction caches.
   
   Protect against this by clamping to the default PartialThreshold
   of 150, which is the same as the default full-unroll threshold
   and half the aggressive full-unroll threshold. Allowing more
   partial unrolling than full unrolling is certainly non-sensical.
   
   (*) I strongly doubt that this is actually correct -- I believe
   this may derive from an incorrect reading of Agner Fog's
   micro-architecture guide. The number 4096 that was originally
   used here is the size of the general micro-op cache, not that of
   a loop buffer. A separate loop buffer is not listed for the Zen
   microarchitecture. Comparing this to the listing for Skylake, it
   has a 1536 micro-op buffer, but only a 64 micro-op loopback buffer,
   with a note that it's rarely fully utilized. Our scheduling model
   specifies LoopMicroOpBufferSize of 50 in that case.

   

   nikic committed May 6, 2024
   Configuration menu

   • View commit details
   • Copy full SHA for 5c7fbce
   • Browse repository at this point

   Copy the full SHA

   5c7fbce View commit details

   Browse the repository at this point in the history