[PATCH] D113798: Add loop unrolling and peeling preferences for RISCV

Wed Dec 8 12:15:57 PST 2021

jrtc27 added a comment.

In D113798#3179913 <https://reviews.llvm.org/D113798#3179913>, @luismarques wrote:

> In D113798#3174400 <https://reviews.llvm.org/D113798#3174400>, @asb wrote:
>
>> - @luismarques reports this is performance neutral for Embench and Coremark on Ibex.
>
> I think that was with an old version of this patch. With the current patch (now reverted) the numbers are:
>
> CoreMark O3 <https://reviews.llvm.org/owners/package/3/>: +11.4% perf, 2.88 times the size
> CoreMark O2 <https://reviews.llvm.org/owners/package/2/>: +9.36% perf, 2.88 times the size
> Embench O3 <https://reviews.llvm.org/owners/package/3/>: no perf change, 23.6% size increase
> Embench O2 <https://reviews.llvm.org/owners/package/2/>: no perf change, 28.6% size increase
>
> No changes for Os and Oz.
> That's almost tripling the CoreMark size.

CoreMark is a pretty tiny benchmark that fits in the L1 <https://reviews.llvm.org/L1> I$ of many processors, so that's probably not hugely surprising, though I suspect you could get a similar performance increase with a much more targeted segment of unrollings...

Repository:
  rG LLVM Github Monorepo

CHANGES SINCE LAST ACTION
  https://reviews.llvm.org/D113798/new/

https://reviews.llvm.org/D113798