Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[X86] LowerBITREVERSE - use AND+CMPEQ+MOVMSK trick to lower scalar types #92236

Draft
wants to merge 1 commit into
base: main
Choose a base branch
from
Draft

[X86] LowerBITREVERSE - use AND+CMPEQ+MOVMSK trick to lower scalar types #92236

wants to merge 1 commit into from
+518 −171

Commits on Jul 17, 2024

  1. [X86] LowerBITREVERSE - use AND+CMPEQ+MOVMSK trick to lower i8/i16/i3… 

    …2 scalar types

By splitting each byte of a scalar type into 8 copies, in reverse order, we can then test each one to see if the bit is set and then use PMOVMSKB to pack them back together to get the bit reversal.

So far I've only managed to make this worth it for i8/i16 with SSSE3 (PSHUFB), i32 with AVX2 (AVX1 would be worth it if we can force all 256-bit operations to be split), and i64 with AVX512BW (which can use VPTESTMB).

Fixes llvm#79794
    @RKSimon
    RKSimon committed Jul 17, 2024
    Configuration menu
    Copy the full SHA
    8cb57cb View commit details
    Browse the repository at this point in the history