[llvm] [SystemZ] Remove getInliningThresholdMultiplier() override (PR #94612)

Nikita Popov via llvm-commits llvm-commits at lists.llvm.org
Tue Jul 16 02:56:15 PDT 2024

Previous message: [llvm] [X86] Fold blend(pshufb(x,m1),pshufb(y,m2)) -> blend(pshufb(x,blend(m1,m2)),pshufb(y,blend(m1,m2))) to reduce constant pool (PR #98466)
Next message: [llvm] [PAC][CodeGen][ELF][AArch64] Support signed GOT (PR #96164)
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

nikic wrote:

Is there something we can reasonably do here in time for the LLVM 19 release? For example, would reducing the multiplier to 2 rather than 1 reduce the benchmark impact? It's not the ideal end-state, but maybe a reasonable intermediate step...

https://github.com/llvm/llvm-project/pull/94612

Previous message: [llvm] [X86] Fold blend(pshufb(x,m1),pshufb(y,m2)) -> blend(pshufb(x,blend(m1,m2)),pshufb(y,blend(m1,m2))) to reduce constant pool (PR #98466)
Next message: [llvm] [PAC][CodeGen][ELF][AArch64] Support signed GOT (PR #96164)
Messages sorted by: [ date ] [ thread ] [ subject ] [ author ]

More information about the llvm-commits mailing list