@@ -983,6 +983,1299 @@ define i64 @mul81(i64 %a) {
ret i64 %c
}
+
+define i64 @mul153(i64 %a) {
+; RV64I-LABEL: mul153:
vacma wrote:
Thanks for your feedback,
tests show that 3 shXadds is _a little_ slower than mul, but for some reason
GCC uses 3 shXad
@@ -983,6 +983,1299 @@ define i64 @mul81(i64 %a) {
ret i64 %c
}
+
+define i64 @mul153(i64 %a) {
+; RV64I-LABEL: mul153:
topperc wrote:
llvm-mca doesn't think this is an improvement for sifive-x280
https://godbolt.org/z/6ahP11xrq
https://github.com/llvm/ll
https://github.com/vacma updated
https://github.com/llvm/llvm-project/pull/68144
>From 64e5deade92206cd04f86e0665a482ea78d0f3bf Mon Sep 17 00:00:00 2001
From: Alexandr
Date: Sun, 1 Oct 2023 19:57:01 +0300
Subject: [PATCH 1/3] [RISCV][Zba] Optimize mul with SH*ADD
Optimized multiplication b
