mikros/rust
1
0
Fork 0
Code Issues Pull Requests Packages Projects Releases Wiki Activity

Fix _mm_sad_epu8 implementation

Browse Source 
Fixes rust-lang/rustc_codegen_cranelift#1426
This commit is contained in:
bjorn3 2023-11-21 17:00:43 +01:00
parent cd7218aa9f
commit a5fcfad890
1 changed files with 3 additions and 1 deletions
Show Stats Download Patch File Download Diff File Expand all files Collapse all files

4
src/intrinsics/llvm_x86.rs
View File

@ -364,9 +364,11 @@ fn select4(
for out_lane_idx in 0..lane_count / 8 {
let mut lane_diff_acc = fx.bcx.ins().iconst(types::I64, 0);
for lane_idx in out_lane_idx * 8..out_lane_idx * 8 + 1 {
for lane_idx in out_lane_idx * 8..out_lane_idx * 8 + 8 {
let a_lane = a.value_lane(fx, lane_idx).load_scalar(fx);
let a_lane = fx.bcx.ins().uextend(types::I16, a_lane);
let b_lane = b.value_lane(fx, lane_idx).load_scalar(fx);
let b_lane = fx.bcx.ins().uextend(types::I16, b_lane);
let lane_diff = fx.bcx.ins().isub(a_lane, b_lane);
let abs_lane_diff = fx.bcx.ins().iabs(lane_diff);