bjoernager/rust
bjoernager/rust
1
Fork 0
Code Activity

Remove my scalar_copy_backend_type optimization attempt

Browse source 
I added this back in 111999, but I no longer think it's a good idea
- It had to get scaled back to only power-of-two things to not break a bunch of targets
- LLVM seems to be getting better at memcpy removal anyway
- Introducing vector instructions has seemed to sometimes (115515) make autovectorization worse

So this removes it from the codegen crates entirely, and instead just tries to use <https://doc.rust-lang.org/nightly/nightly-rustc/rustc_codegen_ssa/traits/builder/trait.BuilderMethods.html#method.typed_place_copy> instead of direct `memcpy` so things will still use load/store for immediates.
This commit is contained in:
Scott McMurray 2024-03-29 00:00:24 -07:00
parent ff24ef91fc
commit b5376ba601
10 changed files with 60 additions and 154 deletions
Download patch file Download diff file Expand all files Collapse all files

8
tests/codegen/array-optimized.rs
View file

@ -16,8 +16,8 @@ pub fn array_copy_1_element(a: &[u8; 1], p: &mut [u8; 1]) {
#[no_mangle]
pub fn array_copy_2_elements(a: &[u8; 2], p: &mut [u8; 2]) {
// CHECK-NOT: alloca
// CHECK: %[[TEMP:.+]] = load <2 x i8>, ptr %a, align 1
// CHECK: store <2 x i8> %[[TEMP]], ptr %p, align 1
// CHECK: %[[TEMP:.+]] = load i16, ptr %a, align 1
// CHECK: store i16 %[[TEMP]], ptr %p, align 1
// CHECK: ret
*p = *a;
}
@ -26,8 +26,8 @@ pub fn array_copy_2_elements(a: &[u8; 2], p: &mut [u8; 2]) {
#[no_mangle]
pub fn array_copy_4_elements(a: &[u8; 4], p: &mut [u8; 4]) {
// CHECK-NOT: alloca
// CHECK: %[[TEMP:.+]] = load <4 x i8>, ptr %a, align 1
// CHECK: store <4 x i8> %[[TEMP]], ptr %p, align 1
// CHECK: %[[TEMP:.+]] = load i32, ptr %a, align 1
// CHECK: store i32 %[[TEMP]], ptr %p, align 1
// CHECK: ret
*p = *a;
}