diff options
| author | Scott McMurray <scottmcm@users.noreply.github.com> | 2022-06-04 16:23:15 -0700 |
|---|---|---|
| committer | Scott McMurray <scottmcm@users.noreply.github.com> | 2022-09-19 23:24:34 -0700 |
| commit | 6dbd9a29c21db63e2c72f5e7f4f8b5ba58023875 (patch) | |
| tree | 983bdd8ac2914903d10871843668fa61cf28fdc3 /src/test/codegen | |
| parent | c773c134c949661874d76313280e76cc3dd60975 (diff) | |
| download | rust-6dbd9a29c21db63e2c72f5e7f4f8b5ba58023875.tar.gz rust-6dbd9a29c21db63e2c72f5e7f4f8b5ba58023875.zip | |
Optimize `array::IntoIter`
`.into_iter()` on arrays was slower than it needed to be (especially compared to slice iterator) since it uses `Range<usize>`, which needs to handle degenerate ranges like `10..4`. This PR adds an internal `IndexRange` type that's like `Range<usize>` but with a safety invariant that means it doesn't need to worry about those cases -- it only handles `start <= end` -- and thus can give LLVM more information to optimize better. I added one simple demonstration of the improvement as a codegen test.
Diffstat (limited to 'src/test/codegen')
| -rw-r--r-- | src/test/codegen/slice-iter-len-eq-zero.rs | 14 |
1 files changed, 14 insertions, 0 deletions
diff --git a/src/test/codegen/slice-iter-len-eq-zero.rs b/src/test/codegen/slice-iter-len-eq-zero.rs index fd19e624cdd..1124028253d 100644 --- a/src/test/codegen/slice-iter-len-eq-zero.rs +++ b/src/test/codegen/slice-iter-len-eq-zero.rs @@ -1,5 +1,6 @@ // no-system-llvm // compile-flags: -O +// ignore-debug: the debug assertions add extra comparisons #![crate_type = "lib"] type Demo = [u8; 3]; @@ -12,3 +13,16 @@ pub fn slice_iter_len_eq_zero(y: std::slice::Iter<'_, Demo>) -> bool { // CHECK: ret i1 %2 y.len() == 0 } + +// CHECK-LABEL: @array_into_iter_len_eq_zero +#[no_mangle] +pub fn array_into_iter_len_eq_zero(y: std::array::IntoIter<Demo, 123>) -> bool { + // This should be able to just check that the indexes are equal, and not + // need any subtractions or comparisons to handle `start > end`. + + // CHECK-NOT: icmp + // CHECK-NOT: sub + // CHECK: %1 = icmp eq {{i16|i32|i64}} + // CHECK: ret i1 %1 + y.len() == 0 +} |
