Auto merge of #3387 - RalfJung:bench, r=RalfJung

make 'invalidate' benchmark shorter This is currently by far the slowest benchmark in our suite, taking >9s, when the second slowest takes 2.7s. So let's speed this up to 2.3s, making it still the second-slowest in the benchmark suite. `@saethlin` any objections? Also, why is this called "invalidate"? It got added in #3083 but I can't figure out the point of that name even after looking at the PR.^^ There should be a comment in the benchmark explaining what it is testing.
rust-lang · Mar 18, 2024 · 850447d · 850447d
2 parents 3e6847c + b3481a4
commit 850447d
Show file tree

Hide file tree

Showing 3 changed files with 4 additions and 3 deletions.
diff --git a/bench-cargo-miri/invalidate/Cargo.lock → bench-cargo-miri/range-iteration/Cargo.lock b/bench-cargo-miri/invalidate/Cargo.lock → bench-cargo-miri/range-iteration/Cargo.lock
diff --git a/bench-cargo-miri/invalidate/Cargo.toml → bench-cargo-miri/range-iteration/Cargo.toml b/bench-cargo-miri/invalidate/Cargo.toml → bench-cargo-miri/range-iteration/Cargo.toml
@@ -1,5 +1,5 @@
 [package]
-name = "invalidate"
+name = "range-iteration"
 version = "0.1.0"
 edition = "2021"
 

diff --git a/bench-cargo-miri/invalidate/src/main.rs → bench-cargo-miri/range-iteration/src/main.rs b/bench-cargo-miri/invalidate/src/main.rs → bench-cargo-miri/range-iteration/src/main.rs
@@ -1,4 +1,5 @@
+//! This generates a lot of work for the AllocId part of the GC.
 fn main() {
     // The end of the range is just chosen to make the benchmark run for a few seconds.
-    for _ in 0..200_000 {}
+    for _ in 0..50_000 {}
 }