From 2c7670bef9abc472b15b7166e4a5587647c35a4c Mon Sep 17 00:00:00 2001
From: not-matthias <matthias@codspeed.io>
Date: Fri, 9 Oct 2026 18:28:28 +0200
Subject: [PATCH 1/2] perf: use mimalloc as the runner's global allocator

The release binary targets musl, whose allocator is slow on many small
allocations and serializes threads. Post-processing a memory run decodes
the whole memtrack artifact with one or more allocations per event, which
on a 62M-event artifact took 8-19 minutes in CI with the musl build against
1-2 minutes with glibc for the same code.

Use mimalloc, as memtrack already does, so the runner no longer depends on
the libc allocator.
---
 Cargo.lock                 | 1 +
 Cargo.toml                 | 2 ++
 crates/memtrack/Cargo.toml | 2 +-
 src/main.rs                | 4 ++++
 4 files changed, 8 insertions(+), 1 deletion(-)

diff --git a/Cargo.lock b/Cargo.lock
index d7bdaa3cd..f0f9e20ca 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -730,6 +730,7 @@ dependencies = [
  "log",
  "memmap2",
  "memtrack",
+ "mimalloc",
  "nestify",
  "nix 0.31.3",
  "object",
diff --git a/Cargo.toml b/Cargo.toml
index f85f895d3..c8df261fe 100644
--- a/Cargo.toml
+++ b/Cargo.toml
@@ -60,6 +60,7 @@ linux-perf-data = { git = "https://github.com/mstange/linux-perf-data.git", rev
 ] } # unreleased main as of 2026-03-19
 debugid = "0.8.0"
 memmap2 = "0.9.10"
+mimalloc = { workspace = true }
 nix = { version = "0.31.3", features = ["fs", "time", "user"] }
 futures = "0.3.32"
 runner-shared = { path = "crates/runner-shared" }
@@ -113,6 +114,7 @@ libc = "0.2"
 log = "0.4.28"
 serde_json = "1.0"
 serde = { version = "1.0.228", features = ["derive"] }
+mimalloc = "0.1"
 ipc-channel = "0.20"
 itertools = "0.14.0"
 rayon = "1.12"
diff --git a/crates/memtrack/Cargo.toml b/crates/memtrack/Cargo.toml
index 20bf77fee..6e9279b1f 100644
--- a/crates/memtrack/Cargo.toml
+++ b/crates/memtrack/Cargo.toml
@@ -40,7 +40,7 @@ byteorder = "1.5"
 rayon = "1.12"
 parking_lot = "0.12"
 typed-builder = "0.23.2"
-mimalloc = { version = "0.1", optional = true }
+mimalloc = { workspace = true, optional = true }
 
 [build-dependencies]
 libbpf-cargo = { version = "0.26", optional = true }
diff --git a/src/main.rs b/src/main.rs
index 518b8d3ef..446d01bb6 100644
--- a/src/main.rs
+++ b/src/main.rs
@@ -1,3 +1,7 @@
+// musl's allocator is slow on the many small allocations made when decoding memtrack artifacts.
+#[global_allocator]
+static GLOBAL: mimalloc::MiMalloc = mimalloc::MiMalloc;
+
 use codspeed_runner::{clean_logger, cli};
 use console::style;
 use log::log_enabled;

From 5ee7e5883fe6b2929a32a4960ed9e9392ea00538 Mon Sep 17 00:00:00 2001
From: not-matthias <matthias@codspeed.io>
Date: Mon, 12 Oct 2026 18:18:24 +0200
Subject: [PATCH 2/2] test(runner-shared): run the memtrack benches on mimalloc

---
 Cargo.lock                                      | 1 +
 crates/runner-shared/Cargo.toml                 | 1 +
 crates/runner-shared/benches/memtrack_reader.rs | 3 +++
 crates/runner-shared/benches/memtrack_writer.rs | 3 ++-
 4 files changed, 7 insertions(+), 1 deletion(-)

diff --git a/Cargo.lock b/Cargo.lock
index f0f9e20ca..c44350fe5 100644
--- a/Cargo.lock
+++ b/Cargo.lock
@@ -3661,6 +3661,7 @@ dependencies = [
  "linux-perf-event-reader 0.10.2 (registry+https://github.com/rust-lang/crates.io-index)",
  "log",
  "memmap2",
+ "mimalloc",
  "rand 0.8.6",
  "rayon",
  "rmp",
diff --git a/crates/runner-shared/Cargo.toml b/crates/runner-shared/Cargo.toml
index 594d9cab8..83d3d16ab 100644
--- a/crates/runner-shared/Cargo.toml
+++ b/crates/runner-shared/Cargo.toml
@@ -28,6 +28,7 @@ zstd = "0.13.3"
 [dev-dependencies]
 divan = { version = "4.2.0", package = "codspeed-divan-compat" }
 memmap2 = "0.9.10"
+mimalloc = { workspace = true }
 rand = "0.8"
 tempfile = { workspace = true }
 
diff --git a/crates/runner-shared/benches/memtrack_reader.rs b/crates/runner-shared/benches/memtrack_reader.rs
index 2215eb643..87a1d0e6e 100644
--- a/crates/runner-shared/benches/memtrack_reader.rs
+++ b/crates/runner-shared/benches/memtrack_reader.rs
@@ -15,6 +15,9 @@ use runner_shared::measurement_mode::MeasurementMode;
 use runner_shared::runtime_env::RUNNER_MODE_ENV;
 use std::io::Write;
 
+#[global_allocator]
+static GLOBAL: mimalloc::MiMalloc = mimalloc::MiMalloc;
+
 fn main() {
     divan::main();
 }
diff --git a/crates/runner-shared/benches/memtrack_writer.rs b/crates/runner-shared/benches/memtrack_writer.rs
index 580424a1a..2e5eec1fc 100644
--- a/crates/runner-shared/benches/memtrack_writer.rs
+++ b/crates/runner-shared/benches/memtrack_writer.rs
@@ -16,7 +16,8 @@ use runner_shared::artifacts::{
 /// there anyway.
 #[cfg(not(codspeed))]
 #[global_allocator]
-static ALLOC: divan::AllocProfiler = divan::AllocProfiler::system();
+static ALLOC: divan::AllocProfiler<mimalloc::MiMalloc> =
+    divan::AllocProfiler::new(mimalloc::MiMalloc);
 
 fn main() {
     divan::main();