#Mimalloc
1/4

Latest Slackers (Repository for Slackware current) update:

Tue, 17 Nov 2026 19:19:54 UTC

docker-compose-5.8.0-x86_64-1cf.txz: Upgraded.
libupnp-22.2.3-x86_64-1cf.txz: Upgraded.
mimalloc-3.7.0-x86_64-1cf.txz: Upgraded.
mosquitto-2.1.4-x86_64-1cf.txz: Upgraded.
November 17, 2026 at 7:31 PM
Awesome results! Haven't read all of it yet but scrolled thru the code and benchmarks and I'm quite impressed. Also cool to see you added CRAM writing! If you want to take it further, what I'd do is use mimalloc (faster than system alloc for me), and clone seqair's property tests :)
November 14, 2026 at 2:09 PM
Just tried mimalloc, and it didn't speed up this particular workload, but good callout!

The last thing that seems to have made a big difference is pre-reserving a few index buffers and using an unsafe append that skips per-push cap checks.

gist.github.com/sstadick/e6e...
Fast pileups: the Perbase paper comparison, Rust pipeline optimizations, Seqair, and Goose — measured results and source history
Fast pileups: the Perbase paper comparison, Rust pipeline optimizations, Seqair, and Goose — measured results and source history - 00-perbase-rust-goose-benchmarks.md
gist.github.com
November 14, 2026 at 4:24 PM
Correction: this wasn't the malloc() mutex, this was MSVC STL checked containers/iterators using application global mutex.
Using mimalloc (as I recommended in my talk) didn't help at all.
It's just that _ITERATOR_DEBUG_LEVEL=2 is designed for computers from another century.
November 8, 2026 at 2:22 PM
2/5

cpuinfo-e9320ce-x86_64-1cf.txz: Upgraded.
dart-sass-bin-1.107.1-x86_64-1cf.txz: Upgraded.
h2o-94dd3be-x86_64-1cf.txz: Upgraded.
kismet-2026_10_R1-x86_64-2cf.txz: Rebuilt.
libwebsockets-5.1.0-x86_64-1cf.txz: Upgraded.
mimalloc-3.6.1-x86_64-1cf.txz: Upgraded.
October 31, 2026 at 2:31 PM
Wait, a Python build is complaining about this? PYTHON?
October 28, 2026 at 9:19 PM
frustrated that nix (>=2.35) has moved to mimalloc for memory allocations. which causes it to crash out from almost every operation under grapheneos's hardened_malloc.
this also impacts lix (>=2.96) and while lix has a helpful override of `useMimalloc` it seems to still crash for other reasons...😫
October 28, 2026 at 1:30 AM
2/6

fzssh-1.5.0-x86_64-1cf.txz: Upgraded.
github-cli-2.103.2-x86_64-1cf.txz: Upgraded.
libfilezilla-0.58.0-x86_64-1cf.txz: Upgraded.
mimalloc-3.6.0-x86_64-1cf.txz: Upgraded.
mongosh-bin-2.14.0-x86_64-1cf.txz: Upgraded.
netdata-2.12.2-x86_64-1cf.txz: Upgraded.
October 24, 2026 at 5:01 PM
Not too surprising, musl's allocator tends to be slower for heavy multithreaded workloads. Swapping in mimalloc or jemalloc often closes the gap.
October 16, 2026 at 2:19 AM
Great read on memory allocation/malloc in games and benchmarks with some drop in memory allocation replacements like tlsf, mimalloc etc #gamedev
Benchmarking Malloc with Doom 3 - ForrestTheWoods
Benchmarking worst case malloc performance with Doom 3
www.forrestthewoods.com
October 5, 2026 at 5:06 AM
4/6

libupnp-22.1.8-x86_64-1cf.txz: Upgraded.
libvirt-12.8.0-x86_64-1cf.txz: Upgraded.
libyuv-eee4597-x86_64-1cf.txz: Upgraded.
luv-1.53.0_0-x86_64-1cf.txz: Upgraded.
mimalloc-3.5.4-x86_64-1cf.txz: Upgraded.
netdata-2.12.0-x86_64-1cf.txz: Upgraded.
October 1, 2026 at 4:31 PM
guy holding a presentation at work keeps pronouncing mimalloc like meme-alloc
September 29, 2026 at 7:09 AM
I have advanced this to GCC 13.5+binutils 2.41+LLVM 23. funny bit, just linking mimalloc into the build I improved performance by 15% because musl "ping pongs" back to the OS by giving it memory back too eagerly. 300k mmap syscalls to compile 1 (one) .c file! musl's allocator is bad at any speed
Last night I had 6-astra-max plan out and begin working on a port of the full source nix/guix bootstrap to raw Buck2 BUILD files and it apparently completed this morning! will be interesting to evaluate how it did here
September 25, 2026 at 3:08 PM
just realized the ai deleted the mimalloc global allocator override 4 months ago and my jit has been using fucking glibc malloc ever since 👍
September 24, 2026 at 7:03 AM
4/6

lyx-2.5.3-x86_64-1cf.txz: Upgraded.
mimalloc-3.5.3-x86_64-1cf.txz: Upgraded.
modulejail-1.6.2-x86_64-1cf.txz: Upgraded.
mongo-c-driver-2.5.4-x86_64-1cf.txz: Upgraded.
netdata-2.11.1-x86_64-1cf.txz: Upgraded.
node-bash-language-server-5.8.0-x86_64-1cf.txz: Upgraded.
September 18, 2026 at 1:31 PM
2/4

mimalloc-3.5.2-x86_64-1cf.txz: Upgraded.
node-nan-2.29.0-x86_64-1cf.txz: Upgraded.
opengrep-bin-2.0.0.alpha.3-x86_64-1cf.txz: Added.
php-antimalware-scanner-0.21.10-noarch-1cf.txz: Upgraded.
phpstan-bin-2.2.14-x86_64-1cf.txz: Upgraded.
September 14, 2026 at 1:31 PM
Memory allocators in 2026: mimalloc vs jemalloc vs tcmalloc. Benchmark against your workload to cut tail latency and memory overhead in production.
The State of Allocators in 2026 - 6 Months Later
Four Signals — The Wire
www.foursignals.dev
September 10, 2026 at 3:00 AM
Yay! mjp41 merged smalloc into mimalloc-bench! github.com/daanx/mimall... Now it is easy to compare smalloc to mimalloc, snmalloc, rpmalloc, jemalloc, and the system allocators on a variety of workloads. 🤗
rpmalloc build fails on Ubuntu · Issue #256 · daanx/mimalloc-bench
$ ./build-bench-env.sh je sn mi2 rp packages bench lean-mathlib ... FAILED: build/ninja/linux/debug/x86-64/test-7e8c09f/main-override-7531094.o clang++ -MMD -MT build/ninja/linux/debug/x86-64/test-...
github.com
September 3, 2026 at 11:40 PM
A major advantage that bench-allocators has over mimalloc-bench is that it produces pretty graphs: github.com/zooko/bench-... (Thanks to a suggestion from Sean Bowe, and thanks to help from Claude Fable 5.)
September 3, 2026 at 11:45 PM
If you love this kind of thing, see also my bench-allocators which is a lot like mimalloc-bench but Rust-oriented instead of C-oriented, and with a completely different set of workloads: github.com/zooko/bench-...
GitHub - zooko/bench-allocators: run various benchmarks of memory allocators
run various benchmarks of memory allocators. Contribute to zooko/bench-allocators development by creating an account on GitHub.
github.com
September 3, 2026 at 11:41 PM
3/6

libheif-1.23.3-x86_64-1cf.txz: Upgraded.
libvhdi-20260901-x86_64-1cf.txz: Upgraded.
libvirt-12.7.0-x86_64-1cf.txz: Upgraded.
libvslvm-20260901-x86_64-1cf.txz: Upgraded.
libyuv-f51ad56-x86_64-1cf.txz: Upgraded.
mimalloc-3.5.1-x86_64-1cf.txz: Upgraded.
September 2, 2026 at 8:31 PM
day 3 — excited for Daan Leijen’s keynote!

Efficient strong functional programming with effects and compiler guided reference counting (and mimalloc)
August 27, 2026 at 1:08 PM
update from the slop mines: it crashes because claude recommended the default allocator to be replaced with mimalloc, and this causes some kind of conflict during startup.

i'm not even sure there's any benefit from replacing the allocator; a logic analyzer doesn't exactly fit the bill for "lots […]
Original post on social.treehouse.systems
social.treehouse.systems
August 27, 2026 at 5:55 PM
Determinate Nix 3.22.2 is now out! 🎉

That means a new upstream base (2.35.2), 5-12% faster evaluation with mimalloc, HTTP/3 binary cache fetching, faster `nix copy` over slow links, GC roots for `nix eval`, Replit auth for FlakeHub, and more ❄️ 🚀 🔗

determinate.systems/blog/changel...
Changelog: a new upstream base, GC roots for nix eval, and a faster nix copy
Determinate Nix 3.22.2 moves onto upstream Nix 2.35, gives nix eval a --drv-link flag that keeps the garbage collector away from freshly instantiated derivations, and makes nix copy fast again…
determinate.systems
August 26, 2026 at 4:17 PM