[PATCH v2 0/6] Improve make and make check parallelism
Sam James
sam@gentoo.org
Wed Jun 17 02:24:19 GMT 2026
Adhemerval Zanella <adhemerval.zanella@linaro.org> writes:
Thank you for doing this. I plan to provide review as well later.
> [...]
> Build results are unchanged: all built objects, archives, and shared
> objects are bit-identical to the serial build across repeated clean
> parallel builds, the installed tree layout is identical, and the
> tests.sum results are identical. Since most of the possible failure
> modes are races, the series was validated with repeated from-scratch
> builds on x86_64-linux-gnu (24 threads), aarch64-linux-gnu (80 cores).
Have you tried make --shuffle [0][1] yet? It's effective in finding some
problems [2] and also very convenient given it produces a seed upon failure.
>
> Benchmarks (cumulative; each percentage is relative to the previous
> row, the first patch only adds data and acts as the baseline):
>
> * x86_64 [1]
>
On x86_64 with Ryzen 3950X (16c/32t), I saw an improvement (first
numbers are skewed, see below):
2026-06-09T11:31:39 >>> sys-libs/glibc-2.43.9999: 2:32:02
2026-06-13T06:50:13 >>> sys-libs/glibc-2.43.9999: 51m44s
I'd also applied the workaround for the gawk issue [3] as I'd upgraded
it prior in the above period, giving:
2026-02-10T08:58:06 >>> sys-libs/glibc-2.43.9999: 1:27:00
2026-02-18T16:03:23 >>> sys-libs/glibc-2.43.9999: 2:30:43
So, your saved me 30-40m with the 3 multilibs on v1. I'll test the speed
again on v2 and with isolation of the gawk variable too later. But very
nice indeed.
[0] https://trofi.github.io/posts/238-new-make-shuffle-mode.html
[1] https://trofi.github.io/posts/249-an-update-on-make-shuffle.html
[2] https://trofi.github.io/posts/294-an-obscure-make-shuffle-bug.html
[3] https://inbox.sourceware.org/libc-alpha/lhuikb5ibey.fsf@oldenburg.str.redhat.com/
> [...]
sam
-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 418 bytes
Desc: not available
URL: <https://sourceware.org/pipermail/libc-alpha/attachments/20260617/61f5baae/attachment.sig>
More information about the Libc-alpha
mailing list