|
Karim Belabas on Sat, 12 Oct 2024 15:08:10 +0200
|
[Date Prev] [Date Next] [Thread Prev] [Thread Next] [Date Index] [Thread Index]
|
Re: PARI with --enable-tls
|
- To: Max Alekseyev <maxale@gmail.com>, Pari Users <pari-users@pari.math.u-bordeaux.fr>
- Subject: Re: PARI with --enable-tls
- From: Karim Belabas <Karim.Belabas@math.u-bordeaux.fr>
- Date: Sat, 12 Oct 2024 15:08:06 +0200
- Delivery-date: Sat, 12 Oct 2024 15:08:10 +0200
- Dkim-signature: v=1; a=rsa-sha256; c=relaxed/simple; d=math.u-bordeaux.fr; s=2022; t=1728738488; bh=Jdu9ODyKTUx+EaKkFwxxqh4CyPgXu4GuOPE4n5o7AgE=; h=Date:From:To:References:In-Reply-To:From; b=TfP59sBOGE9F9NbXvIhk0+zoQBc7hbE5l80+NGhhcexobx+Bqvqlul41JhzVQJbjP VUl9PKHlAmVaOktkAUkGMXKq/vTyZ0sDGwIx8JtEj0dGSArlWDF2R6iwQRWudT4Nju vyVt4yjlsf+uKAEDScRNGPgAJSMoTJA64nS6ULQkxs74URYtLtkROdoLCp+XE8qDJv 32KKxA9y3K6p90ysAIpYQ+17vIHKnzUiyyZgPyqR4zOgMsemg6W2qzX9EIhnRfySQQ zDGbZI4TeJJ2ioRbkR2csUBGctQwff0vX59hFatclFnJmy9GGbRSroTom8mbwTLY/P Fj35ORhiLhs/YLSnq0XlDU1AsNzi5MZbItb2fsM6JEU2aPWSgCv3FxDq2+4I+IfYJy Hjk8rcrkdKpHKdEX4+3P8rcrgahsQp4Du8XD2x3TTmBLnFnOVTlp0qDjoNyBa40uSx C5GwMyq9BxgAek0l/kYK1UZy2LFyfMWZmuSDeJXCZFXznP6DefjiIXC6XVzSxoOnDZ 2WkbpD/2XcPelUX/GPPIOT0FWfgyyxXjKg1thRrXdboU77F3Ah+EgRFw55EAZOKb/d CIfoQjB3SeDEBMVtHPJP9XDhssl0kz0HmXhrbhhctJczLdxrPa0APUdteSITEtI7pv 55kNeqS+ybGlow++rKjxsojA=
- In-reply-to: <Zwpz2sTlW0vQ9edH@math.u-bordeaux.fr>
- Mail-followup-to: Max Alekseyev <maxale@gmail.com>, Pari Users <pari-users@pari.math.u-bordeaux.fr>
- References: <CAJkPp5O_tYaERSa0yCYLSfbVrsf=JOQ89aUc_y5SrmZa-ML-Jw@mail.gmail.com> <Zwpz2sTlW0vQ9edH@math.u-bordeaux.fr>
* Karim Belabas [2024-10-12 15:04]:
> * Max Alekseyev [2024-10-12 14:56]:
> > Hello,
> >
> > Pari can be compiled with the option:
> > --enable-tls enable thread-local stack
> >
> > What are the advantages/disadvantages of using it? (with combination of
> > --mt=pthread if that matters)
>
> It's implied by --mt=pthread.
>
> Without the latter, it builds a thread-safe version of the PARI library.
> The downside is that it becomes a little slower. For instance the
> dynamically linked gp-dyn binary ends up being about 25% slower on
> many architectures; gp-sta is usually about 5% slower.
This is explained in doc/INSTALL.dvi in the sources
(also present in $install_prefix/share/pari/)
Cheers,
K.B.
--
Pr. Karim Belabas, U. Bordeaux, Vice-président en charge du Numérique
Institut de Mathématiques de Bordeaux UMR 5251 - (+33) 05 40 00 29 77
http://www.math.u-bordeaux.fr/~kbelabas/