Re: Request for Sponsorship: VeryFastTree - Parallelized and Optimized Version of FastTree
Hi César,
please also check the very detailed answer of Nilesh.
Am Wed, Jul 05, 2023 at 07:52:07PM +0000 schrieb César Pomar:
>
> More than hardware optimizations, I mean that it may not make sense to
> increase the
> package size with 5 binaries
Also your choice. The binary itself is not extraordinary big in size so
providing 5 of them would be not too expensive. However, since on one
hand we need to respect baseline and on the other hand we want to
support "the fastest" architecture we can usually expect in nearly all
relevant cases its your choice what you want to provide "inbetween".
> when SSE (the minimum required by Debian) and
> AVX2 (more common in current hardware) could be sufficient. AVX512, in the
> tests conducted, doesn't have as much impact as SSE to AVX2, and outside of
> servers, it is still not as common. The intermediate extensions are the
> same. If current computers have AVX2, the previous ones don't make sense,
> and for the rest, SSE is sufficient.
>
> Regarding the performance differences, they are analyzed in the
> supplementary material
> of the VeryFastTree paper, which is open access. You can find it here:
> https://academic.oup.com/bioinformatics/article/36/17/4658/5861530
You might like to provide this as debian/README.Debian to inform those
users who are trained to check this file (which is unfortunately only
a small fraction of our users).
> Should I push it as is or should I delete some binaries?
> Approximately, the size of the binaries would be:
> AVX512: 10MB, AVX2: 5MB, AVX: 5MB, SSE4: 3MB SSE: 2.7MB = 25.7MB
> for the package seems like a lot when perhaps AVX2 and SSE with 7.7MB would
> be sufficient.
As I said, just check Nilesh's mail which is pretty verbose about the
changes that are needed.
Kind regards
Andreas.
--
http://fam-tille.de
Reply to: