blis

blis

[Efraim Flashner]

[-- Attachment #1: Type: text/plain, Size: 918 bytes --]

I admit, looking at the commit, I'm a little jealous that there wasn't
an aarch64 special version. I checked in $repo/config¹ and there's a
config folder for armv7a, armv8a and longsoon3a, among other options. It
might be worth having some of these be the default option for these
platforms, since it should overlap with our supported architectures.

On the other hand, armv8a's make_defs.mk specify a BIG.little (or is
that big.LITTLE?) configuration of A57 + A53. I know some of my boards
are A53 only so I don't know how that works.

The default armv7a looks harmless, although there are options for using
an A9 or A15 specifically.


¹ https://github.com/flame/blis/tree/0.2.2/config

-- 
Efraim Flashner   <efraim@flashner.co.il>   אפרים פלשנר
GPG key = A28B F40C 3E55 1372 662D  14F7 41AA E7DC CA3D 8351
Confidentiality cannot be guaranteed on emails sent or received unencrypted

[-- Attachment #2: signature.asc --]
[-- Type: application/pgp-signature, Size: 833 bytes --]

Re: blis

[Dave Love]

Efraim Flashner <efraim@flashner.co.il> writes:

> I admit, looking at the commit, I'm a little jealous that there wasn't
> an aarch64 special version. I checked in $repo/config¹ and there's a
> config folder for armv7a, armv8a and longsoon3a, among other options. It
> might be worth having some of these be the default option for these
> platforms, since it should overlap with our supported architectures.

Is it really known to be significantly faster than OpenBLAS on those
platforms, and has that been reported if so?  It seems unlikely for
loongsong, for instance.