From: Linus Walleij <linus.walleij@linaro.org>
To: Arnd Bergmann <arnd@arndb.de>
Cc: Dave Airlie <airlied@gmail.com>,
Daniel Vetter <daniel.vetter@ffwll.ch>, Greg KH <greg@kroah.com>,
Leon Romanovsky <leon@kernel.org>,
Laurent Pinchart <laurent.pinchart@ideasonboard.com>,
Thomas Gleixner <tglx@linutronix.de>,
Josh Triplett <josh@joshtriplett.org>,
Mauro Carvalho Chehab <mchehab@kernel.org>,
Jonathan Corbet <corbet@lwn.net>,
ksummit@lists.linux.dev, dev@tvm.apache.org
Subject: Re: [MAINTAINER SUMMIT] User-space requirements for accelerator drivers
Date: Tue, 14 Sep 2021 15:07:50 +0200 [thread overview]
Message-ID: <CACRpkdZ_xwCepDXx+2nywzmBqq0ey_5kC9e06m4+ef9k3JUHeQ@mail.gmail.com> (raw)
In-Reply-To: <CAK8P3a2Fo=O3tM2vrFzbifc9rkmz9b3homwitTg5WNJrpDcLxA@mail.gmail.com>
On Mon, Sep 13, 2021 at 3:20 PM Arnd Bergmann <arnd@arndb.de> wrote:
> One straightforward hardware independent low-level API would
> be the traditional BLAS GEMM call[1] for matrix multiplication
> and its variants (integer, float, bfloat16, ...).
What this (and subsequent posts from Dave and Daniel) show, is
that the general pattern is that what we are accelerating is no longer
the specialized use cases of linear algebra such as 3D "shaders"
or whatever inference linear algebra NPUs are doing, which
appear to include regression, bayesian stuff, gaussian quadrature...
name it.
What we are talking about here is acceleration, using an efficient
data path, of numerical analysis, using tailored hardware.
I'm not even sure we are limited to linear algebra anymore.
Is this what is happening, and should we be thinking numerical
analysis accelerators and their different shapes and sizes
rather than usecase-foo-accelerators, so we don't end up with
this situation again the next time applied math comes knocking
on the door with their next usecase?
Yours,
Linus Walleij
next prev parent reply other threads:[~2021-09-14 13:08 UTC|newest]
Thread overview: 77+ messages / expand[flat|nested] mbox.gz Atom feed top
2021-09-10 21:00 [MAINTAINER SUMMIT] User-space requirements for accelerator drivers Jonathan Corbet
2021-09-10 21:32 ` Josh Triplett
2021-09-13 13:50 ` Christian Brauner
2021-09-13 13:57 ` Daniel Vetter
2021-09-14 2:07 ` Laurent Pinchart
2021-09-14 14:40 ` Jani Nikula
2021-09-14 14:45 ` Geert Uytterhoeven
2021-09-14 14:59 ` Jani Nikula
2021-09-14 15:10 ` Geert Uytterhoeven
2021-09-10 21:51 ` James Bottomley
2021-09-10 21:59 ` Alexandre Belloni
2021-09-10 22:35 ` James Bottomley
2021-09-11 14:51 ` Jonathan Corbet
2021-09-11 15:24 ` James Bottomley
2021-09-11 21:52 ` Laurent Pinchart
2021-09-14 13:22 ` Johannes Berg
2021-09-11 0:08 ` Laurent Pinchart
2021-09-10 22:52 ` Mauro Carvalho Chehab
2021-09-10 23:45 ` Josh Triplett
2021-09-10 23:48 ` Dave Hansen
2021-09-11 0:13 ` Laurent Pinchart
2021-09-10 23:55 ` Thomas Gleixner
2021-09-11 0:20 ` Laurent Pinchart
2021-09-11 14:20 ` Steven Rostedt
2021-09-11 22:08 ` Laurent Pinchart
2021-09-11 22:42 ` Steven Rostedt
2021-09-11 23:10 ` Laurent Pinchart
2021-09-13 11:10 ` Mark Brown
2021-09-11 22:51 ` Mauro Carvalho Chehab
2021-09-11 23:22 ` Mauro Carvalho Chehab
2021-09-11 10:31 ` Leon Romanovsky
2021-09-11 11:41 ` Laurent Pinchart
2021-09-11 12:04 ` Leon Romanovsky
2021-09-11 22:04 ` Laurent Pinchart
2021-09-12 4:27 ` Leon Romanovsky
2021-09-12 7:26 ` Greg KH
2021-09-12 8:29 ` Leon Romanovsky
2021-09-12 13:25 ` Greg KH
2021-09-12 14:15 ` Leon Romanovsky
2021-09-12 14:34 ` Greg KH
2021-09-12 16:41 ` Laurent Pinchart
2021-09-12 20:35 ` Dave Airlie
2021-09-12 20:41 ` Dave Airlie
2021-09-12 20:49 ` Daniel Vetter
2021-09-12 21:12 ` Dave Airlie
2021-09-12 22:51 ` Linus Walleij
2021-09-12 23:15 ` Dave Airlie
2021-09-13 13:20 ` Arnd Bergmann
2021-09-13 13:54 ` Daniel Vetter
2021-09-13 22:04 ` Arnd Bergmann
2021-09-13 23:33 ` Dave Airlie
2021-09-14 9:08 ` Arnd Bergmann
2021-09-14 9:23 ` Daniel Vetter
2021-09-14 10:47 ` Laurent Pinchart
2021-09-14 12:58 ` Arnd Bergmann
2021-09-14 19:45 ` Daniel Vetter
2021-09-14 15:43 ` Luck, Tony
2021-09-13 14:52 ` James Bottomley
2021-09-14 13:07 ` Linus Walleij [this message]
2021-09-13 14:03 ` Mark Brown
2021-09-12 15:55 ` Laurent Pinchart
2021-09-12 16:43 ` James Bottomley
2021-09-12 16:58 ` Laurent Pinchart
2021-09-12 17:08 ` James Bottomley
2021-09-12 19:52 ` Dave Airlie
2021-09-12 7:46 ` Mauro Carvalho Chehab
2021-09-12 8:00 ` Leon Romanovsky
2021-09-12 14:53 ` Laurent Pinchart
2021-09-12 15:41 ` Mauro Carvalho Chehab
2021-09-10 23:46 ` Laurent Pinchart
2021-09-11 0:38 ` Mauro Carvalho Chehab
2021-09-11 9:27 ` Laurent Pinchart
2021-09-11 22:33 ` Mauro Carvalho Chehab
2021-09-13 12:04 ` Mark Brown
2021-09-12 19:13 ` Dave Airlie
2021-09-12 19:48 ` Laurent Pinchart
2021-09-13 2:26 ` Dave Airlie
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=CACRpkdZ_xwCepDXx+2nywzmBqq0ey_5kC9e06m4+ef9k3JUHeQ@mail.gmail.com \
--to=linus.walleij@linaro.org \
--cc=airlied@gmail.com \
--cc=arnd@arndb.de \
--cc=corbet@lwn.net \
--cc=daniel.vetter@ffwll.ch \
--cc=dev@tvm.apache.org \
--cc=greg@kroah.com \
--cc=josh@joshtriplett.org \
--cc=ksummit@lists.linux.dev \
--cc=laurent.pinchart@ideasonboard.com \
--cc=leon@kernel.org \
--cc=mchehab@kernel.org \
--cc=tglx@linutronix.de \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).