public inbox for cygwin@cygwin.com
 help / color / mirror / Atom feed
From: Takashi Yano <takashi.yano@nifty.ne.jp>
To: cygwin@cygwin.com
Subject: Re: [ANNOUNCEMENT] Updated: openblas-0.3.14-1
Date: Wed, 19 May 2021 18:12:18 +0900	[thread overview]
Message-ID: <20210519181218.de23e36e1fcec21656f274a6@nifty.ne.jp> (raw)
In-Reply-To: <announce.00828a5c-2d89-cd92-fa21-a11a3e1f03b8@gmail.com>

On Wed, 31 Mar 2021 21:03:03 +0200
Marco Atzeri wrote:
> New releases 0.3.14-1 of
> 
> openblas         (source)
> libopenblas      (dinamic library)
> 
> are available in the Cygwin distribution :
> 
> CHANGES
> Last upstream release bugfix and performance improvment
> https://github.com/xianyi/OpenBLAS/releases
> 
>   OpenBLAS will now provide enough buffer space for at
>   least 50 threads by default.
> 
> DESCRIPTION
> OpenBLAS is an optimized BLAS library based on
> GotoBLAS2 1.13 BSD version and further extended.
> OpenBLAS is licensed under the 3-clause BSD license.
> 
> HOMEPAGE
> http://www.openblas.net/
> 
> CYGWIN NOTES
> 1) As BLAS it is around 5-10 times faster than Netlib reference,
>     included in the liblapack0 package.
> 
> 2) No devel package is provided as liblapack-devel already provide
>     the needed headers and import.
>     Openblas is fully compatible with Netlib BLAS.
> 
> 3) libopenblas consist of a single file
>       /usr/bin/cygblas-0.dll
>     that will precede in PATH the liblapack0
>       /usr/lib/lapack/cygblas-0.dll
>     and used instead. Removing libopenblas will restore the
>     usage of Netlib BLAS
> 
> 4) CPU architecture covered up to now
> 
>     64 bit
>      ATOM BARCELONA BOBCAT BULLDOZER CORE2 DUNNINGTON EXCAVATOR
>      HASWELL NANO NEHALEM OPTERON OPTERON_SSE3 PENRYN PILEDRIVER
>      PRESCOTT SANDYBRIDGE STEAMROLLER ZEN
> 
>     32 bit
>      ATHLON ATOM BANIAS BARCELONA BOBCAT COPPERMINE CORE2
>      DUNNINGTON KATMAI NANO NEHALEM NORTHWOOD OPTERON
>      OPTERON_SSE3 PENRYN PRESCOTT

After upgrading libopenblas from 0.3.12-2 to 0.3.14-1,
octave 5.2.0 crashes on plot command in 64 bit environment.
In 32 bit cygwin (WoW64), this does not happen. Reverting
libopenblas to 0.3.12-1 resolve the issue.

octave:1> plot([])
fatal: caught signal Illegal instruction -- stopping myself...
Illegal instruction(core dumped)

My environment:
OS: Windows 10 20H2 (64 bit)
CPU: Inte Core i7 870 (Lynnfield)

[yano@Express5800-S70 ~]$ cygcheck -c cygwin octave libopenblas
Cygwin Package Information
Package              Version        Status
cygwin               3.2.0-1        OK
libopenblas          0.3.14-1       OK
octave               5.2.0-1        OK

By starting octave-cli in gdb, it stops at vmovaps instruction
in gemm_(). IIUC, vmovaps is an AVX instruction which is not
supported by Core i7 870.

In another PC with Core i7-4790 which supports AVX, the issue
does not occur.

-- 
Takashi Yano <takashi.yano@nifty.ne.jp>

  reply	other threads:[~2021-05-19  9:12 UTC|newest]

Thread overview: 6+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2021-03-31 19:03 Marco Atzeri via Cygwin-announce
2021-05-19  9:12 ` Takashi Yano [this message]
2021-05-19 10:56   ` Marco Atzeri
2021-05-19 14:25     ` Takashi Yano
2021-05-19 16:24       ` Marco Atzeri
2021-05-19 17:21         ` Takashi Yano

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=20210519181218.de23e36e1fcec21656f274a6@nifty.ne.jp \
    --to=takashi.yano@nifty.ne.jp \
    --cc=cygwin@cygwin.com \
    /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 read-only IMAP folder(s) and NNTP newsgroup(s).