PathScale EKO Path 5 not for FreeBSD anymore?

[O. Hartmann]

A while ago - approximately three years from now, i was looking for a
GPGPU capable solution for usage on FreeBSD and I stepped into the
compilers from PathScale which are supposed to handle OpenACC (like
OpenMP #pragma omp, but in this case #pragma openacc instead).

Well, there was hope since PathScale obviously had a FreeBSD commercial
solution. It was in BETA stage that time, I applied for getting a
testing copy, but never got an answer, even having had contact to
Christopher Bergström, CTO at PathScale.

Looking today at Phoronix,

(http://www.phoronix.com/scan.php?page=article&item=pathscale_ekopath_5beta&num=1),

I read this benchmarking and followed the links which say that PathScale
opensourced their compiler suite a while ago

(http://www.phoronix.com/scan.php?page=article&item=pathscale_ekopath4_open&num=1)

I was curious and looked at PathScales website where I found three years
ago also FreeBSD mentioned as a supported platform, but see foryourself,
what supported platforms today mentioned there:

http://www.pathscale.com/ekopath-compiler-suite

Well, at the end, this means there is simply no binary or package that
could be installed easily for scientists interested in that compiler.

I do not know whether there are motivations to produce a FreeBSD 10/9
compatible package from the newly emitted PathScale EKO Patch 5 Beta
compiler sources, which are available at github:

https://github.com/path64/compiler

Well, the official website of PathScale doens't mention FreeBSD anymore
and this must have a reason why they droped support or any intention to
support that OS. From the perspective of a "user", I'm the lonely
"idiot" within kilometres using FreeBSD for my day-to-day scientifice
work and sacrifice myself not having GPGPU capabilities. Something is
really going into the wrong direction here.

I'm very interested in the reasoning why PathScale droped FreeBSD and I
guess it would be nice to reveal the reasons.

Am I blind or is this again another erosion process of an operating
system usefull even for scientific purposes?

Well, I'm aware that my posting triggers again a lot of emotional
discussions (I guess), since it did in the past. I try to evaluate the
situation from the perspective of a "user", not someone who needs to be
a Os engineer to use an OS. It is a kind of deep running frustration to
see how the next great compiler suite for scientific purpose is simply
vanishing - as it did with the NAG compilers which were offered for
freeBSD as well as other OSs at the end of 1990s. Now there is no
offering anymore.

Regards,
Oliver

-------------- next part --------------
A non-text attachment was scrubbed...
Name: signature.asc
Type: application/pgp-signature
Size: 488 bytes
Desc: OpenPGP digital signature
URL: <http://lists.freebsd.org/pipermail/freebsd-performance/attachments/20130220/e33a0510/attachment.sig>