Fortran - Lahey/Fujitsu Fortran v7.1 released

Author

Lahey/Fujitsu Fortran v7.1 released

John Swenson

2004-03-27, 12:17 am

March 17, 2004 - Incline Village, Nevada, USA
Lahey Computer Systems, Inc. today announced the release of
Lahey/Fujitsu Fortran v7.1 for Windows. This full-featured Fortran
95/90/77 language system includes enhancements for performance and
ease of use. The LF Fortran v7.1 language system is available in four
configurations:

Enterprise (Fortran for .NET / Fortran 95 / Visual Studio)
Professional (Fortran 95 / Visual Studio)
Express (Fortran 95)
Academic (Fortran for .NET)

New Fortran 95 (Win32) Features of LF Fortran v7.1 include the
following:
-Pentium(R) 4 and Xeon(TM) optimizations
-SSE2 (Streaming SIMD Extensions 2) instructions
-New WiSK Project Template (Enterprise and Professional only)
-BLAS for Pentium 4 (Enterprise and Professional only)
-LAPACK for Pentium 4 (Enterprise and Professional only)
-Optimized SSL2 math library (Enterprise and Professional only)
-Enhanced Automake (Enterprise and Professional only)

According to Bill Lassaline, president of Lahey, "We tested v7.1 Win32
optimizations on a 2.4GHz P4 with 512 MB of PC2100 RAM, running
Windows XP, using Polyhedron's (www.polyhedron.com) Fortran
benchmarks. Specifying the new switches -tp4, -sse2, -zfm, and
-inline the 90 benchmarks ran an average of 15.4% faster and the 77
benchmarks 4.4% faster than they did when built with LF Fortran v7.0."

Lahey Computer Systems, Inc.
Now in its 37th year, Lahey Computer Systems, Inc. develops, markets,
sells and supports high-quality Fortran-language programming tools.
For additional information, please visit www.lahey.com.

Contact:
John Swenson
Lahey Computer Systems, Inc.
865 Tahoe Blvd., P.O. Box 6091
Incline Village, Nevada, U.S.A. 89450
Telephone: 800-548-4778 or 775-831-2500
Fax: 775-831-8123
Email: sales@lahey.com
Web: www.lahey.com

Rich Townsend

2004-03-27, 12:17 am

John Swenson wrote:
> March 17, 2004 - Incline Village, Nevada, USA
> Lahey Computer Systems, Inc. today announced the release of
> Lahey/Fujitsu Fortran v7.1 for Windows. This full-featured Fortran
> 95/90/77 language system includes enhancements for performance and
> ease of use. The LF Fortran v7.1 language system is available in four
> configurations:
>
> <snip>

And Lahey continue to neglect their Linux product. There hasn't been a
compiler update for almost three-quaters of a year -- in spite of the
fact that I've submitted a number of bugs, and I'm sure that I'm not the
only one finding problems.

Hell, Lahey doesn't even support the TR yet. As far as I'm concerned,
this leaves their compiler dead in the water, as compared to the
competition.

I really wish Lahey would get their act together, since with a bit of
polish, they could have a winning product. At the moment, they are
losing badly to Intel; the latter have a product which wins on
performance, price and support, and unless Lahey raise their game, I see
them being pushed out of the Linux market, and possibly those for all
x86 platforms.

Rich T

Scott Robert Ladd

2004-03-27, 12:17 am

On Mon, 22 Mar 2004 14:27:33 -0800, John Swenson wrote:
> March 17, 2004 - Incline Village, Nevada, USA
> Lahey Computer Systems, Inc. today announced the release of
> Lahey/Fujitsu Fortran v7.1 for Windows. This full-featured Fortran
> 95/90/77 language system includes enhancements for performance and
> ease of use. The LF Fortran v7.1 language system is available in four
> configurations:
>
> Enterprise (Fortran for .NET / Fortran 95 / Visual Studio)
> Professional (Fortran 95 / Visual Studio)
> Express (Fortran 95)
> Academic (Fortran for .NET)

No Linux version? Or AMD84?

That seems a tad short-sighted, doesn't it? Or are Linux and AMD64
versions in the offing?

--
Scott Robert Ladd
Coyote Gulch Productions (http://www.coyotegulch.com)
Software Invention for High-Performance Computing

B52B@pl

2004-03-27, 12:17 am

Scott Robert Ladd wrote:
> On Mon, 22 Mar 2004 14:27:33 -0800, John Swenson wrote:
>
>
>
> No Linux version? Or AMD84?

They have Linux version too. The newest 7.1 version is
windows "brother" for their 6.2 (for Linux) release.

B52

B52B@pl

2004-03-27, 12:17 am

Rich Townsend wrote:

> And Lahey continue to neglect their Linux product. There hasn't been a
> compiler update for almost three-quaters of a year -- in spite of the
> fact that I've submitted a number of bugs, and I'm sure that I'm not the
> only one finding problems.
>
> Hell, Lahey doesn't even support the TR yet. As far as I'm concerned,
> this leaves their compiler dead in the water, as compared to the
> competition.

Apparently they have problems with TR even though they support
allocatable components of derived types. I think they are aware at least.

>
> I really wish Lahey would get their act together, since with a bit of
> polish, they could have a winning product. At the moment, they are
> losing badly to Intel; the latter have a product which wins on
> performance, price and support, and unless Lahey raise their game, I see
> them being pushed out of the Linux market, and possibly those for all
> x86 platforms.

Hmm... I think that intel's compiler is not "mature" in debugging at
least. Lahey's compiler is very good (www.polyhedron.com) but I must
admit that support for the linux should be better.

B52

>
> Rich T
>

Rich Townsend

2004-03-27, 12:17 am

B52B@pl wrote:
> Rich Townsend wrote:
>
>
>
> Apparently they have problems with TR even though they support
> allocatable components of derived types. I think they are aware at least.

I disagree. The Lahey compiler "permits" allocatable components of
derived types; however, Lahey as a company do not support this
"feature", and any bugs regarding TR functionality are filed as product
suggestions. If anyone can prove otherwise, I'd be willing to listen.

I found out the bug-as-product-suggestion weirdness myself when I
submittted a TR bug last August. Because of Lahey's attitue, I doubt
whether the bug will ever get fixed.

> Hmm... I think that intel's compiler is not "mature" in debugging at
> least. Lahey's compiler is very good (www.polyhedron.com) but I must
> admit that support for the linux should be better.

My point is that at least the Intel compiler is evolving and improving.
Even though I have the free, unsupported version of the Intel compiler,
the level of support I have received with it exceeds that I've received
on my *paid-for* copy of the Lahey compiler. By this, I mean that Intel
have actually fixed some of the bugs I have submitted, rather than

a) suggesting a work-around, which ends up being a terrible kludge

or

b) treating the bug as a product suggestion.

cheers,

Rich T

Herman D. Knoble

2004-03-27, 12:17 am

Rich: I don't doublt your experience. But this has not been
my experience with Lahey.

In the last year or so I've submitted 3 bugs to Lahey and they
acknowledged and fixed each one. The impact of a problem with Laney is
somewhat like the old IBM APAR system. That is something like:
Suggestion, Low Impact, Medium Impact, Severe Impact. So it's
important to include an impact statement with a but report. I'd
guess you did that.

Intel Fotran is coming along. But it is presently nowhere near the reliability
of Lahey compilers, and rightly so as its evolution time is still very short.
Considering this, the Intel compilers may even be ahead of "schedule".
It seems obvious from my experience that it's a lot easier to submit
a bug involving a Lahey compiler than one involving the Intel Compiler.
I've communicated this to Intel and I believe that they heard me.

So while milage may legitimately vary, and while this forum is a good place
to correct flaws in various vendor responses, I'd suggest a slightly warmer
approach in working with vendors. That is, the competition can be friendly
competition, and this is good for the whole Fortran Community.

Skip Knoble

On Tue, 23 Mar 2004 07:46:55 -0500, Rich Townsend <rhdt@barVOIDtol.udel.edu> wrote:

-|B52B@pl wrote:
-|> Rich Townsend wrote:
-|>
-|>> And Lahey continue to neglect their Linux product. There hasn't been a
-|>> compiler update for almost three-quaters of a year -- in spite of the
-|>> fact that I've submitted a number of bugs, and I'm sure that I'm not
-|>> the only one finding problems.
-|>>
-|>> Hell, Lahey doesn't even support the TR yet. As far as I'm concerned,
-|>> this leaves their compiler dead in the water, as compared to the
-|>> competition.
-|>
-|>
-|> Apparently they have problems with TR even though they support
-|> allocatable components of derived types. I think they are aware at least.
-|
-|I disagree. The Lahey compiler "permits" allocatable components of
-|derived types; however, Lahey as a company do not support this
-|"feature", and any bugs regarding TR functionality are filed as product
-|suggestions. If anyone can prove otherwise, I'd be willing to listen.
-|
-|I found out the bug-as-product-suggestion weirdness myself when I
-|submittted a TR bug last August. Because of Lahey's attitue, I doubt
-|whether the bug will ever get fixed.
-|
-|> Hmm... I think that intel's compiler is not "mature" in debugging at
-|> least. Lahey's compiler is very good (www.polyhedron.com) but I must
-|> admit that support for the linux should be better.
-|
-|My point is that at least the Intel compiler is evolving and improving.
-|Even though I have the free, unsupported version of the Intel compiler,
-|the level of support I have received with it exceeds that I've received
-|on my *paid-for* copy of the Lahey compiler. By this, I mean that Intel
-|have actually fixed some of the bugs I have submitted, rather than
-|
-|a) suggesting a work-around, which ends up being a terrible kludge
-|
-|or
-|
-|b) treating the bug as a product suggestion.
-|
-|cheers,
-|
-|Rich T

Herman D. (Skip) Knoble, Research Associate
(a computing professional for 38 years)
Email: SkipKnobleLESS at SPAMpsu dot edu
Web: http://www.personal.psu.edu/hdk
Penn State Information Technology Services
Academic Services and Emerging Technologies
Graduate Education and Research Services
Penn State University
214C Computer Building
University Park, PA 16802-21013
Phone:+1 814 865-0818 Fax:+1 814 863-7049

Dr Chaos

2004-03-27, 12:17 am

Rich Townsend <rhdt@barVOIDtol.udel.edu> wrote:
> My point is that at least the Intel compiler is evolving and improving.
> Even though I have the free, unsupported version of the Intel compiler,
> the level of support I have received with it exceeds that I've received
> on my *paid-for* copy of the Lahey compiler. By this, I mean that Intel
> have actually fixed some of the bugs I have submitted, rather than
>
> a) suggesting a work-around, which ends up being a terrible kludge
>
> or
>
> b) treating the bug as a product suggestion.

I have to agree with this.

Intel takes bugs reports seriously and attempts to fix bugs.

On the other hand there have been many bugs.

NAG has the best 'support' I know, since I have yet to find a bug in 7
years or so. And its run-time checking is precise and reliable,
unlike Intel's. Too bad its performance loses to Intel.

Rich Townsend

2004-03-27, 12:17 am

Herman D. Knoble wrote:

<snip>

> So while milage may legitimately vary, and while this forum is a good place
> to correct flaws in various vendor responses, I'd suggest a slightly warmer
> approach in working with vendors. That is, the competition can be friendly
> competition, and this is good for the whole Fortran Community.

Points taken -- and its worth adding that, for the record, the
customer-facing people from Lahey have always been very friendly and
helpful. I'm just not that enamoured with the backroom tech support.

Its also worth mentioning that I *still* use Lahey's lf95 (6.2 Linux) as
my main compiler, since Intel's ifort isn't quite ready for prime time.
But unless Lahey continue to improve lf95, I see myself gradually moving
over to ifort.

cheers

Rich

Sponsored Links