Lahey Support
08-15-2003, 01:19 AM
I don't know about the latest compilers, but the reason I went with La=
hey 3 years ago was that it executed the code I wrote in about half the tim=
e as the Microsoft Fortran.=0D
-David =0D
=0D
----------=0D
From: Jim Kress=0D
Sent: Wednesday, February 10, 1999 3:30 PM=0D
To: Carlstrom_Rick; [address removed]
Subject: Re: [LF] Fortran 90 vs C++ Valarrays=0D
=0D
Let's be a little careful with the 'everybody will jump ship' comments. Fo=
r=0D
the older codgers like myself, FORTRAN is the language of choice and will=
=0D
continue to be for the foreseeable future.=0D
=0D
I have worked with C++ and find it to be a nasty programming language=0D
designed (IMHO) primarily to restore computer scientists to their previousl=
y=0D
occupied 'god of the machine' position. The syntax seems to be deliberatel=
y=0D
obtuse and the vocabulary associated with the design and implementation of=
=0D
C++ systems is an arcane mishmash of gobbledygook that serves only to=0D
impress the uninformed and confuse the rest of us.=0D
=0D
I am not a fan of any language whose fundamental purpose seems to be to=0D
generate bloated code that no one else can understand or reuse - and yes, I=
=0D
know about classes and inheritance and all the rest of that nonsense. I=0D
have yet to see any C++ project implemented that actually allows ANY=0D
significant reuse of the design much less ANY code.=0D
=0D
I do agree that FORTRAN needs to keep a performance edge on other languages=
.=0D
In particular, Lahey Fortran needs to do that as well as provide some of th=
e=0D
basic amenities offered by other FORTRAN compilers (a list I sent to Bob=0D
Runyan=0D
some time ago) including integrated OpenGL support, integrated basic=0D
graphics primitives, a integrated 'Quick Window' development environment,=
=0D
etc.=0D
=0D
FORTRAN uber alles!=0D
=0D
Jim Kress=0D
=0D
=0D
-----Original Message-----=0D
From: Carlstrom_Rick [address removed]
To: [address removed] [address removed]
Date: Wednesday, February 10, 1999 1:43 PM=0D
Subject: RE: [LF] Fortran 90 vs C++ Valarrays=0D
=0D
=0D
>I think one of the keys here is the use of Powerstation 4.0. Here at my=
=0D
>company we had PS 4.0 and Lahey 3.5. A structural analysis DLL was made=
=0D
from legacy fortran 77 code first with Powerstation and then 4 months later=
=0D
>with LF90. Swapping out the 2 DLL's indicated that Lahey had a factor of =
4=0D
>speed advantage over powerstation and maintained a full trace back table i=
f=0D
>the code crashed.=0D
>=0D
>Now I cannot comment on the effects of compile options used to compile=0D
under=0D
>Powerstation and Lahey. (I do not have the original project files that=0D
>generated the DLL using powerstation)=0D
>=0D
>One Comment I have is that LF903.5 created slower executibles then=0D
>F77L3/EM-32 on mathematically intensize code.=0D
>=0D
>I will agree that if the various FORTRAN compiler venders cannot keep a=0D
>speed advantage over C++ they will looses there client base quickly.=0D
>=0D
>Rick Carlstrom=0D
>Head of Programming=0D
>Nichols Advanced Marine Enterprises.=0D
>=0D
>-----Original Message-----=0D
>From: Ching-Kuang Shene [address removed]
>Sent: Wednesday, February 10, 1999 3:16 AM=0D
>To: [address removed]
>Subject: [LF] Fortran 90 vs C++ Valarrays=0D
>=0D
>=0D
>Hi all,=0D
>=0D
>In a recent article "Comparing Fortran 90 and C++ Valarrays" in C/C++ User=
s=0D
>Journal, the authors (Shyam Bhat and B. Arun) compared MSVC++ 5.0 and=0D
>Microsoft PowerStation v 4.0 with their finite element method program on=
=0D
>a Pentium 150MHz with 48MB. Here is their result:=0D
>=0D
>Test # of eqn matrix size F90 C++=0D
>---- -------- ----------- ----- -----=0D
> 1 4353 520,000 3,244 3,095=0D
> 2 3492 320,000 1,612 1,532=0D
> 3 2088 180,000 921 871=0D
> 4 987 65,000 265 250=0D
>=0D
>The "matrix size" is in # of doubles and the timing values are in msec.=0D
>The authors' conclusion is that C++ has a speed advantage of 5%.=0D
>=0D
>Note that valarrays is a C++ STL implementation.=0D
>=0D
>Does anyone out there have similar results? C++ is a complex language and=
=0D
>STL in general in not very efficient. It is strange that C++ with=0D
valarrays=0D
>can beat Fortran. There are certainly a lot of unknown factors such as=0D
>the quality of PowerStation v 4.0 and the testing program, the switches=0D
used=0D
>for the compilation and others. If LF90/LF95 cannot maintain a speed lead=
,=0D
>many engineers/scientists will jump to C++ quickly. I am afraid that this=
=0D
>is happening in many places, including this university.=0D
>=0D
>CKS=0D
=0D
=0D
hey 3 years ago was that it executed the code I wrote in about half the tim=
e as the Microsoft Fortran.=0D
-David =0D
=0D
----------=0D
From: Jim Kress=0D
Sent: Wednesday, February 10, 1999 3:30 PM=0D
To: Carlstrom_Rick; [address removed]
Subject: Re: [LF] Fortran 90 vs C++ Valarrays=0D
=0D
Let's be a little careful with the 'everybody will jump ship' comments. Fo=
r=0D
the older codgers like myself, FORTRAN is the language of choice and will=
=0D
continue to be for the foreseeable future.=0D
=0D
I have worked with C++ and find it to be a nasty programming language=0D
designed (IMHO) primarily to restore computer scientists to their previousl=
y=0D
occupied 'god of the machine' position. The syntax seems to be deliberatel=
y=0D
obtuse and the vocabulary associated with the design and implementation of=
=0D
C++ systems is an arcane mishmash of gobbledygook that serves only to=0D
impress the uninformed and confuse the rest of us.=0D
=0D
I am not a fan of any language whose fundamental purpose seems to be to=0D
generate bloated code that no one else can understand or reuse - and yes, I=
=0D
know about classes and inheritance and all the rest of that nonsense. I=0D
have yet to see any C++ project implemented that actually allows ANY=0D
significant reuse of the design much less ANY code.=0D
=0D
I do agree that FORTRAN needs to keep a performance edge on other languages=
.=0D
In particular, Lahey Fortran needs to do that as well as provide some of th=
e=0D
basic amenities offered by other FORTRAN compilers (a list I sent to Bob=0D
Runyan=0D
some time ago) including integrated OpenGL support, integrated basic=0D
graphics primitives, a integrated 'Quick Window' development environment,=
=0D
etc.=0D
=0D
FORTRAN uber alles!=0D
=0D
Jim Kress=0D
=0D
=0D
-----Original Message-----=0D
From: Carlstrom_Rick [address removed]
To: [address removed] [address removed]
Date: Wednesday, February 10, 1999 1:43 PM=0D
Subject: RE: [LF] Fortran 90 vs C++ Valarrays=0D
=0D
=0D
>I think one of the keys here is the use of Powerstation 4.0. Here at my=
=0D
>company we had PS 4.0 and Lahey 3.5. A structural analysis DLL was made=
=0D
from legacy fortran 77 code first with Powerstation and then 4 months later=
=0D
>with LF90. Swapping out the 2 DLL's indicated that Lahey had a factor of =
4=0D
>speed advantage over powerstation and maintained a full trace back table i=
f=0D
>the code crashed.=0D
>=0D
>Now I cannot comment on the effects of compile options used to compile=0D
under=0D
>Powerstation and Lahey. (I do not have the original project files that=0D
>generated the DLL using powerstation)=0D
>=0D
>One Comment I have is that LF903.5 created slower executibles then=0D
>F77L3/EM-32 on mathematically intensize code.=0D
>=0D
>I will agree that if the various FORTRAN compiler venders cannot keep a=0D
>speed advantage over C++ they will looses there client base quickly.=0D
>=0D
>Rick Carlstrom=0D
>Head of Programming=0D
>Nichols Advanced Marine Enterprises.=0D
>=0D
>-----Original Message-----=0D
>From: Ching-Kuang Shene [address removed]
>Sent: Wednesday, February 10, 1999 3:16 AM=0D
>To: [address removed]
>Subject: [LF] Fortran 90 vs C++ Valarrays=0D
>=0D
>=0D
>Hi all,=0D
>=0D
>In a recent article "Comparing Fortran 90 and C++ Valarrays" in C/C++ User=
s=0D
>Journal, the authors (Shyam Bhat and B. Arun) compared MSVC++ 5.0 and=0D
>Microsoft PowerStation v 4.0 with their finite element method program on=
=0D
>a Pentium 150MHz with 48MB. Here is their result:=0D
>=0D
>Test # of eqn matrix size F90 C++=0D
>---- -------- ----------- ----- -----=0D
> 1 4353 520,000 3,244 3,095=0D
> 2 3492 320,000 1,612 1,532=0D
> 3 2088 180,000 921 871=0D
> 4 987 65,000 265 250=0D
>=0D
>The "matrix size" is in # of doubles and the timing values are in msec.=0D
>The authors' conclusion is that C++ has a speed advantage of 5%.=0D
>=0D
>Note that valarrays is a C++ STL implementation.=0D
>=0D
>Does anyone out there have similar results? C++ is a complex language and=
=0D
>STL in general in not very efficient. It is strange that C++ with=0D
valarrays=0D
>can beat Fortran. There are certainly a lot of unknown factors such as=0D
>the quality of PowerStation v 4.0 and the testing program, the switches=0D
used=0D
>for the compilation and others. If LF90/LF95 cannot maintain a speed lead=
,=0D
>many engineers/scientists will jump to C++ quickly. I am afraid that this=
=0D
>is happening in many places, including this university.=0D
>=0D
>CKS=0D
=0D
=0D