Intel Fortran Compiler for Mac OS 10.5 Tiger?

[thomaspatko]

I have some programs that need to be compiled from source using a good Fortran compiler for Mac OS (computational chemistry programs specifically) and I understand that Intel is currently the best option available.

Has anyone tested the current version of the Intel Fotran compiler 10 for Mac OS on a beta (developer) release of Tiger with the new version of XCode?

I do not want to purchase the current student version (as it does not come with updates per my understanding) unless it will work on Tiger as I plan to upgrade shortly.

Any thoughts would be appreciated.

--Thomas

[mcneale]

I have a workaround for some of the SIGSEGV, segmentation fault problems. If you save program input files in DOS format, and then run programs using these DOS-formatted input files, the SIGSEGV faults are partly overcome. In a long input file I still hit the problem. Saving in MAC (or I suppose Unix) format seems to elicit the SIGSEGV bug. I write this to alert the programmers to this Leopard-only problem.

[filipesmg]

I'm also having this SIGSEGV problem with 32-bit version of ifort on Leopard, even with the new 10.1 compiler version. And I can't use the /fce/ compiler because it doesn't work on my macbook core duo (32bit).

[Trond_K_]

I am having exactly the same problem. I have tried various compiler options (turning off optimization) etc., but with no luck. I always get a SIGSEV on my macbook dual core with Leopard and Intel Fortran Compiler 10.1.006.

Cheers, Trond

[Steven_L_Intel1]

10.1.007 should be available now - try that.

[Po_D_]

Good morning Steve,

I tried 10.1.007 version and made tow tests :

1) Compile and run a standard Fortran code that read and write files and that does not link to an external library;
2) Compile f90gl librarires and compile/run a Fortran code that uses f90gl.

Task 1 passed but I saw that reading files from a mounted network drive did not work. For task 2, I have this error that I had since I'm running Leopard :

forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
sppr 0005B117 Unknown Unknown Unknown
sppr 0005A5B3 Unknown Unknown Unknown
sppr 00031634 Unknown Unknown Unknown
sppr 00011544 Unknown Unknown Unknown
sppr 000182AE Unknown Unknown Unknown
libSystem.B.dylib 9006F97B Unknown Unknown Unknown
Unknown FFFFFFFF Unknown Unknown Unknown
sppr 00022921 Unknown Unknown Unknown
sppr 00002A57 Unknown Unknown Unknown


Regards,

Pierre-Olivier

[Steven_L_Intel1]

Please submit a test case to Intel Premier Support.

[rewaju]

Does anyone know if the issue below has been resolved yet. I am having the same problem using intel fortran 10.1.007 on Mac OS 10.5 (Leopard).

Thanks.

Lanre


forrtl: severe (174): SIGSEGV, segmentation fault occurred
Image PC Routine Line Source
sppr 0005B117 Unknown Unknown Unknown
sppr 0005A5B3 Unknown Unknown Unknown
sppr 00031634 Unknown Unknown Unknown
sppr 00011544 Unknown Unknown Unknown
sppr 000182AE Unknown Unknown Unknown
libSystem.B.dylib 9006F97B Unknown Unknown Unknown
Unknown FFFFFFFF Unknown Unknown Unknown
sppr 00022921 Unknown Unknown Unknown
sppr 00002A57 Unknown Unknown Unknown

[Steven_L_Intel1]

Lanre,

What you have shown is an error message, not an issue. There are many, MANY possible causes of a segmentation fault, some of which are programming errors in the application. We'd need to see the program that generated this error to tell you if it is an issue in the compiler and whether or not it is resolved. If you have a test case, please submit it to Intel Premier Support and we'll be glad to take a look.

[guerom00]

Same issues here with v. 10.1.006 and Leopard 10.5.2.

In the meantime, I compiled my codes with gfortran. Will this issue be ever addressed ? (because this thread is quite old now)

TIA

[Steven_L_Intel1]

What "same issues"? This is a long thread with a variety of issues, some of which are coding errors.

[Po_D_]

Hi,

From my side I still have problems (libSystem crashes) when trying to compile f90gl, library that I was able to compile before under Tiger with ifort. I get :

libSystem.B.dylib 9479C5EB Unknown Unknown Unknown

when compiling this :

../util/sppr fwrap.f90

I tried everything - I don't know what to do.

Thanks

PO

[Steven_L_Intel1]

Report the problem to Intel Support and include everything needed to reproduce the problem. Make sure that you identify the compiler version as well (10.1.012 is current).

[guerom00]

MADsblionel:

What "same issues"? This is a long thread with a variety of issues, some of which are coding errors.

SIGSEGV while accesing libSystem.B.dylib

Hard to narrow down the issue but I have the feeling it has to do with accessing stdin or stdout

Random example : the following program now crashes, it was fine with OSX 10.5.1.

program H3_JA

implicit real*8(a-h,o-z)

read(*,*) rr,r,th
pi=acos(-1.d0)
x1 = 0.
y1 = 0.
z1 = 0.
q1 = 1.
x2 = r-rr/2.*cos(th*pi/180.)
y2 = rr/2.*sin(th*pi/180.)
z2 = 0.
q2 = 1.
x3 = r+rr/2.*cos(th*pi/180.)
y3 = -rr/2.*sin(th*pi/180.)
z3 = 0.
q3 = 1.
x4 = 0.
y4 = 0.
z4 = 0.
q4 = 0.
if(abs(th).lt.1.d-5) then
y1 = 0.
y2 = 0.
y3 = 0.
end if
write(*,'(9(f10.6,2x))') x1,y1,z1,x2,y2,z2,x3,y3,z3

end

Forgot to mention that in some cases, switching to the fce distribution does solve the problem.
So I guess it would be nice to clarify when to use fc, when to use fce, why, etc

[Steven_L_Intel1]

Re: fc vs. fce:

fc generates 32-bit code, fce generates 64-bit code. That's it. If you're seeing problems when updating MacOS versions, that suggests a bug in MacOS. It would be most interesting to know if the error occurs if you DON'T rebuild the application. Those dylibs in the error messages are Apple's, not ours.

[Ron_Green]

To follow Steve's comments, I would like to add a few points.

- As Steve said, segmentation faults can occur for many many reasons, coding errors, insufficient stack space in the user environment, etc.

- You are using a the 10.1.006 version of the compiler. This was released back in November. Since then there have been 3 more compiler updates, with another due near the end of this month. Registered users go to:

https://registrationcenter.intel.com

to download the latest compiler updates. This applies to evaluation customers too, who have 30 days of access to registrationcenter. I would recommend this - you are using Leopard and .006 was the first compiler to support Leopard. There have been many improvements to the compiler for Leopard in these updates.

Take some time to read through the forum. There are multiple notes on segmentation violation, albeit on Linux. The same Linux concepts apply to Mac OS. And of course, use the 64bit compiler (if you have a 64bit Mac) and try options:

-g -traceback -check all

to help isolate the fault.

ron

[guerom00]

OK, I found a good example. I dug up an old code which I compiled under 10.5.0 I think Run it as it is, without rebuilding it, and it runs OK.

make clean; make (using fc) --> SIGSEGV in libSystem.B.dylib

make clean; make (using fce) --> Runs OK.

Should I conclude that the fc distribution is kinda broken with 10.5.2 ?
IMHO, you should anyway advertise more on these 2 versions of the compiler : I learnd about this fce thingy reading this thread :D

[Steven_L_Intel1]

No, your test does not necessarily mean that the compiler is broken. Please do submit an example to support so that it can be investigated.

Have you read the product release notes or documentation?

[guerom00]

MADsblionel:

Please do submit an example to support so that it can be investigated.

Well, there is always the little program I have posted above

[guerom00]

So, I've installed version 10.1.012 and at first glance, it seems to solve the SIGSEGV problem. I'll continue testing it.

Thanks for the help.

[guerom00]

Almost forgot this thread :D

I now have installed 10.1.014 and still have this SIGSEGV error on some of my codes; codes which I since have compiled with gfortran and which run fine (the reason I forgot this thread. Didn't look into this more deeper)

Hopefully, my production machine under OSX 10.4 with ifort 9.1.036 runs like a charm. Since it is a production machine, I don't upgrade anything ! :D

What's that saying ? If it ain't broken, don't fix it. :)

[alain_hebert]

My legacy application is compiled with ifort 10.1 Build 20070913
on MACOSX 10.5.2. I get a SIGSEGV, segmentation fault as soon
as my code try to use an array allocated by malloc in a C function called by the code. I reported the bug to Intel (Issue number: 478952) with a simple code to reproduce the bug, but I am unable to consult my issue report as the website https://premier.intel.com produces a HTTP 500 - Internal Server Error.

What can I do now? Any idea? I had similar problems with Absoft, 15 years ago but at least they were responding.

Alain Hebert
professor