Fortran - Intel 10 for MAC OSX ?

Author

Intel 10 for MAC OSX ?

Engineering Calculations

2007-08-25, 8:06 am

Is there one or will there be a cross compiler to build for both
Windows and OSX ?

Just currious,

Jim

James Klein
Engineering Calculations
KDP2 Optical Design Program
www.ecalculations.com

jwm

2007-08-25, 7:10 pm

On Aug 24, 11:04 pm, Engineering Calculations
<ecalculati...@ecalculations.com> wrote:
> Is there one or will there be a cross compiler to build for both
> Windows and OSX ?
>
> Just currious,
>
> Jim
>
> James Klein
> Engineering Calculations
> KDP2 Optical Design Programwww.ecalculations.com

this link can give you the answer:

http://softwarecommunity.intel.com/...y/en-US/forums/

Steve Lionel

2007-08-25, 7:10 pm

On Aug 25, 1:04 am, Engineering Calculations
<ecalculati...@ecalculations.com> wrote:
> Is there one or will there be a cross compiler to build for both
> Windows and OSX ?

Cross-compiler between Windows and MacOS? We don't offer one and it
would be tricky as it would also require a cross-linker and some way
to get the appropriate libraries on the other platform. We do have
native compilers on both platforms, of course.

Steve

Gordon Sande

2007-08-25, 10:12 pm

On 2007-08-25 20:08:42 -0300, Steve Lionel <steve.lionel@intel.com> said:

> On Aug 25, 1:04 am, Engineering Calculations
> <ecalculati...@ecalculations.com> wrote:
>
> Cross-compiler between Windows and MacOS? We don't offer one and it
> would be tricky as it would also require a cross-linker and some way
> to get the appropriate libraries on the other platform. We do have
> native compilers on both platforms, of course.
>
> Steve

Any chance of a joint product for less than the cost of the two separately?
One might think of it as one compiler and two runtimes.

Steve Lionel

2007-08-25, 10:12 pm

On Aug 25, 8:41 pm, Gordon Sande <g.sa...@worldnet.att.net> wrote:

> Any chance of a joint product for less than the cost of the two separately?
> One might think of it as one compiler and two runtimes.

I've seen many people request a cost-reduced multi-os license. Some
other vendors do offer this. It's strictly a business decision,
there's nothing technical involved. I will pass on your request to
the appropriate folks.

Steve

Engineering Calculations

2007-08-26, 7:14 pm

Steve Lionel <steve.lionel@intel.com> wrote:

>On Aug 25, 8:41 pm, Gordon Sande <g.sa...@worldnet.att.net> wrote:
>
>
>I've seen many people request a cost-reduced multi-os license. Some
>other vendors do offer this. It's strictly a business decision,
>there's nothing technical involved. I will pass on your request to
>the appropriate folks.
>
>Steve
I'm not sure how many folks will find the new Macs to be the kind of
overall solution that I have found my new Mac Book Pro to be but it
gives me the power to do essentially everything at the choice of an
operationg system to boot into.

I have had Macs and PCs over the years but this is the Cat's PJs of
computing and the price is commensurate with buying two machines or
less. When my Enpower 221 notebook and my home built PC died a few
w

s ago due to a fried surge protector (took out an HP 940 printer
too) we bit the bullet and got the Mac portable and a new iMac for my
wife and it is a lot like having a Porsche 911 S4 and 2-1/2 ton 6x6
truck (sans the quad 50 on the flat bed) all in one. If more folks
find and use this type of solution, it will be time to start buying
some Apple stock again :-)

All the PC codes built using USB dongles also run on the Mac in XP now
and the XP software I use to get into the day job email system also
work flawlwssly.

Real Basic is based on something called a Universal Binary so what
runs on an Intel Mac also runs on a G4 or G5 (there will be a bunch of
these around for a long time). I don't een pretend to understand how
it works, but it does.

Things have come a long way to these tenth generation Intel
processors. My first PC cost the same as my new Macbook but had only
an 8086 and 256K ram and no hard drive in 1985.

8086, 286, 386, 486, Pentium, Pentium Pro, PII, PIII, P4 and now Dual
Core Intel. 22 years and now no more 0 and 1 programming like the
calculator in today's FOXTROT comic.

Whatever the Intel choice becomes, Steve, they will need to pry the
Fortran compiler out of my cold dead hands even if I also start Real
Basic programming.

We will start the process of going to the Intel 10 compilers to
replace the 9s tomorrow at the day job. Keep up the good work.

Sincerely,

Jim Klein

James Klein
Engineering Calculations
KDP2 Optical Design Program
www.ecalculations.com

2007-08-27, 10:44 am

> Is there one or will there be a cross compiler to build for both
> Windows and OSX ?

I don't think such things exist in commercial land. Other people will
certainly correct me if I'm wrong.

In the open source community, it does exist but only in one direction
AFAIK: it is possible to build GCC (and thus gfortran, which is part of
GCC) a compiler that runs on Mac OS X and creates Windows executables. I
have done it myself for a few other platforms (eg compilers hosted on
i386-linux creating Windows executables) and it will work flawlessly on
Mac OS. Guides on how to build cross-compilers targetting Windows (via
mingw) exist on the net.

The other way around is tricky: the compiler and assembler would work
fine in a Windows-hosted cross targetting MacOS, but not the linker. Even
though the Apple linker (which is open source, under the Apple license)
is loosely based on the GNU linker, it has evolved too much (with
addition of darwin-specific code) that it is not cross-friendly, as is
the GNU linker (ie able to be built on one platform and targetting
another). Two years back, I tried to modify it to work as cross, but
never achieved it (a Google search will show other people have tried, and
also failed).

Hope this helps, I've always been amazed so few people actually use
cross-compilers.

--
FX

Greg Lindahl

2007-08-27, 8:11 pm

In article <faue6s$1mn7$1@nef.ens.fr>, FX <coudert@alussinan.org> wrote:

>Hope this helps, I've always been amazed so few people actually use
>cross-compilers.

Build and test machines are now cheap, and fiddling with
cross-compilers is expensive for the end-user, even if it's a
commercial compiler and the heavy lifting is done by the vendor.

-- g

Richard Maine

2007-08-27, 8:11 pm

Greg Lindahl <lindahl@pbm.com> wrote:

> In article <faue6s$1mn7$1@nef.ens.fr>, FX <coudert@alussinan.org> wrote:
>
>
> Build and test machines are now cheap, and fiddling with
> cross-compilers is expensive for the end-user, even if it's a
> commercial compiler and the heavy lifting is done by the vendor.

Except, of course, for the embedded market, where cross-compilers are
ubiquitous. That's because the target systems tend to not have
development environments. In fact, in that market, it is common to do
much of the development with a software simulation of the target CPU.

--
Richard Maine | Good judgement comes from experience;
email: last name at domain . net | experience comes from bad judgement.
domain: summertriangle | -- Mark Twain

Greg Lindahl

2007-08-27, 8:11 pm

In article <1i3iize.1e58h1y11kg34xN%nospam@see.signature>,
Richard Maine <nospam@see.signature> wrote:

>Except, of course, for the embedded market, where cross-compilers are
>ubiquitous.

Right, which isn't what we were talking about.

-- g

James Van Buskirk

2007-08-28, 5:34 am

"Richard Maine" <nospam@see.signature> wrote in message
news:1i3iize.1e58h1y11kg34xN%nospam@see.signature...

> Except, of course, for the embedded market, where cross-compilers are
> ubiquitous. That's because the target systems tend to not have
> development environments. In fact, in that market, it is common to do
> much of the development with a software simulation of the target CPU.

I don't get it. Is IVF suddenly a native compiler? From what I've
seen it's been a cross-compiler forever.

--
write(*,*) transfer((/17.392111325966148d0,6.5794487871554595D-85, &
6.0134700243160014d-154/),(/'x'/)); end

Steve Lionel

2007-08-28, 7:12 pm

On Aug 28, 12:12 am, "James Van Buskirk" <not_va...@comcast.net>
wrote:

> I don't get it. Is IVF suddenly a native compiler? From what I've
> seen it's been a cross-compiler forever.

James, I'm not sure why you think that or what specifically you're
referring to.

IVF includes both native and cross-compilers. The cross-compilers are
IA-32 applications which target the x64 and Intel Itanium platforms
and which run on either IA-32 or x64 hosts, and there is also a native
Itanium compiler. The IA-32 compiler is native. On the Windows
platform, cross-building from Windows on one architecture to Windows
on another is common.

What is trickier is developing a cross-compilation facility that goes
across operating systems and this requires building some
infrastructure for it. Ideally you'd also want some sort of cross-
debug tools.

As Greg suggests, cross-compiling from one desktop OS to another
doesn't seem worth the trouble and the requests for this are more, I
think, from the notion that they'd like the second compiler for free
rather than it being an actual convenience. In Jim's case, he has a
single system that dual-boots MacOS and Windows - this would suit a
multi-OS license better than cross-compiling.

Steve

James Van Buskirk

2007-08-28, 7:12 pm

"Steve Lionel" <steve.lionel@intel.com> wrote in message
news:1188307992.560792.96750@m37g2000prh.googlegroups.com...

> On Aug 28, 12:12 am, "James Van Buskirk" <not_va...@comcast.net>
> wrote:

[color=darkred]
> James, I'm not sure why you think that or what specifically you're
> referring to.

> The IA-32 compiler is native.

The x64 compiler is not. It can easily run out of virtual address
space during compilation, whereas a native x64 compiler would go a
bit longer before this kind of event happened.

--
write(*,*) transfer((/17.392111325966148d0,6.5794487871554595D-85, &
6.0134700243160014d-154/),(/'x'/)); end

Steve Lionel

2007-08-28, 7:12 pm

On Aug 28, 11:09 am, "James Van Buskirk" <not_va...@comcast.net>
wrote:

> The x64 compiler is not. It can easily run out of virtual address
> space during compilation, whereas a native x64 compiler would go a
> bit longer before this kind of event happened.

Right - I said that. You seemed to be implying that all of the
compilers provided by IVF were cross-compilers.

I expect that eventually we'll offer a native x64 compiler as well.

Steve

James Van Buskirk

2007-08-28, 7:12 pm

"Steve Lionel" <steve.lionel@intel.com> wrote in message
news:1188317239.122625.316570@q5g2000prf.googlegroups.com...

> Right - I said that. You seemed to be implying that all of the
> compilers provided by IVF were cross-compilers.

Well, all the compilers unless you want to give up 50% of your xmm
register file, 75% of your integer register file, and about 99.9999%
of your virtual address space just so that you can say you're running
on a native compiler.

> I expect that eventually we'll offer a native x64 compiler as well.

Due to the price increases in IVF, I probably won't be able to afford
it. In the good old days, one could purchase a license and only
upgrade occasionally, but now you have to upgrade every year or buy
new. This is not a workable solution for those of us who do lots of
different things rather than simply program in ifort day in and day
out. Funny how the more of a multi-tasking system you have, the more
single-tasking price structures and installation programs (why can't
we just get along with -m64 or -m32 as gfortran does?) we are faced
with.

--
write(*,*) transfer((/17.392111325966148d0,6.5794487871554595D-85, &
6.0134700243160014d-154/),(/'x'/)); end

Sponsored Links