EKOPath 4 Compiler suite is now Open Source

[Edvard]

Looks like good 'ol GCC and pals are getting some good ol' fashion comp-o-tition...

EKOPath 4 Compiler Suite going open source with support available
Wilmington, Delaware June 13th, 2011

PathScale announced today that the EKOPath 4 Compiler Suite is now available as an open source project and free download for Linux, FreeBSD and Solaris. This release includes documentation and the complete development stack, including compiler, debugger, assembler, runtimes and standard libraries. EKOPath is the product of years of ongoing development, representing one of the industries highest performance Intel 64 and AMD C, C++ and Fortran compilers.
...

http://pathscale.com...opath-compiler-suite

So what is it?

http://www.phoronix....path4_open&num=1

EKOPath 4 is a high-performance compiler that up until now has been proprietary and costs nearly $2000 USD per license, but now it's open-source and can sharply outperform GCC in many computationally-intense workloads.
...
EKOPath 4 carries full support for the SSE2, SSE3, SSSE3, SSE4.1, SSE4.2, SSE4A, and AVX instruction sets. The run-time is GNU compatible and with the GCC tool-chain, provides optimized C/C++ debugging, excellent multi-core support, provides the PathDB debugger, a PathAS assembler, and supports OpenMP 2.5. The PathDB debugger was previously open-sourced and ported to FreeBSD on GitHub as the Path64 debugger.

So, what does it all mean?
Not sure at the moment.
I DO know that it means a previously very expensive compiler suite targeted to supercomputer environments will now be freely available for the average Linux/BSD/Solaris user to play with, that has been touted as a "CUDA Killer" for GPU code crunching, is blazingly fast, and produces optimized executables that in many (not all) cases run more efficiently and faster.

Discussion here:
http://phoronix.com/...th-4-Compiler-Suite/

Download here (nightly build):
http://www.pathscale...opath-compiler-suite

For now, it's only available for 64-bit Linux/BSD/Solaris environments.
However, the download page says Windows and Mac Betas will be available "Q1 2011"; any idea when that is?

* Edvard downloading now...

from Phoronix.com

[worstje]

For now, it's only available for 64-bit Linux/BSD/Solaris environments.
However, the download page says Windows and Mac Betas will be available "Q1 2011"; any idea when that is?

-Edvard (July 06, 2011, 11:57 AM)

Past tense. Q1 means Quarter one, which means January-March. Given the fact we are in July right now, we are in Q3, from which I would interpret that that page is outdated and/or that they are way overdue.

Besides that, it sounds very interesting.

[Edvard]

Hrm... can't get the damn thing to compile "hello world"...
Guess I got some learnin' to do... 

[iphigenie]

Fortran! that brings me back

All that parallel computing cleverness I never got to use once I left Uni...

[Edvard]

Yeah, I've always wondered why fortran is still so widely supported.
Is there THAT much software still done with it?
Does it have features that simply trump similar sets in other, more commonly used languages?
What gives?

[Eóin]

It is still one of the key languages of high performance number crunching software. I believe compilers for it can produce superior code than for similar C++ programs because lax rules in C/C++ regarding pointers limit the compilers scope for optimization.

Many libraries such as Blas, Linpack and Intel's Math Kernel Library are written in Fortran.

[iphigenie]

If I had to do some computational physics or something else that has a lot of dimensions/matrices/factors and I would be hard pressed to think what to use if not Fortran... and why not? No need for closures, aspects, objects etc. for maths modelling...

They *did* get rid of the stupid spaces at the beginning of lines somewhere in the mid 90s... right?

[Edvard]

Dunno.
I kinda figured Fortran was still useful, but yeah it makes sense it does better for high-end number crunching.
After all, Fortran = formula translating.

Apparently Fortran 2003 includes support for Object-Oriented programming and interoperability with C.
If I weren't already getting ankle-deep in learning C++, I'd give it a go just because.

Either way, I still can't get this darn thing to compile anything.
I'm probably doing a whole lot wrong, but community support is so new, and the documentation so sparse (or maybe just unfamiliar), there's nothing I can find as to what the problem might be.
Apparently the IRC channel has been hopping lately (as can be expected with a newly-opened piece of juicy software) so maybe I'll hop in there this weekend.

[Edvard]

Figgered it out...
Instead of running 'pathcc' to compile, I needed to run 'pathCC', which is the C++ compiler.
Apparently, 'pathcc' is the C compiler. o_O

Silly me...

[daddydave]

However, the download page says Windows and Mac Betas will be available "Q1 2011"

-Edvard (July 06, 2011, 11:57 AM)

They have corrected the download page to say "Q4 2011" now. Hey, it's (almost) a NANY 2012 project!

[f0dder]

Instead of running 'pathcc' to compile, I needed to run 'pathCC', which is the C++ compiler.
Apparently, 'pathcc' is the C compiler. o_O

-Edvard (July 12, 2011, 03:37 PM)

I've always found case sensitive filesystems (and the use of them) to be utterly moronic and confusing - this is another fine example

Anyway, interesting development... now (or 'soon'? ) we have EKOPath, LLVM and GCC on the open source front, with Intel and Microsoft on the closed path. Never bad with a bit of healthy competition

[Edvard]

I've always found case sensitive filesystems (and the use of them) to be utterly moronic and confusing - this is another fine example

-f0dder (August 16, 2011, 11:28 AM)

Maybe a bit daunting at first, and I don't know the original justification, but I've gotten quite comfortable with it; not confusing at all.

What I find confusing is this particular choice of file naming.
They could've warned us...

Anyway, interesting development... now (or 'soon'? ) we have EKOPath, LLVM and GCC on the open source front, with Intel and Microsoft on the closed path. Never bad with a bit of healthy competition

I'm actually looking forward to folks hacking other languages into EKOPath...
D, C#, .NET, Pascal, Ada, COBOL anyone? 

[f0dder]

I've always found case sensitive filesystems (and the use of them) to be utterly moronic and confusing - this is another fine example

-f0dder (August 16, 2011, 11:28 AM)

Maybe a bit daunting at first, and I don't know the original justification, but I've gotten quite comfortable with it; not confusing at all.

-Edvard (August 16, 2011, 03:33 PM)

Original justification is probably laziness, or some ill-conceived performance reasons - it's more code+runtime work to make the filesystem case insensitive.

I've never seen a good reason for case sensitivity, though, case preservation is sufficient.

What I find confusing is this particular choice of file naming.

-Edvard (August 16, 2011, 03:33 PM)

I've seen '.C' for C++ files... so it kinda makes twisted sense in that mindset.

[Edvard]

I tend to agree WRT case-sensitive filesystems, but *sigh* that's where I'm at so I just deal with it.
If searches and shell auto-completion could be case-insensitive, I could live with it even better.

Tsk, tsk, so much vitriol:
http://www.raizlabs....nix-case-sensitivity