Slashdot Mirror

The SiCortex website mentions their top of the line machine using about 900mW per 1.4GFlop processor (1.55Glop/W), but the overall system uses 20KW for a 5832 processor system, giving a system figure of (5832 * 1.4) / 20,000 = 0.40GFlop/W.

http://www.pathscale.com/products/high_capability_system_sc5832

Now, consider that DARPA wants 50GFlop/W - a factor of 100x improvement over the SiCortex number.

So, I'd hardly classify SiCortex's products, cool as they may be, as "just this sort" of energy-efficient supercomputer.

In fact the IBM Roadrunner supercomputer TFA refers to does better thsn SiCortex's system, and achieves 0.44GFlop/W.

http://en.wikipedia.org/wiki/IBM_Roadrunner

Just keepin' it real!

Phoronix stated in the second paragraph: "For our testing we had used the final Intel 32-bit releases of the four most recent Ubuntu releases except for Ubuntu 8.10 "Intrepid Ibex" where we used the Intrepid release candidate."

This benchmark would be more valuable if it compared how the ubuntu repository fared when compiled with the different available compilers:
-Intel compiler
http://www.ubuntugeek.com/howto-install-intel-c-compiler-10-on-ubuntu-feisty-fawn.html
-GNU compiler
-SunStudio compiler
http://developers.sun.com/sunstudio/downloads/index.jsp
-Portland group has unifying binary for both intel 64-bit and amd 64-bit
http://www.pgroup.com/about/why_pgi.htm
-Pathscale compiler
http://www.pathscale.com/node/189

It would also be nice to discuss benchmarks for:
-AMD 32-bit Kernel
-AMD 64-bit Kernel
-Intel 64-bit Kernel

5) AMD and Intel continue to beat the crap out of each other with customers gaining but wondering why there is no software that supports those new 8-way processors, as both compilers and third-party developers fail to keep up.

The compilers exist, in the sense that outfits like PathScale and PGI already have compilers that support OpenMP and some degree of automatic parallelization. They need a lot of work to scale to larger numbers of cores but the primary roadblock here is integration with IDEs and moving these technologies into mainstream computing. If these companies and Microsoft figure out how to make these compilers pervasive with Visual C++, etc. things will change quite dramatically. I don't think this will happen in 2007, though. What will happen is that compiler vendors make significant strides improving access to parallel programming models, pareticualrly with support for Co-Array Fortran and UPC.

why is this news?

Agreed. This was explicitly stated as the reason for switching. I also find it amusing that the previous Slashdot headline is titled "Intel Loses Market Share to AMD".

Since there is no topic, I'll go offtopic.

AMD has completely kicked Intel's but in price, performance, and power requirements. Opteron's with their HTX memory controllers and now using them for interconnects http://pathscale.com/infinipath.php is absolutely off the hook. Intel does have better mobile chips though.

For about $200k, you can throw together a very fast general purpose beowulf cluster. About 64 dual-core processors, and save thousands in power every year over anything Intel has to offer.

In 2007, Intel is going to add direct memory controllers to their x86 and Itanium processors, but what they need to do lower their prices. They already have the best manufacturing capabilities of any chip maker, so they need to give the CxOs a pay cut and lower their prices. Time will tell.

Have you tried Pathscale compilers http://www.pathscale.com/ekopath.html? They seem to give quite decent performance on AMD64 chips.

I haven't used their compiler, but this looks fscking sick: http://www.pathscale.com/infinipath.php

AMD has come a very long way the past few years. If they had Intel's fabrication skills they would be almost unstoppable. This HTX stuff simply looks like the bomb.

Have you tried Pathscale compilers http://www.pathscale.com/ekopath.html? They seem to give quite decent performance on AMD64 chips.

I haven't used their compiler, but this looks fscking sick: http://www.pathscale.com/infinipath.php

AMD has come a very long way the past few years. If they had Intel's fabrication skills they would be almost unstoppable. This HTX stuff simply looks like the bomb.

Have you tried Pathscale compilers http://www.pathscale.com/ekopath.html? They seem to give quite decent performance on AMD64 chips.

And I've been very happy with it.

And I've been very happy with it.

There's really only one reason to use Intel's
compiler -- for performance. It's well known that
Intel's compiler generates code that vastly
outperforms everything else for the same platform
(namely Microsoft Visual C++ 6/7 and gcc -- everyone
else (Watcomm, Borland) has long since been
relegated to "also ran" status).

Pathscale's EKOPath is the compiler I see most performance-oriented folks use for AMD64 platforms. It knows about EM64T too.

Quote from the article: A start-up called PathScale offers an open-source compiler that's compatible with GCC 3.3. "Our company is trying to be the GCC alternative for people who care about high performance," said Len Rosenthal, vice president of marketing for PathScale..
From this page and their claims about GCC front end (source code) compatibility I infer that their compiler is just an back-end (code generator) to GCC using GCC's front-end. I think that writing full C/C++/Fortran front end with GCC compatibility would be no small feat (especially that most big companies use Edison Design Group front-end, developing only their back-ends).
What looks a little bit suspicious are PathScale's performance claims. It looks like the PathScale compiler for x86 generates much faster code than Intel's own in-house compiler (which is being developed by people hired from KAI known for creating a very high performance compiler) or the industry-famous Portland Group compiler. I would be more than happy to see that PathScale's product delivers what is promised.

Quote from the article: A start-up called PathScale offers an open-source compiler that's compatible with GCC 3.3. "Our company is trying to be the GCC alternative for people who care about high performance," said Len Rosenthal, vice president of marketing for PathScale..
From this page and their claims about GCC front end (source code) compatibility I infer that their compiler is just an back-end (code generator) to GCC using GCC's front-end. I think that writing full C/C++/Fortran front end with GCC compatibility would be no small feat (especially that most big companies use Edison Design Group front-end, developing only their back-ends).
What looks a little bit suspicious are PathScale's performance claims. It looks like the PathScale compiler for x86 generates much faster code than Intel's own in-house compiler (which is being developed by people hired from KAI known for creating a very high performance compiler) or the industry-famous Portland Group compiler. I would be more than happy to see that PathScale's product delivers what is promised.

You missed PathScale, a compiler for AMD64 that supposedly generates absurdly fast code.

Who said there are only 2 compilers? He only tested 2, that's all. Here's a bit of a list - and notice that some of these are targeted specifically to scientific computing:

1. GCC
2. Intel compiler (Intel only)
3. Comeau
4. PathScale (Opteron only)
5. Portland Group (PGI)
6. Borland

A good review would have pitched the 3.6Ghz nacoma vs an Opteron 150, would have tested both in 32 and 64 bit and tried to use some application benchmarks.

Since the code for these benchmarks is available, it would have been really interesting (for me--as a developer/environmental modeler who compiles his own codes) to see what performance boost these compilers would have given (as compared with default "gcc" builds)... A lot more work, I'll admit.

application availablility along with essentially non-existant tier-1 vendor support.

Let me put it this way: you are living under a rock.

what's the point of upgrading your hardware if you only want to run 32bit x86 binaries?
The performance on legacy code alone is enough to warrant the upgrade.

Show me such scalability with an opteron!
Cray will be doing so soon, I believe with 10,000+ Opterons in one machine, and I'm sure you'll be pleasantly surprised.

no *good* AMD64 compilers
Ah, so you haven't heard of Pathscale then?

This year promises to be very interesting indeed.

Pathscale by former SGI'er does just that.