"Intel's compiler: is crippling the competition acceptable?"

[agenda2005]

Well, someone finally confirm all my rants on this forum concerning the practice of Intel in compiler and benchmark programs.
The guy(http://www.swallowtail.org/) finally wrote an elegant PERL script to disable the rather silly 'GeniuneIntel' check from the compiler. You can copy the script and use it for free with or without modification for Intel FORTRAN compiler running on LINUX. The patch can be found here:
http://www.swallowtail.org/naughty-intel.html

Here's some extract of what he has to say:


It is a shame that the Intel compiler, which use to be almost the no-brainer choice if your primary concern was fast code, is now being coerced into being a marketing tool. Crippling the output for non-Intel chips may mean that some published benchmarks may end up bogusly favouring Intel over AMD, but the cost is that if you want to release fast production code I can't recommend the (unpatched) compiler. There are an awful lot of AMD machines out there!

One of the candidate for this kind of mishievious behaviour is PCMARK. I don't really know what it represent, just that I don't get it why a P4EE at 3200MHz will be a better CPU than A64 at 2840MHz. Will anybody in it's right senses and have some clue about CPU chooses a P4EE at 3200MHZ over an Athlon 64 at 2840MHz. That's why I mention that PCMARK05 is a bogus benchmark here:
http://xtremesystems.org/forums/showthread.php?t=67192&highlight=Intel+compiler

because futuremark compiled it with an Intel compiler which is known to cripple performance on other competitors CPUs.
Sandra 2005 I read somewhere was also compiled with Intel compiler.

[agenda2005]

I've ran the patch myself and got a 14% faster code with my intel compiler over gcc . It was a releif over the hopeless result I was getting earlier, an example of which I mention earlier here:
http://xtremesystems.org/forums/showthread.php?t=67169

There's a discusion about someone who patch the intel C++ compiler for windows here: http://xtremesystems.org/forums/showthread.php?t=28918&page=2&pp=25&highlight=intel+compiler

[masterofpuppets]

GCC is a good compiler, but Intel is even better. I use it to compile pretty much everything on my system (except from the kernel, and packages that suffer breakage). I've used ICC outputted binaries on an AMD system before with no performance hinder. Maybe it is just me, maybe I am using an older version...

[Lithan]

Yeah, we've known for years that PCmark was complete Intel-favoring BS. Now we just know why. What other benchmarks used ICC?

[agenda2005]

I think what AMD need to do now is to start an agressive means of pursuing software engineers, games programmers and benchmark writers to start using a better compiler which is well optimized for their processor (PathScale Compiler @ pathscale.com). Even if they need to start carrying bags full load of $$$ to pay those guys to do so. It's not against the law, INTEL is doing it. If most games are recompiled with AMD optimized compiler I think Intel CPUs will suffer even more setback than it's apparent now.
Those of us who write and compile scientific programs know quite well that Intel CPU are just Inferior to A64 if you have the right compiler at your disposal.