lindahl at pbm.com
Sat May 2 16:42:15 PDT 2009
> [Intel and Shanghai]
> > Is this deliberate?
> In the sense that they have no desire to support
> competitors hardware, yes. Not really surprising,
> if AMD made compilers I doubt they'd try and do
> Intel specific optimisations either..
Actually, that's what a part of the AMD/Intel anti-trust lawsuit is
about. Back in the day of plug-compatible mainframes, IBM would have
been in deep anti-trust doodoo for intentionally reducing performance
on competing hardware -- and their mainframe architecture, just like
x86, had a wide variety of add-on features which could be tested for
individually, just like SSE / SSE2 / SSE3 / etc.
Instead of making compilers, AMD chose to suport PathScale (now
SiCortex), PGI, and gcc, all of which also have Intel-specific
As for the blog posting, you shouldn't draw general conclusions from a
single toy test-case. And I think you'll find all compilers will do a
much better job if they know at compile-time that it's x**(-2)... so
if that's your real app, don't code it like that toy.
More information about the Beowulf