News for hackers. Stuff that matters.

A comment on The end of Wintel. Parent comment

Re:The end of Wintel by e40 on Tuesday August 03, 2010 @11:53AM (#1571586)
From a [native compiler] developers point of view, Intel's fatal flaw: the architecture is nearly impossible to write a compiler/assembler for. That's what Intel told my company (that makes native compilers). They said we wouldn't be able to make a high performing compiler because the instruction scheduling, etc were too complex and not documented well enough for us to do it.

A port of our compiler, which normally takes a couple of months, was estimated on IA64 to be 1-2 years, for a high-performing result.

This is where they failed.

Oh, the AMD64 (what it was called when we started the port) took 2 months, and that was fully optimized.

[ Parent | on HN ]
Re:The end of Wintel by InclinedPlane on Tuesday August 03, 2010 @03:18PM (#1572360)
Yeah, I only mentioned that in passing but it is a serious issue. All the more pressing because a lot of the promise of IA64 hinged on smart compilers. It was a dramatic change of course for CPU architectures since the previous trend had been to make the hardware smart enough to figure out how to take advantage of ILP on the fly, even with unmodified code (thus the superscalar architecture of the P5, the out of order & branch prediction strategies of P6, etc.)
[ Parent | on HN ]