Christopher Smith wrote:
Andrew Lentvorski wrote:
It is the collected wisdom of several generations of microprocessor
designers inside IBM, Motorola, AMD, and DEC.
Ummm.. because it is generational you kind of have a problem there, as
"Unix" and "Windows" have changed a lot over the generations,
particularly Windows.
True. But I'd still bet on it. The fact that gcc compilations under
UNIX are *still* one of the best memory stressors lends credence to that.
Unix was always the toughest to get working on a new microprocessor.
Windows was always the easiest.
Talk to the folks at Intel and AMD folks about the pain they go through
ensuring Windows compatibility vs. Linux compatibility and get back to
me on that. ;-)
Well, since Intel and AMD simply ignore legacy Linux (more than 2 years
old), that's pretty much a no-brainer. Whereas, Intel and AMD will test
compatibility the whole way back to Windows 3.1.
No one ever understood why. It was simply true.
Well, I can give you a very straightforward explanation: until NT came
out, "Windows" didn't imply a preemptively scheduled threaded system
with multiple concurrent processes and a virtual memory system. Unix
did. NT was a PITA to get working on Alphas, PowerPCs, etc.
As an anecdotal counter point on this: Linux rarely crashes when I
overclock my CPU, but Vista... ugh.
Not necessarily Vista's fault. It could be an FPU problem. Linux
doesn't need the FPU. I'm pretty sure Vista does.
-a
--
[email protected]
http://www.kernel-panic.org/cgi-bin/mailman/listinfo/kplug-lpsg
