Matt Thomas wrote:

> At 01:59 PM 6/8/2000 +1000, Simon Burge wrote:
> >Matt Thomas wrote:
> >
> > > (cd usr.bin/make; make clean; csh -c 'time make')
> > >
> > > ds5k:   580.7u 183.2s 13:20.01 95.4% 0+0k  15+923io 1193pf+0w
> > > mp800:  352.1u  68.2s  7:23.12 94.8% 0+0k  12+846io  256pf+0w
> > > vs4k:   ???.?u  ??.?s  5:42.70 ??.?% 0+0k  19+331io  471pf+0w
> > > cobalt:  91.1u  16.7s  2:01.75 88.6% 0+0k 786+974io 1078pf+0w
> >
> >60MHz R4400 (DECsystem 5900/60), src via NFS:
> >
> >143.4u 29.7s 3:05.16 93.5% 0+0k 313+1178io 391pf+0w
> >
> >With the DECsystem, 60MHz is the external clock speed, not the internal
> >doubled clock speed - eg, lmbench reports 119MHz.  I'd say you should
> >be getting about the same performance as my DEC box all things being
> >equal.  I'm guessing that the C compiler doesn't do FP - that'd kill
> >the VR4121...
> 
> The DECstation 5000/125 is very slow.  I'd expect to closer to the VAX
> in terms of performance.  The lack of FP during doesn't matter since
> that's mostly integer stuff.

I think there's some crossed wires - what I was trying to say was
that my DECsystem 5900/60 should be pretty much equal to your mp800
integer-wise (120MHz r4k internal core vs. 131MHz r4k internal core),
but your mp800 doesn't have any hardware FP.  One obvious explaination
for such a real-world benchmark discrepany would be that some FP jumped
in somehow.

For a straight integer R4400 vs VR4121 shootout, what do you get for the
following?

	mona:/tmp 43> cat foo.c
	main()
	{
		int i;

		i = 100000000;
		while (i-- > 0)
			;
		exit(0);
	}
	mona:/tmp 44> cc -o foo foo.c
	mona:/tmp 45> csh -fc 'time ./foo'
	10.9u 0.0s 0:10.93 99.9% 0+0k 0+0io 1pf+0w

Simon.