Re: Merced to suck(?)

Gregory Alan Bolcer (
Mon, 20 Oct 1997 09:03:33 -0700

Intel is claiming a 4-gigahertz chip in their next
generation chip. (4,000MHz for those of you at home).
My response is I'll believe it when I see it. Hopefully
this new line will fix their extremely poor floating
point performance.

I am underwhelmed by the Alphas, they can only manufacture
at 750MHz, so while those are some good Spec95 numbers,
what really impresses me is the 600MHz UltrasparcIII which
sampled at SPECint95 of 35+ and SPECfp95 of 60+.
Now that's fast.

In other interesting news, Sun announced their 200Mhz java processor
which claims a Caffeine Mark 3.0 of ~14k. This is the same
CMark as a Ultrasparc II running at 200Mhz with their JIT
compiler, but in hardware and at 1/5 the cost. I had seen
them press release pre-announce a $99 JavaBlaster plugin board for
386-586 machines that runs independent of the cpu and includes
a 100bT network card. This would be interesting development for
all those bookends I have in my closet.

...speaking of benchmarks, SGI has an OpenGL version of Quake
that runs under Irix 6.x. It's a beautiful port and while
has some texture fragmentation errors which make it not
as optimized as the PC GL versions, it is network compatible
with them. Sun has an OpenGL implementation for Solaris 2.5.1+
and I keep lobbying the SGI folks for the source code so I
can (trivially) recompile it and have a Solaris Quake version
also. Anyone up for some lobbying? Anyone know someone at SGI?


> Am I the only one underwhelmed by the bits leaked at Microprocessor Forum?
> Apparently one of the Intel droids (Fred Pollack) said:
> "wait until the *second* IA-64 chip - it will knock your socks off"
> Sheesh, excuses already for vapour-ware supposed to materialise in two
> years. Lame-o-rama.
> Digital demoed Alphas with the next generation chip, 21264.
> The current generation 21164 gets SPEC 18.8, 29.2 whereas the latest
> 300MHz Pentium II score is 11.9, 8.56.
> -- R.