[Info-vax] RealWorldTech on Poulson

[Michael Kraemer]

Johnny Billquist schrieb:

> The performance difference with some smaller Alpha is not that big, so 
> we still about a 30% advantage.

Rather mediocre for a new architecture which needs to make an 
impression. Compare that with the introduction of HP's PA RISC
or IBMs Power.

> Who cares that the clock rate is twice? 
> That is irrelevant. Or are you saying that the clock ticks themself cost 
> you something?

heat dissipation. electricity bill.
And, more important and fatal: can't use it in "embedded" designs,
i.e. something that saved several other RISC's live.


> You do know that the SpecINT 92 has long been obsoleted because it 
> turned out to not be a good test, partly because it was too easy for 
> compiler writers to detect and explicitly handle the test cases? (There 
> were other problems too, which caused the tests to be replaced.)

Spec92 was a good benchmark in 1993.
It went obsolete when larger portions of code and data fit
into system's caches and was replaced by Spec95 and so forth.

> You seem to place way to much value into the Specmark tests results.

I don't see any reason not to use them as a guideline,
I always found them in sync with my own benchmarking.
What else could replace it, "personal feeling" that box
A is faster than B?

> So the claim that DEC was too small to make something like the Alpha, 
> where does that put Sun, HP and Mips?

They were smarter than DEC and more realistic.
Sun and Mips went fabless right from the start.
HP wasn't heading for high clock rates in the first place but put
an appreciable amount of cache onto the CPU
and so could beat the alphas at half the clock speed.
All three didn't tinker with PC OSs or proprietary
stuff but focussed on Unix when the market demanded it.
And, most important: they got their act together
when the time was right.