Re: [tlug] Is Prime PC a good enough place to buy a PC

[Ian Wells <ijw@example.com>]

On 10/23/05, Lyle (Hiroshi) Saxon <ronfaxon@example.com> wrote:

Stephen J. Turnbull wrote:

>I'll admit that there's a perceptible difference between a 450MHz P2
>and a 2.4GHz Athlon, but the Athlon box also has 1GB RAM where the P2
>has 256MB, and a 133MHz disk controller where the P2 has a 33MHz
>legacy bus.  (I don't have the controller cache size offhand, but that
>matters a lot, too.)
>
>

This is OT, but I'm curious about a hardware question that I've been
wondering about.  I have a P-III 450MHz, and I was wondering what the
highest clock speed of the P-II's was.  Was it 450, 500, or...?  The
last 486 was 100MHz I think, and the first Pentium-I was 90MHz (I
think).  Was there a similar overlap with P-II's and P-III's?
Incidentally, I noticed a sharp increase in usable performance when
going from a P-II 350 to a P-III 450, much more than the 100MHz should
have generated.  According to a friend of mine (I haven't properly
researched it), the design of the P-III does a much better job of
managing memory than the P-II?.

Lyle

The last 486 was a DX2-133,  I believe.  Original Pentiums certainly went to P133, possibly P166, PPros to 200MHz.  PIIIs are difficult to quantify, cos something based on the PIII core was in their Pentium Ms.  But I've certainly used PIII-Xeon-800s.  Somewhere in that sequence they went from 1 instruction per clock to 2 - the first Pentiums, maybe?

I've used 8088s, 188s and 286s but it's far too long ago to remember about such things...

PIIIs were, in my experience, much better at instructions per cycle than the PIIs.  And the P4s as well - the original P4s had faster clock speeds but were a step down in performance on the best PIIIs, and even now don't seem to perform as well as their clock speed would suggest.  Compare a P4 2.8 with a mobile Pentium 1.7 laptop, for instance; there's not anything like a 50% improvement - and AMD are competing with their made-up clock-speed-equivalency numbers which are now tending towards 2x the actual chip speed.

--
Ian.