> On Mon, 2 Feb 1998 21:02:41 +0100 (CET), Kurt Garloff wrote:
>
> >I doubt, that the 6x86MX-233 is faster than a iPII-233. It is possibly
> >faster than a iP233MMX as long as you don't start doing floating point
> >calculations or using the MMX features, but the iPII-233 is superior.
>
> Hardly. It is brute force faster, not technolically superior. A 6x86 P233 only
> runs at 187MHz. A more impotant factor in over all system performance is
> bus speed. I would take a 75x2.5 Cyrix over a 66x3 PII anyday. Drive
> access, and memory I/O WILL be faster then the PII.
The 6x86 has technically very advanced features, such as register
renaming, data bypassing, out of order completion and speculative
execution. That's why it is able to run faster than a Pentium or MMX with
higher frequency. The Pentium (MMX) lacks these features.
The PII however also has some of these features. It has a backside bus for
the L2 cache, whichs runs at half processor speed (116.7 MHz or more).
This effectively decouples the CPU from the mainboard speed. So, the bus
speed of only 66.7MHz is not really a disadvantage. Only a few
applications which work on much more than 512 kB data could be a little
bit faster with SDRams on a 75MHz or 83.3MHz system.
As for I/O: Most I/O interfaces don't nearly use the full bandwidth of the
PCI bus, not the parallel, serial, IDE or SCSI ports. (PCI bandwidth on
a 33.3MHz PCI bus is 133.3MB/s.) The only one really taking the PCI to
it's limits is the graphics cards if you do 3D games or similar. But then
the profit of a higher PCI bus clock is questionable. 37.5 or 41.7MHz is
clearly out of spec and can cause problems e.g. with SCSI controllers.
(However, there are boards with async PCI bus, but then you don't profit,
becausePCI bus clock is 33.3MHz anyway)
Kurt Garloff, Dortmund
<K.Garloff@ping.de>
PGP key on http://www.student.physik.uni-dortmund.de/homepages/garloff