14 Nov
2003
14 Nov
'03
7:16 a.m.
On Nov 14, 5:35, Eric Smith wrote:
uses
There's no challenge in naming a few parts that
have speculative
execution.
No, I suppose there isn't, assuming one knows what the term means at
all, because it's common in modern processors.
However, I may have been a bit hasty. Although I
don't think the
R4600 was out at that time, after thinking about it some more it
occurs to me that some earlier R4K series parts may have had
speculative execution, and probably the DEC Alpha (20164).
I don't know much about the inner workings of Alphas, so I didn't
mention it. Nor do I know all that much about the inner workings of
R4K prior to the R4600, except that I know an R4600 will outstrip an
R4400 at the same clock speed (modulo cache differences) and an R4400
will outdo an R4000 by a fair bit. In fact, the earliest R4000 chips
were embarrassingly slow. I don't think R3000 or R4000 had speculative
execution, but I'm not absolutely sure.
I meant the
Pentium family.
Is there really a Pentium family? Aside from the Pentium Pro,
Pentium II, and Pentium 3, which use similar cores, there seems to be
little microarchitectural similarity between parts for which Intel the "Pentium" name. For instance, there
seems to be more similarity
between the i486 and Pentium cores than between the Pentium III and
Pentium IV cores, despite the fact that the Pentium was superscalar
and the i486 was not.
I'm sure you're right. I've almost deliberately avoided knowing too
much about Intel chips post-8085, except to find out what MMX really
does, and discovering along the way that Intel do seem to have changed
their minds a few times.
--
Pete Peter Turnbull
Network Manager
University of York