Hello, been a while, dual CPU mobos

Lennart Sorensen lsorense-1wCw9BSqJbv44Nm34jS7GywD8/FfD2ys at public.gmane.org
Thu Sep 9 18:49:11 UTC 2004


On Thu, Sep 09, 2004 at 02:28:32PM -0400, Devin Whalen wrote:
> I was going to pipe in with a defence of the P4 design but after
> Lennart's email I now realize that I know nothing about computers and
> that my only possible rebuttal is...."My cat's breath smells like cat
> food"....I will now proceed to go home and throw my brand new Pentium
> machine in the garbage :)

Well the P4 design does work real well on massive data crunching such as
video encoding and the like, when it is written and compiled to use SSE2
instructions to work on multiple pieces of data at once, in long
identical loops.  It's really fast at some things (with optimized code)
and not so fast at others (with or without optimized code).

Now imagine how confusing it will be when intel finishes the switch to
model numbers.

It is pretty obvious to most users (and should generally be tru) that a
530 will be fast than a 525, although there might also be a 527 with
more cache but the same clock speed as a 525 but less clock speed than
the 530 added in, and then you again end up having performance really
depend on your code.  And while a 530 should be faster than a 330 (you
would hope), how will it compare to a 340 or a 350?  So far intel has
3xx, 5xx and 7xx series but are talking about a 6xx series and I believe
an 8xx series as well.  The Pentium-M is probably going to be one of
them, and the Pentium 4 another, and the Celeron-D (P4 derived celerons)
the 3xx series.  Clock speed was never a good indication, but it was
what consumers had.  AMD tried to indicate speed with their model name
rating system, although I am not sure a dual memory channel 2ghz athlon
64 with 1M cache is necesarily always faster than a 2.4ghz athlon 64
with 512k cache, but the former is 3500+ socket 939 and the latter a
3400+ with socket 754 (socket 939 is dual channel ram, socket 754 is
single channel ram, the older 940 is dual channel with ECC used by
opterons (for servers) and the first FX chips (which were opteron 1xx's)).

AMD is doing something similar but not quite the same for the opteron
chips, with 1xx being single cpu chips, 2xx being dual cpu chips, and
8xx being >2 cpu chips.  This is the number of cpus that can be placed
in one machine in this case.  A dual core chip will still only count as
one cpu as far as cpu interconnects are concerned.  But that's next
years problem. :)

Lennart Sorensen
--
The Toronto Linux Users Group.      Meetings: http://tlug.ss.org
TLUG requests: Linux topics, No HTML, wrap text below 80 columns
How to UNSUBSCRIBE: http://tlug.ss.org/subscribe.shtml





More information about the Legacy mailing list