DDC said:
i want to spend a minimum pf 150$ for an upgrade.
By this i would move from an a7nx8-x to that a7n8x-e-dx.
The advantage with it would be the dual memory and it's capacity for
the raid ide drive and it nf2 ssp technology.
With this i think i would get an improvement in video performance and
video encoding. (that is with the new motherboard.)
Or on a second tough, should i get an athlon xp 3200 400fsb. presently
i'm on a 2600 xp 333 Barton core 512megs of ram.
thank for the advise.
The first thing you should do, is find some representative
benchmarks, done with enough motherboards so you can see
what factors influence processing speed. The Tomshardware
Interactive CPU charts are one source:
http://www23.tomshardware.com/index.html
The default chart that comes up for me, is a Divx 5.2 chart.
Now, I don't know anything about video encoding, so you'll
have to decide if that chart applies to what you are doing or
not.
What strikes me most about that chart, is CPU core speed matters
more than anything else. The memory subsystem also makes a
difference, if the memory subsystem is inadequate to keep the
processor busy (that is why some of the P4 results are out of
order in the chart). Buying a 3.8GHz P4 would cost a fortune,
and one way to get there, is via overclocking a slower processor.
On the Athlon64 side, the trick would be to go to the private
forums, and find the cheapest processor known to handle
overclocking well.
P4 570 3.8GHz Prescott 1:40 minutes $639
P4 3.2E Prescott 1:58 minutes $209 <--+-- equal
Athlon 64 4000+ Clawhammer 2:00 minutes $473 <-/ performance
Athlon 64 3000+ Winchester 2:37 minutes $149 <-\
AthlonXP 3000+ Barton 2:40 minutes $116 <--+-- equal
performance
The bottom two entries in the table above, were put there
to show you that the P.R. rating of the two AMD processors
is a pretty good indicator of their relative performance.
In other words, switching from your 3000+ Barton to a
3000+ Winchester, apparently buys you nothing.
The reason I recommend the benchmark route, is the performance
you get is very dependent on how the application is written.
Some applications are optimized for one processor family better
than the other. Sometimes, a great deal of performance can be
gained by a re-write of the application, and that could be a
lot cheaper than buying new hardware. (I.e. Finding a different
coding app could give you a lot more improvement in processing
time than cranking up the hardware. Sometimes this involves
accepting a compromise in output quality.)
Paul
