Barton vs Thoroughbred:
Disabling half the XP3200+'s 512kB L2 cache to measure the effect on performance is not possible, so we are forced to use a real 256kB L2 cache CPU - in this case our trusty XP2200+. This CPU will not run at XP3200+ speeds of course, so we'll have to clock the XP3200+ down to XP2200+ speeds (1800MHz) for this test. To keep things otherwise even we ran both CPU's at 266MHz FSB and 400MHz CAS2 memory. It would be nice to claim that any performance difference we see is exclusively due to the different cache size, but in reality we are comparing the Thoroughbred core and the Barton core. Any greater efficiency of the Barton core will have an effect on the benchmark results in this section. If we were comparing P4's of a certain vintage for example, Hyperthreading would have to be considered. However, we have not been informed of any major architectural changes to Barton over Thoroughbred, apart from the increased FSB and larger L2 cache.
Zero difference here. I would have expected the larger cache of Barton to have an effect.
This is a curious result, with the larger-cache Barton being slightly SLOWER than Thoroughbred. More overhead managing the larger cache?
POVRAY shows an insignificant difference.
3DMark03 also shows an insignificant difference.
Yet again, virtually no difference in performance between the two cores and once again we find ourselves surprised. The increased FSB had little effect and so did the larger L2 cache, or any optimisation Barton has over Thoroughbred.
All original content copyright James Rolfe.
All rights reserved. No reproduction allowed without written permission.
Interested in advertising on OCAU? Contact us for info.
