Paul Elmer, you both are clueless hobbists. When you, together with Uberclockmeister, claim that a score of 55 "is clearly superior" to score of 54, you just prove yourself as clueless idiots.
I can enlighten you a bit. The whole PCI traffic during Business Winstone benchmark takes LESS THAN 3% of the run time, including DISK TRAFFIC!. Therefore even if your lovely AGP would be infinitely fast, a business user will never see his application runnung faster than those 3%. All Windows paintings are CPU-limited, got it? Your lovely hobbist Uberclockmeister is trying to find a difference where there is no differences except in board setup and drivers, and that explains why he got all these contraversal results, all within the jitter of measurements. Help yourself and visit back his PCI vs AGP ramblings:
tomshardware.com
There are no "superior" nor "inferior" implementations of AGP, all must be within specifications. The difference most probably come from the additional quick software patch that need to be applied for VIA AGP port to make it work with current MSFT driver, which is MOST LIKELY optimized for P6 and 440LX. MOST LIKELY that that EXTRA piece of code is responosible for the 1% drop in VIA AGP performance. I hope I do not need to explain to you that any chained driver code does not improve overall performance.
AGP was and is an awkward attempt of Intel to catch up with development of 3D-accelerated cards. For simple applications like games any recent 3D card provides better choice. For serious CAD application the Intel IA-32 cludge is a limiting factor and any faster output bus will not help either. IMO, AMD/VIA are playing this game only because this is a MARKETING GAME, no more, no less, for ignorant pseudo-experts like yourself.
Now go back to Intel for your hobbist's rantings. |
