http://www.ukgamer.com/forum/showthread.php?threadid=296


Unregistered
Guest

Registered: Not Yet
Location: 
Posts: N/A



Nice job getting data on alot cards in all of the different modes. I know what it takes to obtain that kind of data. Being an avid hardware fan I would say based on reading these reviews on a lot of sites that ATI performs better on AMD hardware and Nvidia performs better on Intel Hardware. I have not seen the kind of review where the mobo/cpu was thrown into the mix. I happen to know about sillicon and board design and alot of the benchmarks I see do not make any sense. I really do not have a preference for manufacturer, although I tend to support the source of my bread and butter, I just want what is the fastest. Like I said earlier there appears to be a preference for cpu/mb when it comes to which card is obviously faster. That being the case alot of the problems seen with drivers on different platforms is probably due to the gart (AGP) drivers and how well they move data across the north bridge of the chipset. If you really want to get a good look at how the cards perform with respect to each other you should also use the competitive cpu/mb in the matrix. I really like the work that was done here but you have left out one of the most important parts of the graphics data path. And that is the north bridge. I know that during the benchmarks, almost everything runs out of graphics memory, but that is extremly benchmark dependant and you can see that difference in the quality settings (more objects to render). The slower the FPS (more advanced games) the more data that has to be passed to the graphics memory from the mobo memory. Therefore the chipset drivers play a crucial role in newer games. If you do not believe this try changing the AGP Memory size in the bios. So far all of you hardware guys try to do a good job for the readers, but there has been one sticking point that still gets me. Rarely does anyone take into account the GART drivers(chipset). It would be real interesting to see a complete set of benchmarks on a canterwood system with a 3.2GHz P4.....