Benchmarks IV
Now on to UT2003's in-built benchmark that stresses both the graphics card and subsystem like few games can. I've run it at 800x600 to try and minimise the effect even a top-end graphics card may have on the final results. As it's divided into 2 parts, firstly the fill-rate limited flyby component.
It's all fairly tight apart from the dire performance exhibited by on the on-board solution. That's to be expected, though. Now on to the 'bot matches which usually place a greater emphasis on the subsystem, especially with a high-end graphics card.
More of the same here, apart from the SiS648 and SB51G changing places by an inconsequential amount. The on-board solution continues to show just how inadequate it is with any modern game.
Lastly, let's take a look at Quake 3 running in fastest mode. Although not a benchmark that will stress any decent system, it continues to be one of the best illustrators of just how well system bandwidth is used. Quake 3, v1.30, run at 512x384 Fastest.
Substantiating what we've witnessed earlier, it seems as if both the DDR-based systems from Intel and SiS respectively virtually match each other in the performance stakes. This is no real surprise considering they're running the system memory at identical speeds and timings (166MHz and 2-5-2-2).
