Similar to the Forsaken results, the Quake2 scores recorded were impressive as well, easily playable in resolutions up to 1600x1200.

We can draw some conclusions from the way the results broke down not only as we went up in video resolution, but also as we utilized newer 3D applications in testing.

First off, we need much faster 3D acceleration.

Even at 200/240MHz, the Hercules Dynamite TNT2 Ultra is falling behind the potential of the Athlon 650 in most of the newer benchmarks we used such as Quake3Test 1.08.

Expendable and Descent3, also newer titles, seemed to be very limited on the video side, even at a relatively low 640 x 480 x 16bpp.

Forsaken and Quake2 didn't have this problem as they're not constantly throwing massive levels of polygons or triangles up on the screen each second. Being that they were principally designed/coded in 1997, those two titles were both written largely for 1998 hardware, meaning the Voodoo Graphics and Voodoo2 cards primarily. Because of this they don't bog down the newer and faster TNT2 Ultra video card as much as the more graphically intensive programs like Quake3 and Expendable/Descent3 do.

Only when the synthetic benchmark results are analyzed does it become apparent how much potential Sharky's Machine has to offer at full steam. When the Athlon can stretch its legs and power through synthetic software specifically designed to test its limits (rather than games which are written to work well on almost any system) it shows what raw power can do within a benchmark.

In particular, the 3DMark99 MAX score (using a controversial AMD-modified, Athlon-specific 3Dnow! .DLL file not supported by Futuremark) and both the ZD WinBench99 and SiSoft results, tended to back up and validate the speed that 650MHz of integer and floating point power provides.