Deeko wrote in part ""Riddle me this, how many good benchmarks do you know that can vary by 1000 points with no changes to the system? One time I ran it, I thought I must have left FSAA enabled, nope, 3DMark is just inconsistant. I fired up Q3 to see if everything was scoring low, nope,""
Sorry, but if scores of identical systems are off by 1000 points, then there "IS SOMETHING" wrong or different between the 2. Might be a simple bios setting. Might be memory. But something is different, as that is NOT normal. If you don't want to correct it, don't run the test. It's up to you. But to say these results of 1000 points off is typical is not correct. 5% off is about what you can expect. Run the test a few times in a row, and it should actually score within 2%, which is normal for any benchmark.
Also, so 3DMark (which uses D3D) didn't run like you expected, and you then ran Q3 to see what's up???? Since when does Q3 use D3D?? This is a perfect example of not using correct testing methods. As I keep saying, compare the same things!!!! Comparing a Q3 benchmark (OpenGL) scoring to 3DMark (D3D) scoring, is like comparing ATI to nVidia scores in 3DMark. Can't be done. OpenGL can be at 100%, while your D3D setup can be at 70%. Using UT as a video benchmark, or system checker is also useless, as it's results depend more on the CPU than anything. Plus there is the false UT results that users with 3dfx cards think they are getting, when using Glide and 32bit. But, back to the root. If your system scores vary from other identical systems by more than 300 or 400, or about 5%, then there is some tweaking that could be done. This program is free, and small, unlike the other benchmarking programs you used.