3D Mark 2001SE  is not a way to show the bottom of the line performance, but I still think that it's a good tool when used in conjunction with other benchmarking programs. The scores surprised me a little bit here because I was hoping that I would see some improvement in static tests because they recycle some code, but it seems not enough to make cache make a difference.

3D Mark 2003 as well as 3D Mark 2001 should not be used as a stand alone benchmark, but should be used in conjunction with other benchmarks to tell true system performance. I was hoping that this test would show some difference in performance because 3D Mark 2003 has a CPU test that I was hoping would take advantage of the increase cache.

With my third test, I tried out GLMark.  GLMark is basically the same as 3D Mark but tests Open GL performance instead of DirectX. Like the tests I have done to this point, there still has not been any performance improvement with the extra 256k of cache that the Barton chip uses.

Let's move on to some more tests and see if we can get a better idea of why the extra cache is important.