Just like my findings the first time I ran these tests in December of 2002, I have found that my results came out the same. There just doesn't seem to be any rule that will work on every system. I have always said it's best to set your AGP aperture size as high as you can without matching or exceeding your systems physical memory but this rule may not be the best in all cases. I think the bottom of the line here is that AGP aperture doesn't affect performance as much as some people think it does. In fact, the only tests that I saw any real differences where the static tests. I guess that proves that static tests are very flawed when it comes to testing real time performance and with all the favoritism we have heard about in the news lately I don't think static benchmarks should be trusted at all.

Like I said in my last article about AGP aperture size, it would be great if people would post there system specs in the forum and explain what they think is the best rule for AGP aperture. Post comments here

In conclusion, I would like to state that I still believe that AGP aperture should be set to not match or exceed your system's memory.  For example, if you had 512Mb of system memory, you should set your AGP aperture one setting lower then 512Mb.  This is because you don't want your AGP aperture to exceed your system memory because the AGP card cannot use memory that isn't there and you don't want to match your system memory due to the fact that some programs will try to consume the entire AGP aperture and not leave any physical system memory for the program itself.