Well, I ended up having to get a whole new motherboard (thread).

I like the one I got. A Tyan board (s1864 or something like that), which is good because my prior mobo was a Tyan that I liked and it's still going strong in my print server at home.

This new board I just bought is the first AGP-4x capable board I've owned. I've set the proper settings in the BIOS to allow it to do AGP 4x with my 4x-capable GeForce card.

Everything seems to work, my games run fine, etc. My only question is: How can I tell if AGP 4x is really working or not? What software can I run that'll measure the AGP capabilities?

I've already tried the Ziff-Davis 3D WinBench software. What a piece of crap. It forces you to run an hour's worth of quality tests before it'll do any throughput testing. Then, when you try to run the throughput tests, it fails saying that the GeForce card's HAL doesn't allow the features required to perform the tests. I downloaded 80 megabytes for THAT?

Does anyone have any better (preferably simpler) benchmarks I can play with?

___________
Tony Fabris

How can I tell if AGP 4x is really working or not? What software can I run that'll measure the AGP capabilities?


There are a couple of tools out there that will do the job... But to be honest, it's really going to be a wash. Entech (I think) gives Powerstrip (I would be in Linux, else I'd send you the link). Unless you are using DDR or faster memory, AGP 2x is still within the specs of the transfer rate... There were a couple of big writeups done on the hardware sites when 4x was new.... Basically, Nothing aside from a FEW custom q3a levels even comes close to using that much bandwidth on your video bus for texture transfers.... Everything is done now with the local texture memory on the video card.

I *THINK* I saw a 1fps improvement switching 4x on and off. And I'm not sure if it was faster with or without at this point.

If it's doing good, chances are, it's at least 2x, which is really all you need to worry about....

AGP 2x, 4x, et. all are a lot like the x rating on the cdroms... Sure, it will DO that... under these circumstances, on the outside of the disc, if the disc is perfectly clean.... Only change it to be WITH the right drivers, and fast enough main memory, and no significant load on the processor (agP... It's a direct port on the processor, not a bus), and the bios support, no fast writes, etc....

Does anyone have any better (preferably simpler) benchmarks I can play with?

Here.

Kim

http://gamershq.madonion.com/products/3dmark2001/

I can't imagine why you would recommend that one over all the other ones out there

I use 3dmark, and it does have a benchmark that pushes out massively large textures to the card. :-D

Calvin

Thanks, Kim and Eternalsun. I knew there was another one besides the ZD labs software, I just didn't remember what it was.

___________
Tony Fabris