If you compare the benchmarks it's about the match for a 6600GT.
I don't think its a good comparison to compare the PCIe versions of these cards when considering their AGP counterparts by the way. Its well known that the 6800 GT AGP does better than its PCIe sister card. Thus, I hope you've found a benchmark that compares x1600 Pro AGP to 6600 GT AGP.
Its well known that the 6800 GT AGP does better than its PCIe sister card.
Do you have a link with any info on that? Unless it's clocked higher it should be damn near identical.
As for the X1600 PRO, alot of the benches I've seen are pretty wonky, even losing to an X1300 at times. It does well sometimes and pathetically others, I'd guess there are some driver issues that still need to be worked out.
Okay, I seem to remember reading some article on it a while back and can't seem to unearth it at present. So it could've been a driver thing or I'm just plain wrong.
My point is, I wouldn't compare AGP to PCIe directly, because you're not always dealing with the same card. Take the 6800 GS AGP which uses a different core GPU from the PCIe version, for example. Not to mention you have different bandwiths available on AGP and PCIe.
Well, the 6800GS is a bit of an anomaly as far as that goes.
The X1600 PRO is a new GPU, I can't see them doing a re-taping out of a separate version for the AGP card. So I'd expect them to be identical until we hear otherwise.
And (as mentioned in another post today) the bandwidth difference between PCI-express and AGP is not a factor at this time.
I've still never, ever seen any benchmarks between the two that suggest a performance difference per bus is anything other than completely irrelevant.
Based on that I think an AGP X1600 PRO would be pretty close to it's PCI-express couterpart, so as far as I'm concerned the question isn't how it'll perform, but if a PCI-express X1600 performs well for $130, and frankly, unless I was editing video I'd pick an X800 GT over both the 6600 and X1600 in this price range...