Yes that's very strange. My PSU is a cheapo Itek 550w, but I don't think it's the problem, because the ASUS V9950GE FX5900 which is the very same model (not true the asus has higher memory frequency) works like a charme on the very same configuration.

Nemo1985 wrote on 2022-08-04, 13:41:

Yes that's very strange. My PSU is a cheapo Itek 550w, but I don't think it's the problem, because the ASUS V9950GE FX5900 which is the very same model (not true the asus has higher memory frequency) works like a charme on the very same configuration.

Don't have that Asus card, but if for some reason it draws its power from the 12V rail instead of the 5V rail (like the A350XT), then this would certainly be a huge difference.
Anyway, as I said, it's very unlikely that this is the actual cause (typically the card just complains that it does not receive enough power), but it would definitely be good to eliminate this potential problem (and others) by going with a high quality PSU that has at least 15 - 16 amps on the 5V rail.

bloodem wrote on 2022-08-04, 13:29:

I wouldn't pay too much attention to that. They probably had a faulty card and assumed that it was because of the "earlier" revision.
Both of my rev A2 cards perform flawlessly and have had no issues with them.

That's good to hear.

I ordered a Leadtek 5900XT before seeing that review and it's also revision A2. The card hasn't arrived yet (customs procedures pending) and I'm hoping that it doesn't have any performance issues either.

bloodem wrote on 2022-08-04, 13:29:

My Winfast A350XT cards are both rev A2.

Correction: the card in the Asus A8V Deluxe build is a rev C2 (I accidentally added photos of the same card in my "database").
Compared to the rev A2 version, this one seems to run a bit cooler, but what stands out to me is the fact that it overclocks like crazy (it can go beyond GeForce FX 5950 ultra clocks). It's unclear if this overclocking headroom is just pure luck or something that can be expected to achieve with later revisions of these cards.

In other news, I've benchmarked this build extensively. The results can be seen here.