Artefacts in 3D Games

[Thomas Olivaux]

Hello!

I've got a strange problem on my PC. When I launch a game with high bus speed, I mean 240 MHz or higher, I have video artefacts on all my video games (or even in 3DMarks). I thought that my VGA card was dead, but it's not the case. In fact, if I slow down my memory, for exemple, at 250 FSB, 5:4 so it's 200 MHz, I have no artefacts at all (even with my VGA memory overclocked). I tried with several DDR modules, I currently use Corsair PC4400 (275 MHz certified). What do U think, is it the CPU or the mobo?

Thx for any idea,

Thomas

[Thomas Olivaux]

Currently, I'm running with 270 FSB (4.05 GHz). When I set the memory for 270 MHz (1:1) I can use my PC without freezing, I can launch a couple of benchmarks too. But to play, I must slow down the memory to 216 MHz (5:4).

++

[nicknomo]

Do you have AGP/PCI lock on that board?

[larva]

What are your GAT and timings like?

[flapperhead]

check pci timings gat timings, like mentioned b4, also might want to check lateency.. if all fails then the memory is probably overclocked past its limit or the mb cant kust cant support those overclocked speeds and need to bump the ram down a little..

[Thomas Olivaux]

Hello and happy new year :)

Well, this board (Abit AI7) have AGP/PCI lock, currentrly locked to 66/33. The RAM is not beyond it's limit at all, it can do 277 MHz 3/4/4/8. I've tried with those "bad" timings, GAT set to A/N/A/D/D (the "worst" I think) and a lot of voltage anywhere (1.65v NB, 2.9 even 3.0v DDR) but no way :( The memory have been succesfuly tested on another mobo.

Do you think the CPU is the pb ? But why can I pass an entire SuperPI 32M then ? Can it be the mobo ? But this one seems to be great until 300 MHz FSB (or around)...

Thx,

Thomas

[Thomas Olivaux]

Well, I cheked... the memory is in fact given for 3/5/5/10 at 277 MHz (CPU-Z SPD analysis). But with my mobo, I can't go worst than 3/4/4/8... but I don't think it's the pb, because the PC doesn't freeze, it's only an artifacts pb in 3D games (exactly the same with a dead video memory) and it occures from 235 MHz, not only closed to the 277 MHz limit.

See U,

Thomas