2D test on an old card

[mig]

Dear Australian friends,

I have one doubt about some `video error coruption` that BIT gives to me. I used 5.3, 6.0 and 7.1 and it's the same. Sometimes there is no error for hours, sometes it gives much errors at start and so on. BUT, the card never had a single issue in gaming and in other test batteries. It is AGP 3.0 card MSI GeForce 6800GT w 256MB DDR3 256bit, stock cooled

Beacuse of my curiosity I used several strong tests which NEVER EVER returned an error on this particular card:

Ultra-X QuickTech Pro 5.90 - tested all video mem 256M without error for 6h
M2K Scope v14 - tested all video mem 256M without error for several times
PC doctor v7.06 - tested all video mem 256M without error for 100 cycles

OCCT 4.4.1, FurMark 2.1, ATITool 0.26...stability tests all great for hours
3D mark 2001SE, 2003, 2005, 2006 for days without any glitch in textures

and also Mike's VMT - video memory tester 1.7 which i used for 3 days in a row says 0 errors

http://home.mikelab.kiev.ua/index_en.php?page=PROGRAMS/vmt_en

Rest of the rig is COMPLETELY errorless, tested with the most powerfull tools:

AMD Athlon 3200+
Abit NF7-S2
2x512MB DDR-400 Geil Blue
WD RE 160GB
Enermax 425W

OS used: WinXP SP3 and Win2003 R2 with latest offical ForceWare driver

I don't know how BIT use mem, through DX (DirectDraw?!) or something different, because it is odd that only BIT 2D test randomly gives errors?!

I've notices that in combination with 3D test (which assumes some VRAM) errors are always poping out in 2D and much quicker, but never for 3D test

Thanx on support

[Ian (PassMark)]

BurnInTest tests the Video memory by writing test data to the video memory and reading the test data back and verifying it is correct. It does this via DirectX (and the graphics card device driver).

In the past there have been a few different explanations for this error:
- Incorrect voltages can lead to this problem;
- Device driver issues;
- Nvidia's "P4 cache coherency rules" on old cards.

Please see "http://www.passmark.com/forum/showth...ory-Corruption
I don't know how the other software you mention tests video memory, so I don't know why they don't report issues. I suspect either they are just exercising the video RAM (i.e. just writing to it), or they are using a different mechanism for testing.