I wouldnt discredit it so quickly, we saw the *same exact thing* with the A1 silicon of the R520. It was later attributed to a bad metal layering near the memory controller that when given enough voltage for long enough periods would nuke the card. I myself had almost 6 seperate cards which mysterously died. Some were volted via ATi-tool, after the first card the remaining were not.Originally Posted by The Asgard
Until this is proven one way or another I plan on keeping my memory near default speeds, I can always overclock it later.
_____________________________________
My card BFG 8800GT OC purchased from bestbuy:
(bench suicide *artifact/distortion after long use*)
GPU: 738
Mem: 955
Fan: 90%
Idle: 48c*
Load: 66c*
(max stable *long-term use*)
GPU: 715
Mem: 955
Idle: 48c*
Load: 66c*
__________________________________________________ __________________
Id like to add that for those reading, the G92 is finicky like the 8600GTS was. It has tendency to produce alot of "false-positive" artifacts as well as "false-stable" results. For example this card will artifact and image-distort around 680-690 gpu if I push my CPU high enough. The card itself is fine, somehow the interaction with the motherboard itself with an unstable CPU is producing the visual errors which are noted in both ATi-tool and in 3dmark.
With the processer @ 3.0 and 3.1 with the same voltage, the same exact card and settings are stable to 715, 726mhz and higher produces sporatic "snow" style artifacting. I figured I would add this line in for people who are having trouble clocking their cards, it may not be their GPU at all, but instead they have a CPU that is semi or totally unstable.
Bookmarks