I want to get my side of this whole situation out in the open…
Vapor was in town for some lab projects we have going on and I talked him into benching with me to qualify for MOA. We had a 160L tank of LN2, XPower boards and Ryan’s (GoatEater) untested chip. I setup the OS, drivers and applications and used the 197.75 drivers… I knew FM always lagged behind approving drivers and figured they hadn’t approved the 257.xx drivers yet.
Wednesday and most of Thursday we spent testing a few kits of RAM, how to clock the uncore on the XPower, and so on… Thursday night we went cold. We benched 480’s in SLI with the 197.75 drivers. And as you can see from the ORB links and screenshots (
run 1 ORB /
run 1 HWbot,
run 2 ORB /
run 2 HWbot) we did well with the 480’s and Ryan’s chip was pretty good, but the 197.75 drivers were not FM approved and we later validated that FM QA did not even test those drivers.
So, I have the score to be in second, but my driver choice means the entry cannot be used in the contest and the 480’s are pretty much tapped out score wise. I called in a favor to a buddy here in the Twin Cities to borrow his 5970 and 5870 to go higher in score.
Friday night we tossed the ATI cards on and benched for 7+ hours on Ryan’s chip. During the long session we achieved four scores above 43k and ended up having condensation issues and killing Ryan’s chip early Saturday morning as we were pushing for over 44k.
Saturday afternoon I finally uploaded the scores to ORB and did not even try to make them public at first as I wanted to hold out as long as possible to see what other scores were posted on the leaderboard, but I finally caved and tried to make the lowest score public and it would not go. Here are the screen shots and 3dr files from the Friday night session…
43023 SS /
43023 3DR
43474 SS /
43474 3DR
43553 3DR
43912 3DR
I emailed Pasi at FM and posted in the thread on the FM forums regarding the problem of the scores just not wanting to go Public. During the bench session we renamed 3DMark06.exe to 3DMark05.exe. In pretesting, we observed the score was not very consistent and renaming seemed to improve the consistency and therefore improved the average score slightly, by 50-100 points. Peak scores were unchanged, but low scores weren’t cropping up.
Since we were having problems with the results and had not received any word from FM, we switched over to another 980X and started benching again on Sunday night. However, we were still waiting on FM’s reply to see if they could resolve the ‘Make Public’ issue with the four 43+k ATI results I had uploaded. We ran out of LN2 around 2:00AM Central Time and did not beat our previous high score of 43912 from Friday night/Saturday morning.
We decided to wait up for FM’s reply to see if they could resolve the problem with the results. A few hours before the deadline, an offer came in from Andre for a valid score to secure the 2nd place spot… and out of desperation I took it. In the meantime, Pasi emailed back saying he looked at the results I uploaded previously and saw the the .exe was renamed, which made the checksum invalid and according to FM policy you cannot rename files (modifying the .dlls and .exes at all is forbidden). I uploaded the score from Andre, made it public and notified Pasi of the new project ID to be included in the contest.
The score I uploaded from Andre was pulled as you already know. Taking the 3dr file and uploading it as my own was a complete dumbass move. I exchanged several emails with FM this morning regarding the matter, to which I openly admitted to using Andre’s score as my own.
Regardless of my own stupidity, I want to congratulate the winners and wish them the best of luck in Taipei for the finals.
Bookmarks