Pasi_FM

This is certainly our priority as soon as we get our hands on this tech as based on all the information publicly available right now the effect on benchmark scores is so huge that any benchmark score should come with the information whether this technology has been used or not.

Well the contest was open to anyone and it was very much a contest of OC skill. The specific contest rules and hardware regulations are determined by the sponsors. Note that there was also a more public friendly precision OC category in the contest where 63 different users successfully hit the score target and many more tried (252 different registered users hit within 10 3DMarks of the target). As to the 3DMark 11 submit number as received by Futuremark, the ballpark is easy to verify as due to our current policies most results are visible to anyone on the 3dmark.com website. K404's calculations have an error. 43 weeks times 35000 is 1.5 million submits, not 15 million. Current benchmark submit count is 1501261 3DMark 11 results in the database, 1470951 of which are currently visible on the 3dmark.com website. The link to the first visible submitted result is http://3dmark.com/3dm11/103 The first result within the Gainward competition timeframe is http://3dmark.com/3dm11/1740497 The last result within the contest time period is http://3dmark.com/3dm11/1893992 So, 153495 3DMark 11 results came in during the contest timeframe, 132437 of which are currently visible to anyone on 3dmark.com. The rest are results from unsupported version 1.0.0 (after the patch was released, older 1.0.0 results are still visible), results with pirated keys, results that have been invalidated and results hidden or deleted by users. Feel free to look around the results by changing the ID if you want some verification that there are indeed as many results as stated.

Oof. Maybe El Gappo should subscribe to the Futuremark newsletter or follow us on Facebook to keep up with our OC contest offerings. Our mailing list does contain over 200000 active subscribers accumulated over several years, but to my knowledge we've never claimed that they would all be overclockers. We also receive about 35000 3DMark 11 benchmark results each week, so surely it isn't an outrageous claim that we received 5000 submits (which would actually be fairly close to the actual submit number) that match the contest criteria during the four week period. We don't make stuff up. In the contest Cische is now in the lead with P11039, but Massman is still in second place. There's little benefit to sending scores in early, so there will probably be serious entries coming in over the weekend.

Hey guys, I cross checked his result file with our database and all scores are obviously too high for the detected components (about twice what you'd expect). It probably messes with the system clocks in some way as approximately the same score boost is seen in both GPU and CPU scores, but not sure what the exact behavior of this GPU Booster is. We're currently looking into ways of automatically detecting this cheat. Any information that helps implementing detection to stop this kind of cheating at the doorstep is helpful and can be emailed directly to me at pasi(at)futuremark.com . Goes to show that all serious OC competitions should involve a manual check of the winning scores against the detected components like HWBot and Futuremark competitions do.

Yeah, such results would not show a user name. One more related thing that I don't think was mentioned is that there is now no limit to 3DMark06 scores on an account, so the 1/account limitation is only for 3DMark 11. The new service is now live, so check it out. Still some issues to iron out, but it's mostly functional. http://3dmark.com/login

Small clarification to this. Deleted results can't be viewed or recovered, but if you submit results that end up not being connected to your account (because you already had an older result or just weren't logged in) are viewable by you and others, found on searches and such and checked for result validity. In the future scores will by default be public, unlike in the current ORB.