rtsurfer

Exactly, if he had found some magical tweak or if it was due to a newer GPUPI version, then why couldn't he pass sofos with his 3-way sub. http://hwbot.org/submission/2851407_

Look at those clockspeeds. There is no way that those score are possible at those speeds. Plus in the past I have seen results in 3Dmark Benchs such as Firestrike where the person was running on a single GPU & had other dormant in the system. That gave they a boost in points, which wouldn't be there if it was the sole GPU installed in the system. I'd do a practical test instead of just theory, but I don't have more than 1 290X. Maybe you can get if confirmed with someone else with multiple GPUs. Edit:- Also Hwbot could just ask the guy to reproduce the results with only a single GPU in the system, if he can, then its legit. If not, then my theory is correct.

Mat could you take a look at these scores. http://hwbot.org/submission/2851399_ http://hwbot.org/submission/2851389_ While the GPUs are not actively running the Benchmark, them being in the system is clearly helping. Can it be considered legit.?

Its rebranded Gskill guys. Nothing ground breaking.

Lol... Top 10 seems to be the sweetspot then.

So that's why the server went down again. Couldn't handle this score.

How about some for the U.S guys.?

Yes it is. And it is a standard "tweak".

I believe this flaw can be easily fixed by just disabling mixing GPUs altogether. Obviously only the creator of the Benchmark can tell us how easy it is to do that.

Actually I can see, how the first case wouldn't work. A simple algo would pick the newer card as the faster one & rank the results accordingly. But what if someone ran a 7970 & a 260. The 260 is newer, so the algo would rank the results as 2 x 260, but the 7970 is actually faster. Hmnn.

I thought you had an Algorithm for that.? Didn't know that wasn't the case. Or we can leave it for the user & hope that the cheated results get reported, shaky but these are the only 2 options I can see.

You can combine any two cards, the result will be ranked in the stronger card category. For example if you run 290X & 7970, the result will be counted as 2 290Xs.

Its allowed. But if its not a stable overclock for this bench, then the run will fail. Its same as the "Not in Round error" people get on Super Pi when pushing memory. Lower the frequency a little bit & it will pass. Sometimes you get lucky & are able to get the bench pass at a frequency that was previously unstable.

Is it over clocked.?

I don't see how.

Probably.. The competition this time seems to be less CPU dependent & can be done on Z97 unless the quad channel on Haswell-E gives more Memory Read Bandwidth compared to Tight PSC on Z97. Unlikely, but can't be sure till it's tested.

Lol. That's actually perfectly fair. Or else all you would see is people with 5960Xs on top.

Vietnam 4790Ks have been really good on Ambient & for sub zero there have been a few good chips. Not all of them suck.

The Physics question still stands. Is it going to be divided by the number of Cores.?

If the overclock isn't stable, the run will fail. And this bench has a different definition of "Stability".

I somehow seemed to have entered the wrong time on the Submission(oops). Seems I cannot edit, can a mod please fix it.

This bench allows you to mix cards from the Same Vendor e.g All Nvidia cards. For the ranking, you will be ranked according to the fastest card you used. Otherwise the ranking system would be impossible.

Alright. Thanks mat.

Even for Single GPUs if we look at the Generational increase we might reach to zero fast maybe even with the upcoming 390X Just look at the difference between 7970 & 290X at similar clockspeeds http://hwbot.org/submission/2719953_vadimua_gpupi___1b_radeon_hd_7970_35sec_410ms http://hwbot.org/submission/2760482_d0minat0r_gpupi___1b_radeon_r9_290x_17sec_929ms A 390X on Ln2 might just get to zero. And for multi GPU, you already saw this http://hwbot.org/submission/2820501_pulse88_gpupi___1b_3x_radeon_r9_290x_5sec_907ms 4 x 290Xs on LN2 & its over. In the begining of the thread, people were quite ready for a longer GPU bench, don't know what changed.

Yeah.. Points. Although shouldn't the points have been enabled on the 32B version of the Bench since with the 1B version we will be under 1s with the next gen GPUs or maybe the Gen after. Wouldn't that make this bench short lived, ~ 2 years or so. Just a suggestion.