Unigine Valley 1.0 Benchmark Thread ** Post your scores**

[Face2Face]

Still does not make sense, because even at high frequency the bandwidth is a lot lower than a small OC on the RAM of a 7970 because of the bus width (7970 has 384 and GK104s has 256) GK104s max out at around 250GB/s while 7970 around 360GB/s... Something is not right.

Ok cool I will look into it.

 

[Teizo]

There is no doubt that Valley likes high ram frequency. My scores prove that beyond a doubt. John has his memory at 8ghz.

Maybe you should overclock your cpu and gpu a tad higher higher Ranger :biggrin:

 

[YBS1]

Well, since someone else brought it up...Any scores on here just showing a shot of the html results file should be taken with a grain of salt. Does anyone really think omeds score looks right? Almost 400 points higher than Piklar with similar clocks, about 160 above me with clocks that are way slower??? This is why I started posting camera shots.

 

[Rangerjr1]

Well, since someone else brought it up...Any scores on here just showing a shot of the html results file should be taken with a grain of salt. Does anyone really think omeds score looks right? Almost 400 points higher than Piklar with similar clocks, about 160 above me with clocks that are way slower??? This is why I started posting camera shots.

R2 290? [redacted]

 

[Face2Face]

YBS1 brings up a good point. From now on I will only accept scores posted with a full screenshot. It's not hard to do, press F12 when the bench run is over. I am looking at unusual results on a case by case basis.

R2 290? [redacted]

He is showing how easy it is to modify the results using HTML, hence why a screenshot is preferred now.

I am also working on a new spread sheet to show rankings.

 

[Elfear]

There is no doubt that Valley likes high ram frequency. My scores prove that beyond a doubt. John has his memory at 8ghz.

Maybe you should overclock your cpu and gpu a tad higher higher Ranger :biggrin:

I think Valley favors Nvidia cards too. Not a ton but if you look at Sohaltang and guskline's scores they are very close and clocks are close too. In general a 7970 and 680 at the same clocks will favor the Radeon but that's just the opposite in Valley.

Not really a big deal since different benchmarks favor different architectures. You just have to be careful about extrapolating benchmark performance to general gaming performance.

Nice score by the way Ranger. You definitely have better vram than I do.

 

[KingFatty]

YBS1 brings up a good point. From now on I will only accept scores posted with a full screenshot. It's not hard to do, press F12 when the bench run is over. I am looking at unusual results on a case by case basis.





He is showing how easy it is to modify the results using HTML, hence why a screenshot is preferred now.

I am also working on a new spread sheet to show rankings.

But this just creates one more layer, and it perversely makes you trust the results even more.

What I mean is someone can just edit the HTML, and then take a screenshot of the edited HTML. So you'll be more trustworthy of the results due to being a screenshot, but just as easy to fake.

 

[YBS1]

At Face2Face's request, here is what a 780SLI@1228/1800 score should look like (give or take a few points), keep in mind I'm running this bench on a 3930K@5.0Ghz and Valley does scale at least somewhat with cpu speed unlike Heaven.

 

[YBS1]

What I mean is someone can just edit the HTML, and then take a screenshot of the edited HTML. So you'll be more trustworthy of the results due to being a screenshot, but just as easy to fake.

No, he's saying it can't be a screen grab of just the HTML results file from now on. It needs to be an in program capture or camera (I assume this would still be ok). Yes, it still could be "faked", but at least they'd have to put forth some effort to do it.

 

[Face2Face]

But this just creates one more layer, and it perversely makes you trust the results even more.

What I mean is someone can just edit the HTML, and then take a screenshot of the edited HTML. So you'll be more trustworthy of the results due to being a screenshot, but just as easy to fake.

True it can still be faked but it will be harder to do so. We have enough legitimate results from all types of cards to sort out the fakes. This is supposed to be for fun and not a circus of phony scores...

 

[YBS1]

R2 290? [redacted]

He is showing how easy it is to modify the results using HTML, hence why a screenshot is preferred now.

Or was I??? :sneaky:

 

[Teizo]

Not really a big deal since different benchmarks favor different architectures. You just have to be careful about extrapolating benchmark performance to general gaming performance.

That is true, and I also have to fight the urge to game overclocked with this SLI set up as well. At stock it is just fine....but I always keep thinking...but yeah, instead of 95-120 fps you could be getting 110-135 fps :biggrin: (I'm only playing Blacklist atm)...

I reckon these clocks are safe as long as temps stay ok though since I am not touching voltage measurements.

 

[BallaTheFeared]

For anyone curious, this is what 607w at the wall looks like:

i5 @ 4.8/4.7 1.25v
7950 CF @ 1215/1750 max voltage

My HTML score is legit and I'll fight anyone who says otherwise, to the death!

 

[Durvelle27]

True it can still be faked but it will be harder to do so. We have enough legitimate results from all types of cards to sort out the fakes. This is supposed to be for fun and not a circus of phony scores...

:whiste:

 

[YBS1]

:whiste:

Well, there you have it folks....Slower than 780/Titan.

 

[Durvelle27]

Well, there you have it folks....Slower than 780/Titan.

But it was a stock run :\

 

[Face2Face]

Best I could do with my current setup - Top card is getting too hot and I cannot push the 2nd card any further without temps becoming an issue. These cards are in desperate need of water...

1200/1800

 

[monkeydelmagico]

Best I could do with my current setup - Top card is getting too hot and I cannot push the 2nd card any further without temps becoming an issue. These cards are in desperate need of water...

1200/1800

Only a matter of time now.....

Now that I've taken the plunge it takes every bit of willpower not to flash my card with the next gen bios. I've got dual bios switch and am itching to go over the 1300mv cap on my card.

 

[Elfear]

Best I could do with my current setup - Top card is getting too hot and I cannot push the 2nd card any further without temps becoming an issue. These cards are in desperate need of water...

1200/1800

:thumbsup:

Very nice results on air. Scratching my head a little though about the big difference in scores between you and Balla. A 15Mhz difference on the core shouldn't make such a big disparity in the score. Does Haswell vs IVB explain it?

 

[Teizo]

Good job Face2Face. I wish MSI had a Gaming Series version of the 7950/7970, although those AC Heatsinks you put on yours are nice.

Since someone mentioned that Valley scales with your cpu, I decided to run it ONE more time since I moved my cpu back up to 4.4ghz just to see what happened, and they were right. Got my highest score yet. Once I get an NZXT Kraken I should be able to get this cpu up to 4.8ghz and I can see how things go. That will be a few months though.

88/3680 with clocks @ 1293-1280/1851(7400)...cpu @ 4.4ghz

What I do find interesting is that while Valley has my memory listed correctly, it is reading my core @ 1371mhz....yet both Afterburner and GPU-Z read 1293-1280. I imagine Afterburner and GPU-Z are reading the correct speed though.

*updated*
*updated*

 

[Face2Face]

:thumbsup:

Very nice results in air. Scratching my head a little though about the big difference in scores between you and Balla. A 15Mhz difference on the core shouldn't make such a big disparity in the score. Does Haswell vs IVB explain it?

Good question? I am not sure? I ran my 3570K @ 4.8Ghz and re ran the test a couple of times with the nearly the same results.. The cards are not throttling and I can go higher on the memory (1830Mhz), but it doesn't bring my score up at all. Not sure

Good job Face2Face. I wish MSI had a Gaming Series version of the 7950/7970, although those AC Heatsinks you put on yours are nice.

Since someone mentioned that Valley scales with your cpu, I decided to run it ONE more time since I moved my cpu back up to 4.4ghz just to see what happened, and they were right. Got my highest score yet. Once I get an NZXT Kraken I should be able to get this cpu up to 4.8ghz and I can see how things go. That will be a few months though.

87.1/3644 with clocks @ 1293-1280/1825(7300)...cpu @ 4.4ghz


What I do find interesting is that while Valley has my memory listed correctly, it is reading my core @ 1371mhz....yet both Afterburner and GPU-Z read 1293-1280. I imagine Afterburner and GPU-Z are reading the correct speed though.

Valley doesn't correctly give core clocks on Nvidia cards with boost. I see you got a boost from the CPU OC. Looking good!

 

[YBS1]

What I do find interesting is that while Valley has my memory listed correctly, it is reading my core @ 1371mhz....yet both Afterburner and GPU-Z read 1293-1280. I imagine Afterburner and GPU-Z are reading the correct speed though.

It reads mine wrong too. Sometimes by a lot, one time it showed my classifieds clocked at 1683?? or so on the core...I thought about screenshot'ing it just to mess with you all.

 

[omeds]

As requested by face2face

 

[Face2Face]

I changed the results layout. I added a ranking column, let me know if you guys like this better or of you like the old one.?


NEW RESULTS LINK

 

[KingFatty]

I don't suppose there would be any way to, ah, well, add another column for the GPU voltage? [runs and hides...]