UNIGINE-Valley BENCHMARK scores

[cadaveca]

How about a stock benchmark? 7950 @ 900/1250?



1920x1200 here, 1080 isn't something I'll run normally, but I'll try another in a minute. Note my score compared to erocker's 7970 with 300 more MHz. CPU is at stock, 3.6 GHz turbo all cores, 3.9 GHZ single-core.

Useless bench is useless? Or is 7970 a wasted expense vs. 7950 for such workloads?


Or do I have wrong settings?

now a proper score:

stock, stock, and stock.

Note that the settings in the first screenshot here will make many AMD "BOOST" GPUs throttle, but the used settings for this test here will not.

 

[Wrigleyvillain]

Interesting...does that have something to do with downsampling? erocker run it at 1440P, please.

I can do the same with my 7950 as I am now overclocking it to it's max for quad hd myself.

 

[cadaveca]

I had wrong settings. But, power draw is much higher with the settings in the first score above. I use Valley for stability testing, and Ultra settings aren't good for that, so I had my own settings without realizing it.

I did edit the post, sry.

 

[d1nky]

@cadaveca is that the asus 7950?

 

[Mydog]

Not to bad I think

CPU @5,0 GHz
GPU's @1280/1700

 

[Exceededgoku]

Running an 6GB Sapphire 7970 and AMD FX8350:

Overclocked to 1149Mhz/6450Mhz

System spec is:

AMD FX8350 @ 4.6Ghz, 32GB RAM, 6GB Sapphire 7970 @ "stock" - 1100/6000 overclocked - 1149/6450, RAID0 6 x Sandisk Extreme SSD system drive.

AMD dragging it's feet is hardly surprising!

 

[LAN_deRf_HA]

@LAN DeRF HA - how is a gtx660x2 beating 3 7970's.... can you please check your settings and let me know please... thanks...

My settings are fine. It's all listed in the screenshots. I'm just a bit ahead of a Titan and that's where I should be.

 

[R00kie]

HD5870 ------ 1000/1300

 

[Fluffmeister]

Just tried a run @ 2560x1440, not too bad all things considered:

 

[LAN_deRf_HA]

Best I can do. 1267/7450

 

[Wrigleyvillain]

Oh sweet I missed this news when it was released and assumed was Heaven we were running here. So tired of that one and looks look nice too. Downloading now!

 

[purecain]

cool, im just updating the scores...

my maximus is away on rma and im using a stock system with a crappy low end gigabyte board....

I'm wondering how my system would of performed with the maximus board now.... :/

 

[LAN_deRf_HA]

Not to bad I think

http://www.techpowerup.com/forums/attachment.php?attachmentid=51060&stc=1&d=1368217957

CPU @5,0 GHz
GPU's @1280/1700

It's really weird you're higher than GamerGuy running 8x. I know he ran a very slightly higher res but on nvidia you lose 1000 points going to 8x. Maybe nvidia needs to optimize better for valley??

 

[Wrigleyvillain]

OMG this is easily my new favorite bench and probably one of my ultimate favorites ever! It really looks phenomenal on this display. Here's first run native res on high quality:

CPU 4.3
GPU 1200/1250

 

[purecain]

overclocked the gddr5 to 1770mhz and left the core at 1100mhz... giving me 2229 points...

@Wriggleyvillain could you please run at 1920x1080... Its interesting how much our cards all vary...

 

[rtwjunkie]

It took me a few days to get to it. Wow, I'm so ashamed. Stock GTX 580: GPU 772; Memory 2004; Shaders 1544. Still, you'd think with this motherboard and the cpu overclocked to 4Ghz it would be pushing more frames than this:

 

[Shou Miko]

since i like the Valley Benchmark from Unigine i decided to do a little run with my Asus GTX670-DC2-4GD5 overclocked a little ^^ (it boost to 1135mhz on the gpu)

only thing that brings me down a little is that it can't oc higher when i try the driver crashes so i guess i found the limit

or that was on my old IV Gene-Z/Gen3 board maybe my V Extreme will do better i dunno ^^;

 

[spectatorx]

Ok guys, here come my scores from my not so latest hardware but i think you all will be interested with them. As you all know valley is multiplatform benchmark using both APIs: dx and opengl so i decided to run benchmark in few modes.

At first a little bit about computer specification:
CPU: AMD Phenom II x4 B35 2,9@3,4GHz
RAM: 4GB DDR2
GPU: Gigabyte HD7790 OC 1GB
PSU: OCZ ZS550
Windows: 7 Enterprise SP1 x64
Linux: Ubuntu 13.04 amd64 (Kernel 3.8.0-19)
AMD Catalyst: Linux@13.4, Windows@13.5 beta 2

I wasn't taking screenshots, was just saving scores to htm files so i will type scores in here manually.

Every test done with extremeHD preset.

Linux@opengl:
FPS: 4.0
Score: 166
Min FPS: 2.2
Max FPS: 7.4

Windows@opengl:
FPS: 6.1
Score: 256
Min FPS: 2.2
Max FPS: 18.8

Windows@opengl (graphics card oced to gpu:1250/memory:1650):
FPS: 8.9
Score: 371
Min FPS: 2.6
Max FPS: 22.0

Windows@dx11:
FPS: 17.8
Score: 743
Min FPS: 6.7
Max FPS: 36.8

Windows@dx11 (oc: 1275/1650):
FPS: 20.8
Score: 872
Min FPS: 8.2
Max FPS: 43.3


As you see there is a difference in oc between opengl and dx11. Every time i tried to run benchmark in opengl with gpu at clocks higher than 1250 i was getting artifacts and after few scenes benchmark was crashing. In dx11 i clocked it to 1275MHz and still no artifacts nor benchmark crashing.

 

[The Von Matrices]

I'm actually really surprised at my (relatively) low score, and it is consistent across multiple runs. I think the benchmark is CPU limited because my score is about 500 points lower than systems with comparable graphics. Or else, something with my system is wrong. Does anyone have suggestions?

 

[GamerGuy]

Unlike Heaven which emphasize GPU, OCing your CPU results in negligible gains......OCing the GPU nets you better score. Valley seems to be quite reliant on CPU as well, my FX8350 @4ghz (stock) + 2x GTX670 SLi'ed nets me ~3300 only.....which is why I find the two GTX660 Ti score to be strange.

 

[Melvis]

My Mini PC (budget machine) run all at stock

Sorry I cant run 1080 but 1600*1200 isn't far off.

 

[Crap Daddy]

I say why not. The resolution is lower, native 1680*1050 / GTX570 at 825MHz

 

[LAN_deRf_HA]

Valley seems to be quite reliant on CPU as well, my FX8350 @4ghz (stock) 2x GTX670 SLi'ed nets me ~3300 only.....which is why I find the two GTX660 Ti score to be strange.

You know you were beaten by a Titan as well and that shouldn't happen. It must be cpu that's making the difference. That would explain Mydog's score being so high even with 8x. If this benchmark is that bad at isolating the graphics it's probably not the best choice for a benchmark. Granted sometimes games are cpu sensitize.

Might I suggest tpu get a proper Heaven 4.0 thread? If you want to test just the gpu that's it. Nothing is more consistent than that. I'd also suggest breaking the score up by resolution because people seem to have trouble running non-native res. That way every time someone uses a new res you make a new list and everyone competes against other people who can run that res. Define the other presets but leave res up the user.

Like this basically

 

[toilet pepper]

Sapphire 7650 1130/1450

 

[Arctucas]

560Ti SLi 950/2150 @ 1.050V