UNIGINE-Valley BENCHMARK scores

[geovolara]

hey guys this is my best stable OC so far with the SAPPHIRE R9 280 DUAL X @ 1116/1500

 

[EasyLover]

Here is my score of Valley under required criteria.

CPU: Intel i7 5930k @ 4.2GHz
GPU: SLI Asus Strix GTX 980 Ti @ 1652/1903
Score: 6132

 

[johnspack]

Yep dam, I guess my cpu is bottlenecking my gpu now, best I can get so far at oced settings:

 

[Vellinious]

Yep dam, I guess my cpu is bottlenecking my gpu now, best I can get so far at oced settings:

What do you get with the Extreme HD Preset? And what clocks are you running on your 970?

 

[Vellinious]

Did a couple of quick runs with clocks I knew would work.

Single card run, at 1591 / 4065 -- 3835:

SLI run, at 1581/4055 -- 6500:

 

[MrGenius]

3570K @ 4.8GHz + MSI 280X Gaming 3GB @ 1207/1850 = 2727

 

[trog100]

not fully benching maxed out.. just gaming stable..

tis win 10 as well.. valley seems to think its win 8..

trog

ps..

 

[johnspack]

What do you get with the Extreme HD Preset? And what clocks are you running on your 970?

2500.... boosting to 1517 +163 and ram at +204. See almost no cpu usage, no not quite sure why I'm so low.

 

[Mercennarius]

3570K @ 4.9 GHz + MSI 280X Gaming 3GB @ 1207/1850 = 2708
View attachment 68833 View attachment 68834

Wow highest I can get my MSI R9 280X is 2255 so far and clocks around 1175/1650. How did you get your ram clocked so high?

 

[Vellinious]

2500.... boosting to 1517 +163 and ram at +204. See almost no cpu usage, no not quite sure why I'm so low.

PCIe 2.0 lanes, maybe?

 

[johnspack]

Possibly, but just bumped up clock to 1525, it seemed to like that:

I'll try to up clocks a bit more.....

 

[Jborg]

3570K @ 4.9 GHz + MSI 280X Gaming 3GB @ 1207/1850 = 2708
View attachment 68833 View attachment 68834

Needs to be 8x AA

 

[Aquinus]

Needs to be 8x AA

Wrong benchmark. That's the Heaven thread that calls for 8x AA, 4x AA is used on this one.

Preset: Custom
API: DX11
Quality: Ultra
Stereo 3d: Disabled
Monitors: Single
Anti Aliazing: x4
Full screen: yes
Resolution: 1920x1080

 

[trog100]

valley dosnt scale that well in sli on my system with a stock bios.. the boost dosnt go up relative to higher core clocks ether.. it seems to max out around 1442 no matter how high i go on the core clock.. valley shows a higher 1680 but its not right the card isnt running at the speed valley says it is..

this is with one card running at +144 on the core clock and plus 340 on the memory.. with two cards running i get a score 6300.. maybe a 20% gain which is not a lot..

trog

 

[the54thvoid]

valley dosnt scale that well in sli on my system with a stock bios.. the boost dosnt go up relative to higher core clocks ether.. it seems to max out around 1442 no matter how high i go on the core clock.. valley shows a higher 1680 but its not right the card isnt running at the speed valley says it is..

this is with one card running at +144 on the core clock and plus 340 on the memory.. with two cards running i get a score 6300.. maybe a 20% gain which is not a lot..

trog

Check your OC software logs (the graphs you can detach) - I don't think one of your cards is running right. You should be a lot higher than that.

 

[Aquinus]

Check your OC software logs (the graphs you can detach) - I don't think one of your cards is running right. You should be a lot higher than that.

Agreed, that score sounds like only one GPU is actually running.

 

[Jborg]

Wrong benchmark. That's the Heaven thread that calls for 8x AA, 4x AA is used on this one.

Oh good call, I looked at the screenshot above my post, I see he posted both 4x and 8x, I made a quick assumption. Thanks for the correction.

 

[the54thvoid]

valley dosnt scale that well in sli on my system with a stock bios.. the boost dosnt go up relative to higher core clocks ether.. it seems to max out around 1442 no matter how high i go on the core clock.. valley shows a higher 1680 but its not right the card isnt running at the speed valley says it is..

this is with one card running at +144 on the core clock and plus 340 on the memory.. with two cards running i get a score 6300.. maybe a 20% gain which is not a lot..

trog

Agreed, that score sounds like only one GPU is actually running.

Double checked - 2nd GPU is not running - look at temp. Although Valley usually messes up clocks speeds, temps are normally accurate. If your 2nd GPU is at 24 degrees, it's not running.

 

[MrGenius]

Wow highest I can get my MSI R9 280X is 2255 so far and clocks around 1175/1650. How did you get your ram clocked so high?

It's got the Hynix chips rated @ 7000MHz. So I didn't have to do anything except crank it up another 400MHz past that. And/or as far as it will go. 1851 or higher causes immediate instability. As does anything from 1501 to 1849. It likes to run @ 1500(stock) or 1850. Which makes no sense to me. But that's fine. I'll take 1850 stable and be happy for sure.

BTW...the memory voltage on my card isn't adjustable(software wise). And I feel no need, currently, to hard mod it for more volts to see if that even makes a difference.

On your score...I'm on the list, the same card and my old E8600, with 2197. I guess vram speed accounts for much. And CPU core/thread count x speed probably even more so.

 

[trog100]

"Double checked - 2nd GPU is not running - look at temp. Although Valley usually messes up clocks speeds, temps are normally accurate. If your 2nd GPU is at 24 degrees, it's not running."

####

its running i have checked everything many times over.. he he.. temps is one of the things i keep a close eye and all other sensor readings..

there is something wrong with the valley benchmark.. a single card scores about what it should do (least i think it does) but the scaling with two is abysmal.. maybe only 20%..

heaven scales well enough up from 2600-ish with one card to 4600-ish with two.. valley only goes from 5400-ish to 6400-ish.. its just with valley.. even with only 50% scaling i should be looking at closer to 8000 not 6400..

i can only quote the results i get and when i get results i dont expect i do check them pretty well before opening my mouth..

its something i only noticed when decided to post valley results in this thread.. everything is as it should be.. the cards are boosting the same with one or two they are not clocking down due to temps or load.. the boost clock speed is around 1440 as it should be one card or two.. the only things that is odd is the rubbish 20% scale up when running two cards with valley..

trog

 

[Vellinious]

There's nothing wrong with the Valley benchmark. It works just fine in SLI runs. Your second card looks like it's idling.

SLI run

Single GPU run

 

[trog100]

i have shed a little light on my sli scaling problems.. its the 4 x AA mode.. with only 4 x AA i go from 5300-ish with one card to 6300-ish with two.. poor sli scaling..

with 8 x AA (the same as the heaven setting) the scaling makes a lot more sense.. 4300-ish to 6300-ish..

for me the 4 X AA setting isnt loading the two cards enough.. its ether a cpu limit or something else at work.. as for people who think i am dumb enough not to know that i have only one card running.. i cant really comment.. he he..

in sli mode my 4 x AA scores are the same as my 8 x AA scores.. with just the one card the AA setting makes a large score difference.. as to what is bottle necking things i havnt a clue but it sure aint me being dumb enough not to know whether or not i have one or two cards running.. he he

i would welcome some sensible suggestions as to what is bottle necking things with only 4 x AA.. i would think cpu but it only reads around 15% usage and certainly aint working hard.. memory usage is also very low valley dosnt use much of that..

i expect i will see exactly the same results in the heaven benchmark if i run that at 4 xAA instead of the 8 x AA i have always run it at..

trog

 

[the54thvoid]

i have shed a little light on my sli scaling problems.. its the 4 x AA mode.. with only 4 x AA i go from 5300-ish with one card to 6300-ish with two.. poor sli scaling..

with 8 x AA (the same as the heaven setting) the scaling makes a lot more sense.. 4300-ish to 6300-ish..

for me the 4 X AA setting isnt loading the two cards enough.. its ether a cpu limit or something else at work.. as for people who think i am dumb enough not to know that i have only one card running.. i cant really comment.. he he..

in sli mode my 4 x AA scores are the same as my 8 x AA scores.. with just the one card the AA setting makes a large score difference.. as to what is bottle necking things i havnt a clue but it sure aint me being dumb enough not to know whether or not i have one or two cards running.. he he

i would welcome some sensible suggestions as to what is bottle necking things with only 4 x AA.. i would think cpu but it only reads around 15% usage and certainly aint working hard.. memory usage is also very low valley dosnt use much of that..

i expect i will see exactly the same results in the heaven benchmark if i run that at 4 xAA instead of the 8 x AA i have always run it at..

trog

Nobody is accusing you of being dumb. Chill.

PCI lane bandwidth perhaps? Wizzards latest test of PCI-e shows that higher fps rate suffer under restricted bandwidth. Is the 2nd card at x8 or x4?

 

[Vellinious]

Nobody is accusing you of being dumb. Chill.

PCI lane bandwidth perhaps? Wizzards latest test of PCI-e shows that higher fps rate suffer under restricted bandwidth. Is the 2nd card at x8 or x4?

SLI doesn't operate at x4.

i have shed a little light on my sli scaling problems.. its the 4 x AA mode.. with only 4 x AA i go from 5300-ish with one card to 6300-ish with two.. poor sli scaling..

with 8 x AA (the same as the heaven setting) the scaling makes a lot more sense.. 4300-ish to 6300-ish..

for me the 4 X AA setting isnt loading the two cards enough.. its ether a cpu limit or something else at work.. as for people who think i am dumb enough not to know that i have only one card running.. i cant really comment.. he he..

in sli mode my 4 x AA scores are the same as my 8 x AA scores.. with just the one card the AA setting makes a large score difference.. as to what is bottle necking things i havnt a clue but it sure aint me being dumb enough not to know whether or not i have one or two cards running.. he he

i would welcome some sensible suggestions as to what is bottle necking things with only 4 x AA.. i would think cpu but it only reads around 15% usage and certainly aint working hard.. memory usage is also very low valley dosnt use much of that..

i expect i will see exactly the same results in the heaven benchmark if i run that at 4 xAA instead of the 8 x AA i have always run it at..

trog

Try putting more clock on the 4790k. 4ghz isn't much. I don't think it's going to solve the problem, because...honestly, I don't know why it would work in one program and not the other, but....it helps to run a "benchmark" clock on my CPU for Heaven, certainly, and a little bit for Valley too.

Have you tried to uninstall, and reinstall Valley?

 

[Aquinus]

I managed to get a run in at 1165Mhz/1640Mhz. It seems to be about the upper bound of what my 390 will handle with respect to video memory clocks.