Time Spy Score

[D@nte]

http://www.3dmark.com/3dm/13210550?
2790 , 2654 GPU , 3935 CPU
GTX Titan(stock),Core i7 3770k(stock),16 GB Ram

[The Keeper]

I win (for now).

Score: 9,886
Graphics: 11,152
CPU: 6,017
http://www.3dmark.com/spy/22154

2x GTX 980 Ti's, i7 3930K @4.6GHz. My scores might be a little lower than possible, since two graphics cards usually scored lower on a g-sync display with FireStrike, but less so on higher resolutions. Not that this likely matters, but this is using drivers from May.

One thing noticed is in the run details, it shows "SLI/Crossfire:" with either On or Off as options. Does this mean this benchmark isn't using DX12 multi-GPU support? If so, that's a bit disappointing, since theoretically, DX12 multi-GPU support should produce better scaling and allow for different GPU combinations.

[Vellinious]

980ti @ 1562 / 2140 | 5820k @ 4.625
Overall Score: 6608
GPU Score: 6489
CPU Score: 7375

http://www.3dmark.com/spy/22710

[UL_Jarnis]

Ursprünglich geschrieben von St. Ramsay:

I win (for now).

Score: 9,886
Graphics: 11,152
CPU: 6,017
http://www.3dmark.com/spy/22154

2x GTX 980 Ti's, i7 3930K @4.6GHz. My scores might be a little lower than possible, since two graphics cards usually scored lower on a g-sync display with FireStrike, but less so on higher resolutions. Not that this likely matters, but this is using drivers from May.

One thing noticed is in the run details, it shows "SLI/Crossfire:" with either On or Off as options. Does this mean this benchmark isn't using DX12 multi-GPU support? If so, that's a bit disappointing, since theoretically, DX12 multi-GPU support should produce better scaling and allow for different GPU combinations.

"SLI on/off" is just SystemInfo asking the driver if it has that enabled or not during SystemInfo scan. Has actually nothing do with the mGPU mode used in 3DMark Time Spy.

I actually don't know what NVIDIA driver does if you disable SLI in drivers and then try to run DX12 linked adapter mGPU mode. You could try :)

Time Spy simply tells the driver to render the work using all available GPUs in linked node mode.

Note that current drivers actually have some interesting issues with the DX12 mGPU and full screen mode and in some cases you can see higher performance in Windowed Mode than in the default fullscreen mode... so I'd venture a guess that the drivers still have some ways to go with the DX12.

[The Keeper]

Ursprünglich geschrieben von FM_Jarnis:

Ursprünglich geschrieben von St. Ramsay:

I win (for now).

Score: 9,886
Graphics: 11,152
CPU: 6,017
http://www.3dmark.com/spy/22154

2x GTX 980 Ti's, i7 3930K @4.6GHz. My scores might be a little lower than possible, since two graphics cards usually scored lower on a g-sync display with FireStrike, but less so on higher resolutions. Not that this likely matters, but this is using drivers from May.

One thing noticed is in the run details, it shows "SLI/Crossfire:" with either On or Off as options. Does this mean this benchmark isn't using DX12 multi-GPU support? If so, that's a bit disappointing, since theoretically, DX12 multi-GPU support should produce better scaling and allow for different GPU combinations.

"SLI on/off" is just SystemInfo asking the driver if it has that enabled or not during SystemInfo scan. Has actually nothing do with the mGPU mode used in 3DMark Time Spy.

I actually don't know what NVIDIA driver does if you disable SLI in drivers and then try to run DX12 linked adapter mGPU mode. You could try :)

Time Spy simply tells the driver to render the work using all available GPUs in linked node mode.

Note that current drivers actually have some interesting issues with the DX12 mGPU and full screen mode and in some cases you can see higher performance in Windowed Mode than in the default fullscreen mode... so I'd venture a guess that the drivers still have some ways to go with the DX12.

Thanks for the quick response! Good to hear that this test is actually using DX12 mGPU. I'll definitely play around with the test in various settings, as I did with Fire Strike. Hopefully I can break an overall of 10,000 with my current system with some tweaking and overclocking.

[ReflexShift]

GTX 1080, i7 5820k 4.0 ghz http://www.3dmark.com/spy/24949 score: 7302
Crossfire R9 Fury X, i7 5820k 4.0 ghz http://www.3dmark.com/spy/22325 score: 9065

[J๏нη Ꮤɨ¢Ꮶ™]

http://www.3dmark.com/3dm/13229573?



Graphics Score
6 855

CPU Score
4 420

[RyuHazuki]

Total score:5997
Graphic score 7207
CPU score 3074

Asus roc 1080 oc
i5 4670 stock
8 GB ram

Edit: http://www.3dmark.com/spy/27088

[TG SHINODA]

http://www.3dmark.com/3dm/13233014?

all stock, validation warning due to 362.00 drivers not approved

[UL_Jarnis]

I strongly recommend using something newer than 362.00 - those have some known bugs related to DX12.

[Jonny2Bad]

http://www.3dmark.com/spy/14914
3DMark Score7924
Graphics Score8045
CPU Score7306
crossfire 290's 1122/1454
5820k@4.5
32g ram @3000

[no3y3h4nd]

score: 7282

http://www.3dmark.com/3dm/13233470?

i7 4770K @ 4.2
32GB DDR3 @ 2133
2 x GTX 980 SC @ stock overclocks.

[AzzA]

http://www.3dmark.com/3dm/13238315

Total - 5061
GPU - 5633
CPU - 3125

SLI 780 @ 1.18ghz
i7 3820 @ 4.89ghz

Good timing with this test im about to rip these out and replace them with a water cooler 980ti
I'll post a comparision later

One Last run for good luck

http://www.3dmark.com/3dm/13241040?

Total - 5492
GPU - 6639
CPU - 2776

SLI 780 @1.3ghz

Not too sure what happened to the CPu score there , i did notice on the last run it was only detecting and using 8 gig of ram , reseated the modules and got 16gb back and the score dropped ??? Anyways not caring too much about that draining this pc and swapping the cards now :)

[wayne_959]

Score 6575 = http://www.3dmark.com/3dm/13210531

crossfire R9-290 / e5 2690 @3.3Ghz / 32GB DDR3 @ 1600Mhz
All @ stock stettings

[The Keeper]

New score with up-to-date Geforce drivers and on a non-g-sync display.

Score: 10,462
Graphics Score: 11,996
CPU Score: 6,068

GPU(s): 2x GTX 980 Ti's 1518MHz/1,841MHz
CPU: i7 3930K @ 4.6 GHz

Using a g-sync display even with g-sync disabled, expect a 5-10% performance hit when using multiple GPUs. This is something I've documented for quite some time, which applies not just to benchmarks, but also games. For the highest score possible, make sure you're plugged into a non-g-sync display and not plugged into any g-sync displays. Yes, this is a thing and I'd like to see if free-sync has this issue. I have a feeling this is just an Nvidia thing.

[Yonix]

I get a nice 7.231 with a Intel Core i7 4790K (@4.7GHz) and a Palit GeForce GTX 1080 GameRock 8GB (@2.063MHz/11.016MHz).

http://www.3dmark.com/3dm/13205222?