DX12 Time Spy Benchmark Thread

[Silent_Scone]

3DMark Time Spy Benchmark Thread


Many games now offer support for DX12 in one way or another. One of the most talked about features within the last two years has been Asynchronous Compute, which TimeSpy implements heavily.

The term Async Compute has become a coined marketing phrase, encompassing DX12 and its granular control of queues and command lists for execution in parallel. With Turing, independent integer and floating-point calculations result in a large boost to performance using these parallel workloads, so now is a good time to compare with previous generations of GPU to see how things stack up.






Future Ray Tracing Support

https://benchmarks.ul.com/news/watch-our-new-directx-raytracing-tech-demo

Raytracing is not a new technique, but until recently it has been too computationally demanding to use in real-time games.
With modern GPUs, it's now possible to use rasterization for most of the rendering and a smaller amount of raytracing to enhance shadows, reflections, and other effects that are difficult to achieve with traditional techniques.
Our DXR tech demo runs in real-time on a single current-generation GPU. As it builds on existing methods, it was relatively easy to implement into our DirectX 12 game engine.
We are proud to be one of the first developers chosen to work with DirectX Raytracing and excited about the opportunities for this new API.
We are happy to announce that we will be using DirectX Raytracing in a new 3DMark benchmark test that we hope to release towards the end of the year.

Submit your score

The rules are simple, please post your a link to your validated 3DMark Score along with the following information:

Standard or Extreme Test
CPU - Model & Overclock
GPU - Model and Boost Clock
Driver Version
Validated 3DMark URL & screenshot
Optional - Memory Speed & Timings


7900X @ 4.7---1080Ti Strix OC @ 2025MHz---4000MHZ----16-17-17-41 1T
Time Spy Standard
Driver 411.63
https://www.3dmark.com/3dm/28856954?










Time Spy Standard - 1 GPU

1. DooRules----18280---7980XE @ 5.1GHz----2080Ti @ 2130MHz/2050MHz---4000MHz---16-17-16-36-1T---Time Spy Standard---Driver *---Link

2. CN7----17752---7960X @ 5.3GHz----2080Ti @ 2235MHz/8000MHz---4000MHz---15-16-15-35-1T---Time Spy Standard---Driver *---Link

3. Silent Scone----16207---7900X @ 4.7GHz----2080Ti @ 2130MHz/8000MHz---4000MHz---16-16-16-41-1T---Time Spy Standard---Driver 411.70---Link

4. Rob W.----15870---7980XE @ 4.9 GHZ----TITAN V @ 2137MHz/1030MHz---3600MHZ --- *---Time Spy Standard---Driver 416.16---Link

5. Menthol----14955---8700K @ 5.3 GHZ----2080Ti @ 2040MHz/7700MHz---4266MHZ --- 17-18-18-38 1T---Time Spy Standard---Driver 411.63---Link

6. ThrashZone----11639---7900X @ 4.5 GHZ----Titan Xp @ 2050MHz/1628MHz---3200MHZ --- 16-18-18-38 2T---Time Spy Standard---Driver 416.16---Link

7. Eclipt ----11192---9700K @ 4.9 GHZ----2080 Strix OC @ 2070MHz/7450MHz---3200MHZ --- 16-18-18-36 2T---Time Spy Standard---Driver 416.---Link

8. tunc8686----10588---8700K @ 5.2 GHZ---- Strix 1080Ti OC @1790MHz/*MHz---3300MHZ --- 16-18-18-36 2T---Time Spy Standard---Driver 411.70---Link

9. ThrashZone----10498---5930K @ 4.6 GHZ----1080Ti ftw3 @ 2050MHz/1621MHz---3200MHZ --- 14-14-14-34 1T---Time Spy Standard---Driver 397.64---Link

10. Nate152----10445---7700K @ 5.2 GHZ----TITAN Xp @ 2038MHz1628MHz---300MHZ --- 15-15-15-35 1T---Time Spy Standard---Driver 399.64---Link

11. JPMBoy----8465---8086K @ 5.2 GHZ----1070Ti @ 2100MHz/2250MHz---3200MHZ --- 16-16-16-38 1T---Time Spy Standard---Driver 399.24---Link

12. x-rated----7386---9900K @ 5.0GHZ----2080Ti @ 2190/1945MHz---4200MHZ --- 18-18-18-36 2T---Time Spy Standard---Driver 417.71---Link

Time Spy - 2 GPU

1. JustinThyme----23257---9940X @ 4.8 --- 2080Ti Strix OC @ 2085MHz/1170MHz--- 38000MHz 19-19-19-39 2T---Time Spy Standard---Driver 391.35---Link





Time Spy Extreme- 1 GPU

1. JPMBoy----8464---7980XE @ 4.8 --- Titan V @ 2092MHz/1067MHz--- 4000MHz 16-17-17 1T---Time Spy Standard---Driver 391.35---Link

[Silent_Scone]

jpmboy wrote:
for sure. lower rez, more cpu bound. IDK, I'm itching for a pair of Titan Ts... tho the rtx6000 does not seem to be anything to get real excited about according to the comp guys, but the folks I know in the SFx business are snapping them up!

Yes, I know for a fact the 2080Ti is selling, but as per previous gen the cut SMs point to a further card down the line, or possibly just poor yields.


https://www.3dmark.com/spy/4570052 with 380W bios and quick run with shunt on 6-8 pin (EK water block). Sadly, this reference board taps out on ambient around 2130MHz, where others can maybe do 2165MHz or slightly more.

Outside of benchmarking, I'd expect most cards to tap out around the same for stability, though.

[jpmboy]

Hard to judge contemporary GPU performance with timespy (vs ts extreme)... scores rank with CPU overclock with the same GPU at below 4K with these GPUs. Kinda like comparing the single core performance of an 8700K or 7700K with a 7980XE or 7900X. eg, gaming perf =/ benching numbers. 🙂

[Silent_Scone]

jpmboy wrote:
Hard to judge contemporary GPU performance with timespy (vs ts extreme)... scores rank with CPU overclock with the same GPU at below 4K with these GPUs. Kinda like comparing the single core performance of an 8700K or 7700K with a 7980XE or 7900X. eg, gaming perf =/ benching numbers. 🙂

The standard test is more indicative of GPU score than CPU score overal. Look at my 7900X compared with most of the XE in the top 10. Extreme CPU test is far more multi thread intensive. GT 1 and 2 are certainly not CPU limited, either

[jpmboy]

Silent Scone@ASUS wrote:
The standard test is more indicative of GPU score than CPU score overal. Look at my 7900X compared with most of the XE in the top 10. Extreme CPU test is far more multi thread intensive. GT 1 and 2 are certainly not CPU limited, either

... nah. 😛

lol - But sure, overall score weighs physics quite a bit. More threads with the same gpu drives the overall ( via physics score), but with the same cpu and gpu, lower rez is basically a cpu benchmark, in all these benches, FM, Unigine, SquareE... etc. Thread use during the GT portions is not 36 with the XE (this I know) and I'd be surprised if it maxes out the 7900X's 20 threads during the graphics tests in the lower rez settings. 2 or 4 threads depending on the core priority last I looked. If ya just look at the graphics scores - even this, within the same GPU class ranks with cpu frequency (barring LN2 on the GPUs). Lot's of data in the OCN Time Spy threads - actually more than in the HOF since only has the top 100.

Anyway, I find FS Ultra and the 8K tests in Superposition really level the field to the GPU performance... until we have a good ray trace benchmark.

I'm still surprised that the 2080Ti is not just crushing the Titan V. I really hope a Turing Titan is launched (with SLI capabilities - please).

[Silent_Scone]

jpmboy wrote:
... nah. 😛

lol - But sure, overall score weighs physics quite a bit. More threads with the same gpu drives the overall ( via physics score), but with the same cpu and gpu, lower rez is basically a cpu benchmark, in all these benches, FM, Unigine, SquareE... etc. Thread use during the GT portions is not 36 with the XE (this I know) and I'd be surprised if it maxes out the 7900X's 20 threads during the graphics tests in the lower rez settings. 2 or 4 threads depending on the core priority last I looked. If ya just look at the graphics scores - even this, within the same GPU class ranks with cpu frequency (barring LN2 on the GPUs). Lot's of data in the OCN Time Spy threads - actually more than in the HOF since only has the top 100.

Anyway, I find FS Ultra and the 8K tests in Superposition really level the field to the GPU performance... until we have a good ray trace benchmark.

I'm still surprised that the 2080Ti is not just crushing the Titan V. I really hope a Turing Titan is launched (with SLI capabilities - please).

Strange how the HOF isn't dominated by 8700K, in that case. FS Ultra is good if you're strictly testing DX11 performance.

Don't forget GV100 is a massively powerful GPU. I guess from NVIDIA's point of view, it makes Turing look appealing at the price point lol. https://www.3dmark.com/compare/spy/4574376/spy/3355469

[jpmboy]

Silent Scone@ASUS wrote:
Strange how the HOF isn't dominated by 8700K, in that case. FS Ultra is good if you're strictly testing DX11 performance.

Don't forget GV100 is a massively powerful GPU. I guess from NVIDIA's point of view, it makes Turing look appealing at the price point lol. https://www.3dmark.com/compare/spy/4574376/spy/3355469

yep,
that's my point - the physics score weighs too heavily. I usually check the graphics score HOF for GPU contribution (see the pic attached). But either way bro, it seems the 2080Ti is performing as NV expected - killin the GPU scoreboard! Good you snagged two. I cancelled my preorder, wrongly influenced by prelims on the RTX6000.
That said, I may grab a couple if stock ever returns... IDK, in my business launch a product and can't make stock = trashed company stock even if it is one of a 1000 products you sell.

[Silent_Scone]

jpmboy wrote:
yep,
that's my point - the physics score weighs too heavily. I usually check the graphics score HOF for GPU contribution (see the pic attached). But either way bro, it seems the 2080Ti is performing as NV expected - killin the GPU scoreboard! Good you snagged two. I cancelled my preorder, wrongly influenced by prelims on the RTX6000.
That said, I may grab a couple if stock ever returns... IDK, in my business launch a product and can't make stock = trashed company stock even if it is one of a 1000 products you sell.

Prelim leaks on the 2080Ti were vastly based on cards that were heavily power limited, the difference can be staggering depending on the game/benchmark. Hoping the second card does a bit more in the frequency department!

[tunc8686]

8700K @5,2Ghz / 1,380V
32 Gb Corsair Dominator 3300Mhz 16-18-36
Asus Rog Strix 1080Ti OC @1790
-------------https://www.3dmark.com/3dm/29043745-------------

[Silent_Scone]

Nice, solid graphics score for a 1080Ti. 🙂

[Rob_W_]

latest bios and nv driver, that seems as far as I can go on water.
due to temps it keeps clocking back, hitting 45deg,

tested titan in different slots!, no real difference in scores.