New Unigine Superposition VR Benchmark Available

[Zarathustra[H]]

Canned benchmarks have never been our way of determining performance here at the [H], as we feel they rarely offer much insight into real world game performance which is what we should all care about, but Unigine has a new VR enabled benchmark out which really looks impressive.

If you'd like to try it out for yourself, you can nab it here. Guru3d also has some preliminary performance results running the benchmark. Thanks to forum member Revdarian for bringing the Guru link to our attention.

While this may not be a good predictor of how your system will perform in VR titles today, just like their past benchmarks, as well as the classic 3DMark benchmarks it is a nice tech demo of what modern systems are capable of.

KEY FEATURES

• Top-notch visuals powered by UNIGINE 2 Engine
• Extreme hardware stability testing
• GPU temperature and clock monitoring
• Unique SSRTGI (Screen-Space Ray-Traced Global Illumination) dynamic lighting technology
• VR experience (Oculus Rift and HTC Vive)
• Free exploration mode with mini-games
• Over 900 interactive objects
• Global leaderboards integration
• Support for Windows and Linux

 

[heatlesssun]

Tried this out last night. At top settings this thing will crush any current GPU. Under DX mode, I assume 11, a single 1080 Ti overclocked @ 120/400 only manages about 8 FPS average at 4k at max settings. You have to pay to get the VR mode though the benchmark will run on screen.

 

[0x4452]

Glad to see it is DX11. For me that is a sign of an independently developed benchmark, not sponsored by M$ or AMD.

Unigine are clearly brilliant programmers achieving very high triangle counts with modest CPU usage.

 

[0x4452]

And big thumbs up for Linux support.

 

[pendragon1]

megalith posted aboot this in jan and there is a thread here: https://hardforum.com/threads/unigi...ing-out-their-new-hardware-detection.1923337/
oh and its not just a VR benchmark...

 

[FrgMstr]

 

[FrgMstr]

yup it was posted by ocellaris last night in the other thread. there just seems to be a lot of duplicate threads goin on lately....

This is the Front Page News forum. When we post an news item on the front page news, we post it in here as well. If you feel as though our moderation standards for the forum are not up to your standards, I would ask that you use the REPORT POST button that we supply on every post. That way you can draw our attention to it instead of posting off topic in our FPN threads. Thank you.

 

[FrgMstr]

It seems that this bench doesn't read FSB OCing either just like 3DMark.
View attachment 21704

My 8350 is at 4.74Ghz with 237FSB @20 multi. Gotta say this has to be the most stable bench I have seen for graphics, as in no wild swings.

it just reads the cpu label...mine is not correct either

 

[heatlesssun]

4k optimized +125/400 on the 1080 Tis, no SLI support in this yet:

 

[Sparky]

 

[azrael.arach]

Anding mine in.

Card was at +150/+425. Cpu is at 4.512ghz. The card seemed to sit between 1987 and 1940 boost clocks during test.

 

[Supercharged_Z06]

It seems those with SLI aren't getting any love with this benchmark yet... Damn but it is pretty at 4K. I'm also thinking I'll definitely skip upgrading to the Titan Xp variant for that extra 1.5% bump in overall performance. I guess water helps... max temp of 30C.

 

[JustReason]

It seems those with SLI aren't getting any love with this benchmark yet... Damn but it is pretty at 4K. I'm also thinking I'll definitely skip upgrading to the Titan Xp variant for that extra 1.5% bump in overall performance. I guess water helps... max temp of 30C.

View attachment 21733

Yours is the first I have noticed only hit 99% on GPU utilization.

edit add: Never mind seems the 4K ones show only 99%. Most of all the 1080p show 100%. Weird

 

[azrael.arach]

People should post overclocks as I am going back to edit mine now.
Guess time to move to water on the card being it is just a FE version.

 

[JustReason]

People should post overclocks as I am going back to edit mine now.
Guess time to move to water on the card being it is just a FE version.

Not sure the CPU matters much but the GPU clocks would go miles for reference. A guy with a 390X and a 4790K got just a tick higher than me with a 290 and FX8350.

 

[SvenBent]

Im kinda disapointed about how badly it scales across threads
1 miam thread and 1 secondary thread (1/3 the load of main) and thats about it. the rest do miniscule amount of work

 

[castlefire]

 

[JustReason]

Im kinda disapointed about how badly it scales across threads
1 miam thread and 1 secondary thread (1/3 the load of main) and thats about it. the rest do miniscule amount of work

Unigine engine is almost entirely GPU. They haven't ever really stressed the CPU and I gather that they don't want to as they want to get a full record of GPU abilities with out relying on CPU capabilities as well. This is a true GPU bench.

 

[Budwise]

1080 @ 2050ish core 1250 memory and 7700K at 4.8

 

[JohnnyGatt]

It looks amazing even on medium on my budget PC with the RX480. will be interesting to see what we get in VR on the engine the benchmark has a lot more going on than any game I have seen in VR.

 

[heatlesssun]

It looks amazing even on medium on my budget PC with the RX480. will be interesting to see what we get in VR on the engine the benchmark has a lot more going on than any game I have seen in VR.

View attachment 21756

I'd love to see the VR results as well but not enough for $20.

 

[deaedius]

 

[SvenBent]

Unigine engine is almost entirely GPU. They haven't ever really stressed the CPU and I gather that they don't want to as they want to get a full record of GPU abilities with out relying on CPU capabilities as well. This is a true GPU bench.

Not much of a GPU bench when you get CPU bottlenecked because they cant figure out how to distribute load. If it was to be a true GPU bencmhark they would ensure a better CPU utilization so its not becoming the determining part.

 

[JosiahBradley]

It warned me I shouldn't even attempt 4K optimized due to VRAM limitations, but it looked great and stayed around 30fps. 4.8Ghz/1.1Ghz (Edit: Added 8K just because why not?)

 

[umeng2002]

yuge performance impact between 1080p high and 1080p extreme... at least on a 970.

 

[JustReason]

Not much of a GPU bench when you get CPU bottlenecked because they cant figure out how to distribute load. If it was to be a true GPU bencmhark they would ensure a better CPU utilization so its not becoming the determining part.

How do you know there is a CPU bottleneck? The framerates are very tight from min to max and CPU utilization never maxes that one or 2 cores. My frame rates of mid 20s with max of 30s is guaranteed not to be CPU limited. Look at the plathora of CPUs used by members here with the same GPUs and how close their results are, with variations due to clocks on GPU.

 

[Dadebraafsie]

it looks nice , but i like more of those nature/island tests
ive got like 15000 on 1080 medium with 1080 and 3770k on stock

 

[Dadebraafsie]

yuge performance impact between 1080p high and 1080p extreme... at least on a 970.

indeed a huge impct , but i like to know if anyone can see a visual difference between those 2 :s

 

[JustReason]

Yours is the first I have noticed only hit 99% on GPU utilization.

edit add: Never mind seems the 4K ones show only 99%. Most of all the 1080p show 100%. Weird

Ok now mine doesn't follow the norm.

I have 100% GPU utilization on 4K whereas most I have seen are 99%. confusing.

And for kicks:

So going to go with a hunch here and say I AM NOT 8K ready, lol.

 

[magictoaster]

 

[noko]

For SLI seems like the VR tests seems to used both 1070's the most:

 

[SvenBent]

How do you know there is a CPU bottleneck? The framerates are very tight from min to max and CPU utilization never maxes that one or 2 cores. My frame rates of mid 20s with max of 30s is guaranteed not to be CPU limited. Look at the plathora of CPUs used by members here with the same GPUs and how close their results are, with variations due to clocks on GPU.

I know it from seeing the threads CPU usage. Off cause its dependent on the system. but some people have a CPU bottleneck in it on their ryzen system.
Looking at Core utilization is an not a good way to see corespeed CPU bottleneck. Just as bad as when Kyle looked at Core utilization to establish how threaded a game is.

let me give you an example of why. this is a theoretical one thread running that is maxed by core speed aka its running 100% because it eats all it can take from the CPU

Now if we looked at it the way you do we would just see.:
0% on one core and 12.5% on six of the cores. and one core with 25% usage.
In your way you would think it is not bottlenecked when indeed it is.

Is the same mistake that 99% of people do when they try to see if their CPU is bottlenecked or not.
you need to look it from the threads viewpoints not the (logical) cores.


now i got the distribution among the thread by measuring the total amount of cycles that each thread has been running and it clearly shows only 2 threads with really load on them. The main had 3 time as many cycles as the second.
The rest had around 1/3 of the second or even less then that
So this piece of software is never going to benefit from much more than just 2 cores


and yes in these cases there was not much difference between different graphics cards.

 

[ChoGGi]

1080 @ 2025-50/5355 | 5930K @ 4.4

 

[Nenu]

6700K @ 4.6GHz
1080ti FE with Accelero III cooler, 2000/5940 clocks during bench

 

[Sparky]

GPU Clock Offset: +188
MEM Clock Offset: +785
Stock cooling

 

[Wysockisauce]

[/IMG]

3770k @ 4.5
1080ti @ +160/450

 

[mbze430]

I can't get the VR Ready? benchmark to run... I have a Rift but all it does is start and quit

Here are my 4k benchmark result

 

[LeeU]

Hi all,
Newbe here so here is mine.
Lee