Unigine Heaven 4.0 Benchmark Scores

[Wester71]

Relax....I'm sure they'll get updated when whomever does it, has time to. Not everyone lives, breathes and eats in front of their PC so they can make sure scores get updated every time someone posts.......my goodness.

really coming from Iowa

 

[Vellinious]

really coming from Iowa

Of the two of us, I'm not the one acting like an impatient child..... /wink

 

[n0tiert]

 

[GreiverBlade]

YES finally broke 2k point barrier, good thing that the multiple post and notification here reminded me to run it again

i5-6600K @ 4.4 and GTX 980 @ 1378(1479 boost)/1903 (boost logged by GPU-Z during the benchmark 1516.4mhz ) GPU didn't got higher than 49° funny enough i never got a reading of max fps high as that one before that run (on the phase where the camera orbits the flying island)

 

[Retrorockit]

I'm having a strange problem with Unigine benchmarks. My computer scores much slower with my CPU overclocked 3.72GHz. than at 2.93GHz base speed. In 3Dmark it scales up nicely in overall and physics scores. My O/C is in software, but it doesn't matter if it's on before the test loads, or changed afterwards. I'm using Throttlestop 6.00. I raised volts to see if it was instability but no change. In Heaven basic @2.93GHz. I get 99fps, 3.72GHz. I get 90fps. On higher settings, or in Valley it's even worse. Does anyone have any insight into this?

 

[rob cherry]

FX 6300 @ stock 3.5ghz
8gb (2x4gb) kingston hyperX savage @ 1600mhz
r9 290 4gb graphics 947mhz
memory 1250mhz
windows 7 home premium 64-bit

 

[tvamos]

i5-4670k @ 4.2, HD7970 @ 1200/1600

 

[GreiverBlade]

I'm having a strange problem with Unigine benchmarks. My computer scores much slower with my CPU overclocked 3.72GHz. than at 2.93GHz base speed. In 3Dmark it scales up nicely in overall and physics scores. My O/C is in software, but it doesn't matter if it's on before the test loads, or changed afterwards. I'm using Throttlestop 6.00. I raised volts to see if it was instability but no change. In Heaven basic @2.93GHz. I get 99fps, 3.72GHz. I get 90fps. On higher settings, or in Valley it's even worse. Does anyone have any insight into this?

9fps is what i call a standard frame variance ... and also CPU OC has little to do with Unigines Benchmarks ... so the culprit is not the CPU OC afaik

 

[Vellinious]

9fps is what i call a standard frame variance ... and also CPU OC has little to do with Unigines Benchmarks ... so the culprit is not the CPU OC afaik

Ooh, no true at all. CPU overclock makes a HUGE difference on Valley. Not so much on Heaven, but....disable all but 2 cores on an i7, and disable hyperthreading, to add a bit more overclock, can make quite the difference in Valley scores. Like I said, Heaven not so much, but.....it does help keep the minimums up, and will add a little bit on the top as well.

For instance. This was run with all 6 cores of a 5820k at 4.2ghz. 290X at 1303 / 1806

This was run with 2 cores, hyperthreading off with the clocks at 4.9ghz. 290X at 1303 / 1802.

I haven't done the same thing with Heaven, and while it's not as dramatic, it does still make a difference.

 

[GreiverBlade]

Ooh, no true at all. CPU overclock makes a HUGE difference on Valley. Not so much on Heaven, but....disable all but 2 cores on an i7, and disable hyperthreading, to add a bit more overclock, can make quite the difference in Valley scores. Like I said, Heaven not so much, but.....it does help keep the minimums up, and will add a little bit on the top as well.

For instance. This was run with all 6 cores of a 5820k at 4.2ghz. 290X at 1303 / 1806

This was run with 2 cores, hyperthreading off with the clocks at 4.9ghz. 290X at 1303 / 1802.

I haven't done the same thing with Heaven, and while it's not as dramatic, it does still make a difference.

ok then valley is ... CPU and GPU bound ... but we are in HEAVEN topic but thanks for the precision anyway

btw my 6600K at 4.4 has a higher minimal fps than my previous 4690K at 4.5

 

[Retrorockit]

9fps is what i call a standard frame variance ... and also CPU OC has little to do with Unigines Benchmarks ... so the culprit is not the CPU OC afaik

It's too consistent to be a variance, and in Valley it's from 50fps down to 40fps or 25%. CPU should have little to do with Heaven benchmark. It not only does, but in the opposite direction that one would expect. If it was the same I wouldn't be surprised, but the loss is hard to explain.

 

[Vellinious]

ok then valley is ... CPU and GPU bound ... but we are in HEAVEN topic but thanks for the precision anyway

btw my 6600K at 4.4 has a higher minimal fps than my previous 4690K at 4.5

Like I said, it's not as pronounced in Heaven, but, the core clock on the CPU does matter. If you're pushing for high scores, and not pushing the CPU for the unigine benchmarks, you aren't doing yourself any justice. = P

 

[Vellinious]

It's too consistent to be a variance, and in Valley it's from 50fps down to 40fps or 25%. CPU should have little to do with Heaven benchmark. It not only does, but in the opposite direction that one would expect. If it was the same I wouldn't be surprised, but the loss is hard to explain.

Are you sure your clock is stable? Does anything else change, like the bclk? Memory? The only thing I can think of, as to why it would go backwards would be an unstable OC.

 

[Retrorockit]

14x266 fsb all clocks are stock. Just Multi and volts. It runs 3x Prime95 OK, and other stress tests 100% on all 4 cores. It's very stable and runs Firestrike w/o problems scoring 6892 (58%). Firestrike is 6428 w/o O/C (47%). Benchmarking CPU temps. are 66-67*C. That's with Physics tests in 3Dmark.
In Unigine they don't go that high 63*C.
There's a validation link in my sig. for 4GHz. now that was unstable!

 

[Vellinious]

14x266 fsb all clocks are stock. Just Multi and volts. It runs 3x Prime95 OK, and other stress tests 100% on all 4 cores. It's very stable and runs Firestrike w/o problems scoring 6892 (58%). Firestrike is 6428 w/o O/C (47%). Benchmarking CPU temps. are 66-67*C. That's with Physics tests in 3Dmark.
In Unigine they don't go that high 63*C.
There's a validation link in my sig. for 4GHz. now that was unstable!

I really don't know. Could have to do with the age of the architecture? I really don't have a clue.....

 

[Retrorockit]

I just ran a series of Unigine benchmarks at all available speeds including underclocking to 2.66GHZ.
Speed Score FPS
3.72GHZ. 2303 91.4
3.45GHz. 2373 94.2
3.2GHz. 2431 96.5
2.93GHz 2495 99.0
2.66GHz 2377 92.4
Basically Q6700@ 2.66 GHz. is as fast in Unigine as an overclocked QX6800 @ 3.72 GHz.
I turned Unigine off, then reset speeds, then opened Unigine again each time.
At least it didn't get faster with an underclock!
I'll edit this in here.
Unigine Valley test.
speed.........score......fps
3.72GHz. 1911 45.7
2.93GHz. 2098 50.1
I guess it's the same 10% loss from 26% overclock.

 

[GreiverBlade]

Like I said, it's not as pronounced in Heaven, but, the core clock on the CPU does matter. If you're pushing for high scores, and not pushing the CPU for the unigine benchmarks, you aren't doing yourself any justice. = P

well i see what you mean ... still for me it does little to near nothing ... and it's known since quite a while ... even more than 4 core are not a game changer on that bench...

off topic: i still don't see what's going on for the QX6800 of @Retrorockit ... and that's bug me beyond all ... (and you remind me to assemble some S775 build sooner or later ... i still have a E8500 and a XFX 650i ultra sitting on the shelve ... jsut need to find some DDR2 and a batch of GPU to test ... )

OH FCK ... i just noticed that i've put the bench on high instead of ultra ...

DAMN! i have to rerun it ...

and i can't edit or delete the initial post ... ok, don't wait me ... i gonna hang myself in a corner of my livingroom, i'll be back later once i get my cervical back in place ...

 

[Vellinious]

well i see what you mean ... still for me it does little to near nothing ... and it's known since quite a while ... even more than 4 core are not a game changer on that bench...

also compare my 4690K @ 4.5 to my 6600K @ 4.4 result .... same GPU even a lil lower clock on the 2nd run ... altho the result get a boost and minimum fps is still higher (ok ,... i know the ipc or a skylake is better than a haswell ... even with 0.1ghz less )
View attachment 72028

off topic: i still don't see what's going on for the QX6800 of @Retrorockit ... and that's bug me beyond all ... (and you remind me to assemble some S775 build sooner or later ... i still have a E8500 and a XFX 650i ultra sitting on the shelve ... jsut need to find some DDR2 and a batch of GPU to test ... )

OH FCK ... i just noticed that i've put the bench on high instead of ultra ...

DAMN! i have to rerun it ...

and i can't edit or delete the initial post ... ok, don't wait me ... i gonna hang myself in a corner of my livingroom, i'll be back later once i get my cervical back in place ...

The unigine engine only utilizes 2 cores..... And, like you said.....the reason your 6600k at a slightly lower clock is doing better, is because it has a stronger IPC, on the 2 threads that it's running. Which is why it's handy to disable all but 2 cores, disable hyperthreading (on an i7), and overclock the living piss out of the remaining 2 cores.

Yeah, I have no idea what's going on with Retrocockit's build. I've never seen anything like that...

 

[GreiverBlade]

The unigine engine only utilizes 2 cores..... And, like you said.....the reason your 6600k at a slightly lower clock is doing better, is because it has a stronger IPC, on the 2 threads that it's running. Which is why it's handy to disable all but 2 cores, disable hyperthreading (on an i7), and overclock the living piss out of the remaining 2 cores.

Yeah, I have no idea what's going on with Retrocockit's build. I've never seen anything like that...

wrong the reason behind my 6600K is stated on the edit .... "high" instead of "ultra" .... a Skylake is not really a step above a haswell ... and specially with a 4.4 and a 4.5 difference clock ... you would not notice 10% increased performance between SL and HW even if they were at the same clock ... i should retract my previous post but i can't even delete it

and nope OC still does nothing for Heaven and 94pts more in valley is not what i call a huge difference, rather a almost non existent (almost in the "regular performance variance between 2 benchmark runs" case )
placebo effect ... much ...

just fo'laugh'n giggle i will try a 2 core OC on my 6600K (tho i am not sure of the accuracy of that method .... )

also Heaven is defined as 100% GPU bound by Unigine devs themselves CPU OC : does indeed near to nothing ... at last nothing worth a extreme OC (4.4ghz for me is a base OC ) funny enough Valley is also almost GPU exclusive bound ... so again here too a CPU OC produce a result not worth the work demanded behind

and lastly where did you hear that unigine engine only use 2 core? i would like the source if you have it because i can find any info on it more on it benchmark not engine since the engine seems to be scalable, aka it will surely use more core than 2 and even all core of a HEDT CPU if needed

quoted from their official website
"The engine detects the number of CPU cores available and distributes the load among all of them, thanks to an efficient multi-threaded design. As a result, the engine is capable of handling hundreds of thousands of dynamic objects on-screen simultaneously.​

SCALABILITY​

UNIGINE utilizes all hardware resources available in the most efficient way; it is scalable from integrated GPUs to top-tier multi-channel PCs.
UNIGINE supports up to the past five generations of GPUs with performance optimizations in the renderer for various vendors. There is also support for both consumer-level and professional GPUs (Quadro and FireGL series).
Our UNIGINE-powered GPU benchmarks are installed by millions of users worldwide and are considered one of the best applications available to utilize all GPU."​

(9fps loss between a OC and a stock is not really what i call a issue ...rater a non stable OC in Unigines benchies ... )

ok back to the topic ... my cervical are back in place...

 

[Retrorockit]

I can run 4X Prime 95 all day at 3.45GHZ. I don't know how stable it would need to be to improve on that.

 

[Vellinious]

wrong the reason behind my 6600K is stated on the edit .... "high" instead of "ultra" .... a Skylake is not really a step above a haswell ... and specially with a 4.4 and a 4.5 difference clock ... you would not notice 10% increased performance between SL and HW even if they were at the same clock ... i should retract my previous post but i can't even delete it

and nope OC still does nothing for Heaven and 94pts more in valley is not what i call a huge difference, rather a almost non existent (almost in the "regular performance variance between 2 benchmark runs" case )
placebo effect ... much ...

just fo'laugh'n giggle i will try a 2 core OC on my 6600K (tho i am not sure of the accuracy of that method .... )

also Heaven is defined as 100% GPU bound by Unigine devs themselves CPU OC : does indeed near to nothing ... at last nothing worth a extreme OC (4.4ghz for me is a base OC ) funny enough Valley is also almost GPU exclusive bound ... so again here too a CPU OC produce a result not worth the work demanded behind

and lastly where did you hear that unigine engine only use 2 core? i would like the source if you have it because i can find any info on it more on it benchmark not engine since the engine seems to be scalable, aka it will surely use more core than 2 and even all core of a HEDT CPU if needed

The engine detects the number of CPU cores available and distributes the load among all of them, thanks to an efficient multi-threaded design. As a result, the engine is capable of handling hundreds of thousands of dynamic objects on-screen simultaneously.​

SCALABILITY

UNIGINE utilizes all hardware resources available in the most efficient way; it is scalable from integrated GPUs to top-tier multi-channel PCs.
UNIGINE supports up to the past five generations of GPUs with performance optimizations in the renderer for various vendors. There is also support for both consumer-level and professional GPUs (Quadro and FireGL series).
Our UNIGINE-powered GPU benchmarks are installed by millions of users worldwide and are considered one of the best applications available to utilize all GPU.​

(9fps loss between a OC and a stock is not really what i call a issue ...rater a non stable OC in Unigines benchies ... )

ok back to the topic ... my cervical are back in place...

lol, if you say so....I'll just take my high scores for 970 and 290x and smile, cause I know my benchmarks. /wink

1739 - 970
1862 - 290X

 

[GreiverBlade]

I can run 4X Prime 95 all day at 3.45GHZ. I don't know how stable it would need to be to improve on that.

please make a thread about that issue elsewhere, it's not the place to discus of that.

and you can be stable in prime and not stable anywhere else.

there might be another issue underlying that doesn't concern only Unigines bench.

lol, if you say so....I'll just take my high scores for 970 and 290x and smile, cause I know my benchmarks. /wink

1739 - 970
1862 - 290X

mmhhh? ... it's not "if i say so" ... i wrote that my difference of "oh my god 35pts" was due to the use of "high" instead of "ultra" settings ... keep in mind that my GPU OC and CPU OC are lower ... and nope a 4.4 Skylake would not give a edge over a 4.5 Haswell ...specially with a slower clocked GPU in a GPU bound game.

tho ... answer my question: your source on the Unigines engine use only 2 cores... i'd like to know, monitoring?

and the end of your post has nothing to do with me indeed those are good score for a 970 and a 290X and only due to the GPU
i know my benchmark also ... in Valley or Heaven a 4.0ghz i7-920 produced the same result as a 2.66ghz... 94pts in valley is a minimal difference, in heaven it would be different but ... it's not the case, so if we scale the "huge difference" you get in Valley we actually reach the "it does near nothing" in Heaven.

 

[Vellinious]

please make a thread about that issue elsewhere, it's not the place to discus of that.

and you can be stable in prime and not stable anywhere else.

there might be another issue underlying that doesn't concern only Unigines bench.


mmhhh? ... it's not "if i say so" ... i wrote that my difference of "oh my god 35pts" was due to the use of "high" instead of "ultra" settings ... keep in mind that my GPU OC and CPU OC are lower ... and nope a 4.4 Skylake would not give a edge over a 4.5 Haswell ...specially with a slower clocked GPU in a GPU bound game.

tho ... answer my question: your source on the Unigines engine use only 2 cores... i'd like to know, monitoring?

and the end of your post has nothing to do with me indeed those are good score for a 970 and a 290X and only due to the GPU
i know my benchmark also ... in Valley or Heaven a 4.0ghz i7-920 produced the same result as a 2.66ghz... 94pts in valley is a minimal difference, in heaven it would be different but ... it's not the case, so if we scale the "huge difference" you get in Valley we actually reach the "it does near nothing" in Heaven.

If you watch the cores in Unigine while it's running, it's only ever loading up 2 at a time. If you go in and disable all but 2, kill hyperthreading, and overclock the crap out of the remaining 2 cores, it helps to score higher. Adding any more cores than that doesn't do anything. Anything less, lowers the scores a bit.

You believe whatever you want, man. I'm done arguing with ya. You sprinkle your magic dust to have a good run, and I'll sprinkle mine. = D

G'day

 

[krazytoneride]

i5 4690k @ 4.4 GHz, 2 x GTX 680 @ 1200/3004 MHz

 

[Retrorockit]

I found a way to get my O/C recognized by Unigine. If I reboot I need to apply the O/C in Windows, but with a restart it's applied already.
I got 115FPS, and 2902 score in Heaven Basic. About the 15% gain I'd expect from a CPU limited computer overclock.. If I check in CPUZ it's there, But if I open the overclock software to monitor temps. etc, it slows back down. I guess Unigine sets it's real time clock from old data at boot and not by current speed.
In Valley I'm getting 58.9 fps in basic, and 50.5 in Extreme.

In Heaven Extreme full screen I'm getting 59.2 FPS. Seems about right for my hardware.