Unigine Heaven 4.0 Benchmark Scores

[MxPhenom 216]

Nvidia cards have ALWAYS been better with tessellation, and this is a tessellation heavy benchmark, that's why the Green Team is doing better across the board.

 

[freeleacher]

3DMark(2013) is thought to favour neither brand, if not a little AMD skewed. Go and bench on that.

http://www.techpowerup.com/forums/threads/3dmark-2013.179767/

All ready have

 

[HammerON]

New Table

**Updated**

Cool

 

[GreiverBlade]

New Table

**Updated**

thanks looks nice!

 

[devilpepper]

Hi, everyone

I just replaced my 2x GTX 680s with a Gigabyte GTX 780Ti Windforce OC card and I have to say I am pretty happy with it. Just ran a quick bench with a very mild overclock on the card.

 

[devilpepper]

Bumped up the CPU and the gfx memory a bit but didnt make much of a difference ...well I guess better than nothing ;-)

 

[the54thvoid]

Thought I'd have another go, see if a few more micro volts will help.

I had to run the bench and then re-run with all the settings and filed score. It's quite awkward to screenshot for some folk - it doesn't work

What's weird is that my clocks were set to be 1227MHz but Afterburner OSD reports it as 12MHz lower each increment i tried. But Nvidia Inspector and GPU-z both report correctly as 1227MHz.

So, GTX Titan at 1227MHz core, 1752MHz (7008) memory. Core i7 3930k at 4.4GHz.

I was using the voltage hack on afterburner and was set up as 1.263v but with vdroop (or LLC) it was fluctuating between 1.231-1.238v (0.025-0.032 vdroop?). Core temp only got to 43max after 3 consecutive runs. I figure my VRM's should be quite cool as well. I tried 1240MHz but it crashed to a system restart. Wonder if i plugged in 1.3v (1.268v with LLC) I could get closer to 1300MHz?

 

[GreiverBlade]

Thought I'd have another go, see if a few more micro volts will help.

I had to run the bench and then re-run with all the settings and filed score. It's quite awkward to screenshot for some folk - it doesn't work

What's weird is that my clocks were set to be 1227MHz but Afterburner OSD reports it as 12MHz lower each increment i tried. But Nvidia Inspector and GPU-z both report correctly as 1227MHz.

So, GTX Titan at 1227MHz core, 1752MHz (7008) memory. Core i7 3930k at 4.4GHz.

I was using the voltage hack on afterburner and was set up as 1.263v but with vdroop (or LLC) it was fluctuating between 1.231-1.238v (0.025-0.032 vdroop?). Core temp only got to 43max after 3 consecutive runs. I figure my VRM's should be quite cool as well. I tried 1240MHz but it crashed to a system restart. Wonder if i plugged in 1.3v (1.268v with LLC) I could get closer to 1300MHz?

nice, btw is it not millivolt ? mV microvolts should be µV (i know im nitpicking)

 

[MalcomXT]

Hello ,

FX-8350
HD7970 3GB
i could get more score with higher GPU clocks but then artifacts was seen like red and green lights random show so 1180/1680 is the maximum stable OC for my HD7970.

 

[Paladone]

Sorry, I should have mentioned this but my CPU was (and is) clocked at 4.4GHz during the test

 

[Paladone]

Hello All,

I just finished my new build. Here is my information. I was not able to print screen while it was running (not sure on boost clock), but everything is stock.

I don't understand how you got such a higher score than I did, my 4930k is at 4.4 vs. 4.0GHz and and both my cards were overclocked :O Nice score none the less.

EDIT: Are you sure that you had Anti-Aliasing enabled and on x8? I disabled my Anti-aliasing and got a score almost exactly the same as yours. It's just that 4000 seems abnormally high for 780Ti SLI

 

[freeleacher]

now you say that i see your point,
he actually got more than double than his single card test by a huge margin,
That kind of goes against the general rule of less than 100% efficiency wile in sli,
His getting 30% more in sli but maybe thats how its suppose to be who knows.

 

[Paladone]

now you say that i see your point,
he actually got more than double than his single card test by a huge margin,
That kind of goes against the general rule of less than 100% efficiency wile in sli,
His getting 30% more in sli but maybe thats how its suppose to be who knows.

The reason that it baffles me is because I'm running a 4930k at 4.4GHz vs his 4770k at 4.0 and my graphics cards are also overclocked and I got 800 points less

 

[GreiverBlade]

The reason that it baffles me is because I'm running a 4930k at 4.4GHz vs his 4770k at 4.0 and my graphics cards are also overclocked and I got 800 points less

if the bench only use 4 core (i am not sure even if Heaven show 6 core or 4 core in the result pannel if it really use cores above the 4th one ) afaik a 4930X is on same level of a 4770K (luckly it is not a 4820X, they even perform under a 4770k) also the CPU has little to do in a 100% GPU bound benchmark

also i wonder if since it's Haswell versus Ivy-E the ipc has something to do

i once read this about 3.0 i think 4.0 do same
"Unigine Heaven3.0 scored exactly the same with 4 cores vs 8 cores."

 

[Paladone]

if the bench only use 4 core (i am not sure even if Heaven show 6 core or 4 core in the result pannel if it really use cores above the 4th one ) afaik a 4930X is on same level of a 4770K (luckly it is not a 4820X, they even perform under a 4770k) also the CPU has little to do in a 100% GPU bound benchmark

also i wonder if since it's Haswell versus Ivy-E the ipc has something to do

i once read this about 3.0 i think 4.0 do same
"Unigine Heaven3.0 scored exactly the same with 4 cores vs 8 cores."

I understand that but regardless, my 4930k is still 400MHz higher and both of my graphics cards are over clocked (while his are not) yet he scored 800 Pts higher. 4k for 780Ti SLI does not seem right.

 

[GreiverBlade]

I understand that but regardless, my 4930k is still 400MHz higher and both of my graphics cards are over clocked (while his are not) yet he scored 800 Pts higher. 4k for 780Ti SLI does not seem right.

well the 400mhz clock more does nothing. my old i7-920 did the same score at 2.6ghz with a SLI of GTX580 Matrix @950/1170 and at 4.4ghz

after that i don't know about the GPU OC ... maybe your cards throttle down which could explain a 800pts less versus a stock clocked

also i noticed that the screen resolution line show 1920x1080 Fullscreen normally it shows 1920x1080 8xAA Fullscreen, so as you pointed he might have forgot to put 8xAA so his score is voided by that.

aka : nothing to see nothing to bother ...

so Durvell27 should remove the 1st rank multicard setup from him until we see a ingame shot with proof of FxAA x8 active

 

[Durvelle27]

Table

**Updated**

 

[JThorpe]

I apologize, I will double check my settings and run it again.

Sorry it was not intentional

 

[Paladone]

I apologize, I will double check my settings and run it again.

Sorry it was not intentional

It's not biggy dude, it happens

 

[JThorpe]

It looks like I did forget that setting. I ran the benchmark again with the 8xAA. Same stats as last time.

Here is my results (definitely lower).

Sorry again

 

[GreiverBlade]

It looks like I did forget that setting. I ran the benchmark again with the 8xAA. Same stats as last time.

Here is my results (definitely lower).

Sorry again

as Paladone wrote: no biggies, mistake can happen that's still a nice score

 

[MxPhenom 216]

I might need to do a rerun at my clock I did before at 1320. Some of these posts aren't making much sense to me. same GPU, lower clocks, higher score.........

 

[freeleacher]

re done due to missing clocks in image
A low 4.8 ghz on a i7 2700k

Single gpu

 

[HammerON]

I am not sure if most of you are aware of this or not but CPU over clocks have little effect on Heaven and Valley. GPU over clocks is where you want to focus
3DMark on the other hand...

 

[Paladone]

It looks like I did forget that setting. I ran the benchmark again with the 8xAA. Same stats as last time.

Here is my results (definitely lower).

Sorry again

Still a nice score