• Welcome to Overclockers Forums! Join us to reply in threads, receive reduced ads, and to customize your site experience!

Heaven score showed no increase with OC?

Overclockers is supported by our readers. When you click a link to make a purchase, we may earn a commission. Learn More.

Obsidus

Member
Joined
Aug 18, 2012
I have an issue that is just driving me nuts at the moment. You can see in my sig what my system is, and I will say that the CPU is overclocked to 4.6 and 2400 frequency on the NB, and the GPU is overclocked to 1220 on the core and 1350 on the mem. I'm fairly certain my CPU OC is stable, and I know that my GPU overclock is stable (I messed with it a ton, particularly the memory due to the fact that unstable memory overclocks on AMD cards tend to not cause you to crash, but instead lose FPS because it resends data).

With that out of the way I recently benchmarked my computer with Heaven DX11 after updating to the 12.8 drivers and updating my bios to the most recent release. When I did this, I accidentally reset my CPU overclock to 4.2, and I benched it AT 4.2. This is with Heaven running full screen, with high shaders, x4 AA, normal tess, and 2048 x 1152 resolution. I also ran the test 4 times back to back and the scores were broadly similar:

Min FPS: between 7.1 and 8.1
Max FPS: between 115 and 121
Avg FPS: between 48 and 53
Score around 1220 to 1228

Fair enough I guess, my score is largely killed by the minimum FPS due to Bulldozer not being able to hack it. Ok fine. So I overclocked my CPU back to 4.6 ghz and 2400 on the NB, with the same GPU overclock listed above; made sure it was stable and then ran Heaven 5 more times after the CPU overclock. I should also say that my GPU never gets hotter than 64 during Heaven, and my CPU never goes above 49.

The results? Almost IDENTICAL to the ones above. I think I gained a tad because the average FPS was 2 fps higher, but the min and max stayed roughly the same.

????

How the HELL does going from 4.2 to 4.6 on my CPU give me the SAME score? No increase at all, really? Especially on the minimum FPS, which as I understand it heavily depends on your CPUs ability to get the benchmark running up to speed when you first start it?

Can anyone explain this to me? It's getting really depressing to see people with similar systems to mine getting higher scores. It seems like people with my exact system with lower overclocks on the GPU, but even with a stock i5 2500k are getting up to 500 points more than I am. I realize that's just because Bulldozer is :rolleyes: compared to Sandy Bridge. But how come going from 4.2 to 4.6 saw NO gain? Makes me feel like I put in tons of work and extra stress on my CPU for NO reason. I realize that a lot of people take benchmarks with a grain of salt, but still.
 
Heaven tends to lurch right at the beginning of the benchmark, which causes a very low minimum FPS box despite it not being the GPU's fault.
I don't really know what a 7850 ought to be getting at 2048x1152 as I've never run one there.
Heaven is extremely GPU dependent, the higher the resolution the more GPU it is.


If you can download the HWBot version that forces a set of spec settings then we can compare your 7850 to the 7850s on HWBot and see how it's doing.

Are the people getting higher scores at the same resolution and settings? If all the settings aren't identical the scores aren't comparable.
 
Heaven tends to lurch right at the beginning of the benchmark, which causes a very low minimum FPS box despite it not being the GPU's fault.
I don't really know what a 7850 ought to be getting at 2048x1152 as I've never run one there.
Heaven is extremely GPU dependent, the higher the resolution the more GPU it is.


If you can download the HWBot version that forces a set of spec settings then we can compare your 7850 to the 7850s on HWBot and see how it's doing.

Are the people getting higher scores at the same resolution and settings? If all the settings aren't identical the scores aren't comparable.

You know that is a good point, 99% of the benchmarks I've seen were done at 1920 x 1080. I've done a few at that resolution as well and the scores tend to be 100 points higher (so around 1325 instead of 1225). Still doesn't explain why going from 4.2 to 4.6 on the CPU showed no gain. I'll try the HWbot version.
 
you will get more out of increasing bclock in heaven than increasing multi aswell.

Mhz does help with score but not alot after a certain point its pure gpu and mem speed on card.
 
you will get more out of increasing bclock in heaven than increasing multi aswell.

Mhz does help with score but not alot after a certain point its pure gpu and mem speed on card.

Ahh ok, I see. I guess I was working under an incorrect assumption. Many people on another forum were saying that the minimum FPS is heavily reliant on the CPU, which is why with people that have Sandy/Ivy bridge CPU's you'll see minimums of 30-40 FPS; but with Bulldozer its really low.

With HWbot Heaven, should I run basic or extreme?
 
Extreme.

Minimum FPS in my experience is due to the lurch at the start.
Once the bench is running beyond that lurch CPU speed might be involved.
 
Well, just ran the extreme bench on HWbot Heaven, it didn't give me a min/max/average, just a score (although I looked at the monitor about 10 seconds in an it said min fps 16).

The score I got was 783.265. Sounds terrible, but I have no clue if that sounds about right for my computer.
 
running the test more than once can help with that .

also if you can do set infinity
 
running the test more than once can help with that .

also if you can do set infinity

Looked around on the test and I saw no way to change anything, at least in the HWbot version, just step 1,2,3,4; so I'm not sure how to set it to infinity or whatever.

I also ran it two more times and got 773.27 and 774.34 respectively. These scores can't be good. :rain: But, I have no frame of reference so. I didn't and don't expect my computer to be a heavyweight or anything, but these scores seem really crappy even still.
 
Haha wow, that made a big difference. 1458.102, that sounds much better.
 
Yeah, Tess being off makes a huge difference in the benchmark (ATI is lucky that they can turn it off, as nVidia can't via drivers).

As was stated, Heaven is largely GPU dependent, and the minimum FPS marker is because (as Bobnova stated) of the lurch/lag spike at the beginning. I've always had it.
 
Yeah, Tess being off makes a huge difference in the benchmark (ATI is lucky that they can turn it off, as nVidia can't via drivers).

As was stated, Heaven is largely GPU dependent, and the minimum FPS marker is because (as Bobnova stated) of the lurch/lag spike at the beginning. I've always had it.

Hmm, that's strange. There was a LONG (33 page long) thread on another forum, that was FILLED with tons of screen shots of peoples Heaven results. 98% of the people in that thread, regardless of their GPU (some even had things like r6550's) were using Intel CPU's, mostly i5 2500k's, but also some i7's. EVERY single person with an Intel CPU, regardless of their GPU, got between 25-40 MINIMUM FPS, and consequently scored a fair bit higher due to that. I always thought this was indicative of the Intel CPU's being that much better than the Bulldozer and Phenom II processors. But if you're saying that Heaven is very GPU reliant, then I wonder what the cause is. Especially since you have an i7 and you say you still get the slowdown at the start.
 
Eh, my i7 920, i5 2500k/i7 2600k/and i7 3770ks have always had that lurch at the beginning that caused a low minimum (with ATI 6950/6970/7970, GTX 580, etc). I look at it as more of a problem with the benchmark. That's why I prefer to use the HWBot version as I don't believe it really cares about that 2-second of slowness much.

But yeah, Heaven is one of the most GPU-reliant benchmarks available. All of the rest of them (3DMark 01/03/05/06/Vantage/11) benefit from a strong CPU overclock to get a great score (or at least most do).
 
Eh, my i7 920, i5 2500k/i7 2600k/and i7 3770ks have always had that lurch at the beginning that caused a low minimum (with ATI 6950/6970/7970, GTX 580, etc). I look at it as more of a problem with the benchmark. That's why I prefer to use the HWBot version as I don't believe it really cares about that 2-second of slowness much.

But yeah, Heaven is one of the most GPU-reliant benchmarks available. All of the rest of them (3DMark 01/03/05/06/Vantage/11) benefit from a strong CPU overclock to get a great score (or at least most do).

Gotcha. I'm new to the whole overclocking/benchmarking thing, so I don't know squat about all this. Nice to learn.
 
Yep, always happy to give any help at all, feel free to ask us anything (regarding benching :))
 
Back