i5 3570k low physics score in 3dmark

[RDGnome]

Hi,
I've noticed that my 3dmark score on Firestrike (default) for physics is pretty low for my processor. Every test I've seen with the same cpu (stock clock) has at least 7500 if not 8000 and I'm getting 6000-6700. I haven't overclocked but I know from these comparisons that I should be getting quite a bit higher. Even 3dmark says "Your CPU doesn't want to play"...
Here are some screenshots of the best result I've got and a recent one:
i.imgur.com/i36GmHz.png
i.imgur.com/NI2plzs.png
I've made sure my RAM is at the correct frequency and that turbo boost and speedstep are enabled (despite 3dmark thinking they're not), I'm really not sure what could be causing this to be so low.
I can post some more info if needs be.
Any help would be appreciated, thanks.

(System specs are on the screenshots)

[mikeo01]

I wouldn't exactly worry about it, if the games you run run fine then I don't see the problem?

A quick Google suggests 3DMark has some issues on Windows 8 causing a lower score. Power saving features or throttling can cause a lower score.

... but yeah unless you are experiencing bad gameplay (which I assume that's what you are using SLI for ) then I wouldn't worry. They're just benchmarks at the end of the day.

[CAT-THE-FIFTH]

Try running one graphics card only and see what score you get.

[RDGnome]

Yeah I would say all my games run fine (although I have been experiencing some odd fps drops occasionally) and it's true these are just benchmarks, but you know the feeling when you spend a lot of money you want your parts to perform how they should be performing?
I think you might have a point about Windows 8, however having said that, looking back on my benchmarks from when I had W7 and the same hardware, the physics score is essentially the same.
I also made sure I tested when power saving features were disabled and they were set to high performance.
Thanks for the reply

[RDGnome]

Try running one graphics card only and see what score you get.

Alright I'll test that now

[RDGnome]

Try running one graphics card only and see what score you get.

Tried running with one gpu disabled and the results are similar sadly: 3dmark.com/fs/2188611

[CAT-THE-FIFTH]

Could your CPU be throttling??

[RDGnome]

I guess it's a possibility but I did test with high performance power setting to see if that would make a difference and it didn't. What else would cause it to throttle do you think?

[CAT-THE-FIFTH]

I guess it's a possibility but I did test with high performance power setting to see if that would make a difference and it didn't. What else would cause it to throttle do you think?

Is the CPU getting too hot??

[RDGnome]

Is the CPU getting too hot??

I wouldn't say so, I have a Hyper 212 Evo and temperatures never reach 60 degrees and rarely breach 55 degrees. Idles around 35 degrees.

[Stuen4y]

Try running something that is not 3dmark, like Cinebench or other CPU benchmark and post the result in a screenshot with an open CPU-Z window.

[RDGnome]

Try running something that is not 3dmark, like Cinebench or other CPU benchmark and post the result in a screenshot with an open CPU-Z window.

I ran cinebench, here are the results with the CPU-Z screenshot while the test was running: http://i.imgur.com/UG0wQOB.png
I noticed that the turbo frequency wouldn't hit 3.80GHz which is the max and stopped short at 3.60GHz

[Bonebreaker777]

I ran cinebench, here are the results with the CPU-Z screenshot while the test was running: http://i.imgur.com/UG0wQOB.png
I noticed that the turbo frequency wouldn't hit 3.80GHz which is the max and stopped short at 3.60GHz

That is all right.

The i5-3570k under load boosts the frequency depending on the number of cores utilized.
1-2 cores utilized: 3.8Ghz turbo
3 cores utilized: 3.7Ghz turbo
4 cores utilized: 3.6Ghz turbo

http://www.intel.com/support/process.../CS-032278.htm

[Stuen4y]

I ran cinebench, here are the results with the CPU-Z screenshot while the test was running: http://i.imgur.com/UG0wQOB.png
I noticed that the turbo frequency wouldn't hit 3.80GHz which is the max and stopped short at 3.60GHz

The result looks about fine for me. At 4.5GHz my 3570K does ~620 points and if you interpolate this, for 3.6GHz you should be looking at something like
~500 points, so +/- 10 points is absolutely fine. I can see your memory is at 1333MHz, this is the main reason for the difference. I would try overclocking it to at least 1600MHz as 1333 is quite a bottleneck in a modern system. I also see that you have a Z motherboard, go and set the multiplier at something like 40, it should work like a charm with default voltage and box cooler and will stick all 4 cores at 4GHz when loaded. This is in case you do not want to oveclock in general.

[RDGnome]

That is all right.

The i5-3570k under load boosts the frequency depending on the number of cores utilized.
1-2 cores utilized: 3.8Ghz turbo
3 cores utilized: 3.7Ghz turbo
4 cores utilized: 3.6Ghz turbo

http://www.intel.com/support/process.../CS-032278.htm

Very interesting thanks, that would explain that.

The result looks about fine for me. At 4.5GHz my 3570K does ~620 points and if you interpolate this, for 3.6GHz you should be looking at something like
~500 points, so +/- 10 points is absolutely fine. I can see your memory is at 1333MHz, this is the main reason for the difference. I would try overclocking it to at least 1600MHz as 1333 is quite a bottleneck in a modern system. I also see that you have a Z motherboard, go and set the multiplier at something like 40, it should work like a charm with default voltage and box cooler and will stick all 4 cores at 4GHz when loaded. This is in case you do not want to oveclock in general.

That's good to know, I also did a Passmark test and it performed fine, higher than average even. My worry is the direct relationship to games and the physics aspects. I consistently see scores of 8000+ and my score is at 6300 which is stupidly low.
Strange about my memory, I know for a fact I set it at 1600MHz (it is 1600MHz RAM so I wouldn't need to OC it), I will double check that however.
I did have it OC'd a while ago and I'll probably will be overclocking it again at some point, I already have a Hyper 212 Evo so temps are no issue as I'm not planning on going all out. I haven't tried to get beyond 4.1GHz though, do you think OC'ing to 4.5GHz would be okay with my cooler? Oh and what voltage do you need for your 4.5GHz?

Edit: double checked my RAM frequency and it's definitely at 1600MHz (where did you actually see that it said its running at 1333MHz?)

[Stuen4y]

In the Furutemark report it was recognised as 1333MHz but it may be wrong. You can always check with CPU-Z what is the current frequency. Do not leave it at auto as it will go by the SPD values which are always 1333MHz, load the XMP profile for 1600MHz, otherwise load all the primary timings and voltages as on the box/memory sticks.

About the overclock, I would say 4.5GHz is not possible with the Hyper 212 Evo unless you have an extraordinary sample that does it on like 1.15V. 4.3-4.4 though will be more than possible with good temperatures under load (under 90C° in stress testing and under 75C° in normal use). Memory is a bottleneck but not massive so the fact that a synthetic physics benchmark scores lower points does not mean your system will perform 30% lower in a Physics enabled application.
P.S. I am really sorry about the super late reply, I have not checked the forum recently.