Intel 3570K and Batman Arkham City GOTYE

[Swarnendu]

I am getting very low fps on my system running Batman Arkham City GOTYE, benchmarked using the ingame benchmarking. At all settings maxxed but physx off, I am getting 43 fps average. My friends are running 3770K on Maximus V Gene with stock Zotac GTX 680, they are getting 81 fps average at maxxed settings but physx off, at 1920x1080 resolution. I am running 3570K on Big Bang MPower Z77 with Sapphire Vapor-X 7950 at 1280x1024 resolution. online benchmarks show stock 7950 to produce 70 fps at full settings but physx off at 1080p using 3770K cpu. If that is so then I am supposed to get at least above 80 fps average at maxxed settings and physx off, with my 7950 vapor-x (150mhz oc) running 1280x1024 resolution.

One of my friends used to run 2700K with Zotac GTX 680 stock at 1080p resolution, and use to get 55 fps average with physx off but all others maxxed. When he upgraded his cpu to 3770K, his fps shot through 80, without any other change in hardware config. I just cannot imagine such difference in fps between 2700k and 3770k.

Anyone has any clue?, Please help me understand this whole situation. Now batman ac gotye is twice as fast with 3770K compared to 3570K? I just cannot imagine that.

Please Help..

 

[Rammy]

It's kinda impossible to compare two systems which have different configurations and different ages. They will exist outside of the margins of benchmarking due to all the things people do which gradually eat away at performance.
It's also worth remembering that many games (Arkham City included) have an awful lot of scaling options. One persons "maxxed" isn't necessarily the same as another persons. From what I remember about Arkham City, the difference in an average FPS and a minimum versus other games is quite dramatic, this may have a knock-on effect on what results you get.
Having said that, 43fps seems very low. I'm going to assume you've done all the standard stuff like drivers, cpu/gpu load and heat.
I'd check how well your scaling is working, ie - turn details down (like MSAA, Tesselation) and see if you get a dramatic leap in performance that brings you more into line.
As for how your friend got 15fps from a tiny cpu upgrade - he can't. I'm not saying he is lying, but it's likely that there is another factor which he hasn't considered, a GTX680 can probably get 55fps at 2560*1440, 80 is much more like it for 1080P.

 

[Swarnendu]

It's kinda impossible to compare two systems which have different configurations and different ages. They will exist outside of the margins of benchmarking due to all the things people do which gradually eat away at performance.
It's also worth remembering that many games (Arkham City included) have an awful lot of scaling options. One persons "maxxed" isn't necessarily the same as another persons. From what I remember about Arkham City, the difference in an average FPS and a minimum versus other games is quite dramatic, this may have a knock-on effect on what results you get.
Having said that, 43fps seems very low. I'm going to assume you've done all the standard stuff like drivers, cpu/gpu load and heat.
I'd check how well your scaling is working, ie - turn details down (like MSAA, Tesselation) and see if you get a dramatic leap in performance that brings you more into line.
As for how your friend got 15fps from a tiny cpu upgrade - he can't. I'm not saying he is lying, but it's likely that there is another factor which he hasn't considered, a GTX680 can probably get 55fps at 2560*1440, 80 is much more like it for 1080P.

If I disable dx11 features and tessellation and extreme preset with 8X MSAA and physx disabled, then there is a dramatic leap in fps with respect to if I enable dx11 features and tessellation, I am getting 200 fps average at the end of ingame benchmark. turning MSAA down does not hurt fps much.
Do not worry about my system's hardware and software health, they both are top notch, you might agree with me if u had seen and used my system..
My question is what can be the possible cause of this low fps?

 

[Rammy]

I'm not trying to be condescending, just pointing out that a direct comparison with two machines like that is a bit ...awkward.
As I don't know you, or your friend, I'm just going on the obvious issues. If that seems like I'm insulting you, it's not intentional.

The reason I asked about scaling is because it can potentially rule out issues. If I am understanding correctly, you go from 43fps to 200fps by disabling DX11 features and tessellation? On launch Arkham City had an issue where DX11 features were destroying FPS, here's a benchmark from Toms around that period which illustrates the problem - http://www.tomshardware.com/reviews/geforce-gtx-560-ti-448-core-benchmark,3082-5.html While you aren't having quite that extreme a problem, it seems more than coincidental.
Most benchmarks I've seen of this game use Normal tessellation. I'm wondering if that would put you into a more reasonable ballpark in terms of what you (and I) would expect to see.

 

[Swarnendu]

I'm not trying to be condescending, just pointing out that a direct comparison with two machines like that is a bit ...awkward.
As I don't know you, or your friend, I'm just going on the obvious issues. If that seems like I'm insulting you, it's not intentional.

The reason I asked about scaling is because it can potentially rule out issues. If I am understanding correctly, you go from 43fps to 200fps by disabling DX11 features and tessellation? On launch Arkham City had an issue where DX11 features were destroying FPS, here's a benchmark from Toms around that period which illustrates the problem - http://www.tomshardware.com/reviews/geforce-gtx-560-ti-448-core-benchmark,3082-5.html While you aren't having quite that extreme a problem, it seems more than coincidental.
Most benchmarks I've seen of this game use Normal tessellation. I'm wondering if that would put you into a more reasonable ballpark in terms of what you (and I) would expect to see.

No no please do not think like that, you were never offensive and moreover why would you be. Yes you are very correct about comparing two systems like that has discrepancies, but there should not be so much performance difference between these two types of systems, provided the hardware and software configuration of the two systems are properly done such as to be able to produce best performance possible, for the given system. My question is how can there be so much performance diff between to such much high end system? I wonder if this is due to my cpu being 3570k instead of 3770k, if that is the case then also it is insane. Consider that me and my friends installed the exact same copy of the game in the exact same way, I have done it myself. And also I did set up their systems exactly in my own way, both the bios and windows settings and with same copy of windows. Just want to know that if anyone can tell me surely if this is due to the cpu, as I am suspecting to be that. If you can assure me different like it being a software issue or any other system settings, then good for me..

Thanks a lot..

 

[Rammy]

Unless there is something wrong with your CPU, it's not the CPU causing the difference. You'll presumably have different overclocks which will skew the results a little, but not to that level. It's almost undoubtedly due to to graphics, one way or another.

It's an Nvidia branded game which doesn't necessarily mean you are going to get a big difference in performance, but it might explain certain features favouring Nvidia cards.

You said yourself that you are jumping from 43-200fps by tweaking a couple of settings which suggests that for whatever reason, your card is struggling. It might be a driver issue, or it might be poor optimisation on these settings as I alluded to earlier. Frankly, it seems mad that any game would force a slowdown on a 7950 at that resolution, but it seems like that is what is happening. To be clear though, we aren't talking about a huge performance gap. If you compared both systems at 1080P, with max settings but normal tessellation, you'd probably get a very close result.

To be honest, I own Arkham City and max tessellation cripples my card too, but I genuinely can't see the difference in game between max-normal and even off completely.

 

[Swarnendu]

Unless there is something wrong with your CPU, it's not the CPU causing the difference. You'll presumably have different overclocks which will skew the results a little, but not to that level. It's almost undoubtedly due to to graphics, one way or another.

It's an Nvidia branded game which doesn't necessarily mean you are going to get a big difference in performance, but it might explain certain features favouring Nvidia cards.

You said yourself that you are jumping from 43-200fps by tweaking a couple of settings which suggests that for whatever reason, your card is struggling. It might be a driver issue, or it might be poor optimisation on these settings as I alluded to earlier. Frankly, it seems mad that any game would force a slowdown on a 7950 at that resolution, but it seems like that is what is happening. To be clear though, we aren't talking about a huge performance gap. If you compared both systems at 1080P, with max settings but normal tessellation, you'd probably get a very close result.

To be honest, I own Arkham City and max tessellation cripples my card too, but I genuinely can't see the difference in game between max-normal and even off completely.

Yes I did not check at 1080p myself. I checked online benchmarks of this game with stock reference 7950 at 1080p with extreme detail level and all settings maxxed but tessellation set to normal and physx off, the average fps was 78. This is why I am worried. Since all the benchmarks I have seen had tessellation set to normal and physx off and all others maxed, I did also try this at this settings but there was no difference in fps, that is it was still at 43 average. There was no diff in fps with tessellation set to high, as it was when I previously benchmarked.

 

[Rammy]

Right, so when you produced 200fps was with "DX11 features" turned off too? Presumably that's the problem then.
I don't see what you have to be worried about really. A well made PC game usually has enough graphics options to cripple even the best graphics card. I (as you) am surprised that that game can do it at such a low resolution to that card, but it's one game and the settings in question don't have much (if any) impact on the game experience.

Remember all these online benchmarks tend not to provide specifics of their settings, it's quite possible they aren't using "DX11 features" either. Most use a detail preset (Ultra, Extreme or whatever the game says) which usually has additional components not included such as MSAA and in the case, DX11 stuff.

Also, this is a single game. If you are consistently getting crappy performance in other games then it's possible there is something amiss, but as far as I can tell there is nothing wrong.

 

[Swarnendu]

I did recheck and benched Batman Arkham City GOTYE with two different cpus 2700k and 3770k with gtx 680 reference on asus maximus v gene with corsair vengeance 1866 8gb kit and corsair tx750 v2. Now I am sure that such performance gain is not because of anything else but he cpu. From the benchmarks (physx gpu gtx 680 selected form nvidia control panel) I get to see that there is around 31% performance improvement with 3770k over 2700k all at stock speeds and asus multicore enhancement enabled. This is insane. And as I have already told you that 3770k has 100% performance improvement over 3570k, all at stock.
Any one please clarify me, If I am correct. Rammy if u did run Batman AC GOTYE on your system, please tell me the min, max and avg fps u get at the end of the ingame benchmark, and also tell me ur system spec.

 

[Rammy]

That's very strange indeed. As I'm sure you are aware there isn't going to be much direct performance difference between the two CPUs. It's more likely in that case to be some type of feature (like PCIe3.0) which is skewing the results.
Lets be clear though, there are quite a few "holes" in your assessment.
- You have been testing on two different systems, making a direct comparison impossible
- This is one game, which may or may not favour things indiscriminately.
- The Arkham City benchmark tool is not a "true" performance test, as it flicks between scenes and this counts as part of the test.

I don't have the game installed at the moment, but from what I remember I was running at 1080P with more or less everything turned on. I'd assume I didn't have Physx or the DX11 stuff turned on. My system is a basic i5 3450 and GTX660. My framerate was never in question, I'd assume 60+ but I never really checked.

 

[Swarnendu]

That's very strange indeed. As I'm sure you are aware there isn't going to be much direct performance difference between the two CPUs. It's more likely in that case to be some type of feature (like PCIe3.0) which is skewing the results.
Lets be clear though, there are quite a few "holes" in your assessment.
- You have been testing on two different systems, making a direct comparison impossible
- This is one game, which may or may not favour things indiscriminately.
- The Arkham City benchmark tool is not a "true" performance test, as it flicks between scenes and this counts as part of the test.

I don't have the game installed at the moment, but from what I remember I was running at 1080P with more or less everything turned on. I'd assume I didn't have Physx or the DX11 stuff turned on. My system is a basic i5 3450 and GTX660. My framerate was never in question, I'd assume 60+ but I never really checked.

Yes u r right, to be sure about pcie 2.0 bottlenecking in case of 2700k, I did also run the gtx 680 in pcie 3.0 x8, but there was no diff in fps atleast. Yes I know already Batman AC ingame bench is not a true performance test, as u said, but atleast I am using the same benchmarking for both cases. And moreover to be double sure I did also checked diff parts of the game playing with fraps turned on, fps diff was same or very close, more or less, to as shown by the ingame benchmark.
Thanks for your cooperation.

 

[Alexander Tsodikov]

Has anybody tried Hialgoswitch for Batman Arkham City to increase fps?