Skyrim CPU OC Scaling?

[Stupid_Newbie]

Well I saw on THG that an oc to the 2500k gave a good boost to FPS. This was an OC to 4ghz. Im wondering where the gains stop. Obviously there will be a ceiling where more speed, means little to no FPS gain. Anyone know where that is. I have tried a search here, skyrim! forums (official), and google. People talk about OC helping. But cant find anything saying how much matters.

 

[PrincessFrosty]

First of all, the game is capped at 60fps with vsync on, so no matter how fast your CPU is you'll never see more than 60fps (for a 60hz monitor) unless you disable Vsync (iPresentInverval=0 in the skyrimprefs.ini file).

The problem with vsync is it behaves as if it's rounding down your FPS to certain values:
FPS >60 is rounded down to 60fps
FPS <60 but >30 is rounded down to 30fps
FPS <30 but >20 is rounded down to 20fps
etc...

So with vsync on, going from something like 40fps to 50fps will show little or no benefit because they'll both behave as if you're getting 30fps.

Assuming you have disabled vsync AND you're definitely CPU limited then an OC will help to some degree, I don't know exact FPS gains as I've not measured myself, my 2600k is more than enough, honestly I'm surprised the 2500k struggles at all, I'm starting to think this is either a bug or people are GPU limited and don't realise it.

Basically you will keep seeing improvement with a CPU overclock until you are no longer CPU limited and become GPU limited, at which point additional CPU speed wont help. This ceiling will happen at different frame rates depending on 2 things:
1) What GPU you have
2) What graphics settings you've picked for the game.

For example if your GPU can mange 100fps and your CPU can only manage 50fps then you'll see improvements in CPU overclocking until your frame rate hits 100fps at which point you switch back to becoming GPU limited.

 

[bigdogchris]

First of all, the game is capped at 60fps with vsync on, so no matter how fast your CPU is you'll never see more than 60fps (for a 60hz monitor) unless you disable Vsync (iPresentInverval=0 in the skyrimprefs.ini file).

The problem with vsync is it behaves as if it's rounding down your FPS to certain values:
FPS >60 is rounded down to 60fps
FPS <60 but >30 is rounded down to 30fps
FPS <30 but >20 is rounded down to 20fps
etc...

So with vsync on, going from something like 40fps to 50fps will show little or no benefit because they'll both behave as if you're getting 30fps.

I'm not really sure what you're trying to say here. I have VSYNC on and my FPS is anywhere from 40-60 depending on where I am. It never 'rounds it down'.

 

[Brackle]

First of all, the game is capped at 60fps with vsync on, so no matter how fast your CPU is you'll never see more than 60fps (for a 60hz monitor) unless you disable Vsync (iPresentInverval=0 in the skyrimprefs.ini file).

The problem with vsync is it behaves as if it's rounding down your FPS to certain values:
FPS >60 is rounded down to 60fps
FPS <60 but >30 is rounded down to 30fps
FPS <30 but >20 is rounded down to 20fps
etc...

So with vsync on, going from something like 40fps to 50fps will show little or no benefit because they'll both behave as if you're getting 30fps.

Assuming you have disabled vsync AND you're definitely CPU limited then an OC will help to some degree, I don't know exact FPS gains as I've not measured myself, my 2600k is more than enough, honestly I'm surprised the 2500k struggles at all, I'm starting to think this is either a bug or people are GPU limited and don't realise it.

Basically you will keep seeing improvement with a CPU overclock until you are no longer CPU limited and become GPU limited, at which point additional CPU speed wont help. This ceiling will happen at different frame rates depending on 2 things:
1) What GPU you have
2) What graphics settings you've picked for the game.

For example if your GPU can mange 100fps and your CPU can only manage 50fps then you'll see improvements in CPU overclocking until your frame rate hits 100fps at which point you switch back to becoming GPU limited.

That is what triple buffering fixes.

 

[UtopiA]

The only "roof" that matters is the one on your processor.
Just overclock the damn thing until you can't go any higher.

 

[PrincessFrosty]

I'm not really sure what you're trying to say here. I have VSYNC on and my FPS is anywhere from 40-60 depending on where I am. It never 'rounds it down'.

I'm simply describing how Vsync works in a simplified way, it doesn't actually round down the frame rate but for all intents and purpose that's how it's behaving to anyone observing it.

You don't see rounding down in your frame rate measurement because the tool you use to measure frame rate takes an average over some set period of time, during that period you might have frames rendered at different speeds (60fps or 30fps or 20fps) and the average of those frames is going to be some number in between them.

That is what triple buffering fixes.

Triple buffering mitigates the problem of "input lag" somewhat by allowing the video card to remain busy rendering additional frame ahead of the display so that if newer infromation is available before the refresh occurs then it can be displayed, mainly helpful for cutting down input lag when your frame rate is higher than your refresh rate, but it doesn't "fix" the problem.

If one specific frame takes too long to render for whatever reason then vsync is still going to keep the current frame in the buffer while the monitor does another refresh meaning the same frame is displayed for 2 sequential refreshes essentially halving the frame rate for at least that 1 frame. Triple buffering or not, 2 things are still true.

1) You only have at most 1 unique frame per refresh so a max of 60fps
2) If a frame isn't ready in time for the next refresh, the previous frame is used, this forced duplication of frames causes artificial delays and lowers your effective frame rate from that of a non-vysnc enabled configuration.

 

[Stupid_Newbie]

Well a gtx570 at 1920x1080/16af/4aa gained 10fps on the bottom end and 12 on the top end by a 4ghz oc on the 2500k. Seems fairly CPU demanding aswell

But on games like scII (very CPU demanding) you don't gain anything from 4.5ish up. So just wonder if anyone has seen a full review with full cpu scale or tested it.

 

[PrincessFrosty]

Well a gtx570 at 1920x1080/16af/4aa gained 10fps on the bottom end and 12 on the top end by a 4ghz oc on the 2500k. Seems fairly CPU demanding aswell

Again it depends on the settings, by modern standards 1920x1080 is a fairly easy resolution for that card to render at you would expect the CPU to be the bottleneck, if you cranked that resolution to something like 2560x1600 or above (multi-monitor) and enabled 8xAA you'd see it become GPU limited.

 

[Stupid_Newbie]

Yeah I'll be 1080 for now. Till I can get a curved monitor (equal to 3 monitor setup) with an appropriate res. hate the bezels in the middle. I'll just run it at 4.8 or so.

 

[BababooeyHTJ]

Again it depends on the settings, by modern standards 1920x1080 is a fairly easy resolution for that card to render at you would expect the CPU to be the bottleneck, if you cranked that resolution to something like 2560x1600 or above (multi-monitor) and enabled 8xAA you'd see it become GPU limited.

This engine doesn't work like that. That can and will increase the load on the cpu as well kind of like GTA4 or FSX or even wow. A lot of open world games are like that.

Don't get me wrong its possible to be limited by the video card in this game but more often than not you'll be cpu limited and thats going to be even more true with next gen cards. Bethsoft games are always like that and mods will make things more cpu limited.

 

[PrincessFrosty]

This engine doesn't work like that. That can and will increase the load on the cpu as well kind of like GTA4 or FSX or even wow. A lot of open world games are like that.

Don't get me wrong its possible to be limited by the video card in this game but more often than not you'll be cpu limited and thats going to be even more true with next gen cards. Bethsoft games are always like that and mods will make things more cpu limited.

Depends what you change, simply changing the resolution will have negligable impact on the CPU but will have a massive effect on the GPU, I've already posted screenshots of my GPU at 99% and my CPU at 50% as an example of that.

It depends entirely on the setting you use, you can make it CPU or GPU limited by simply setting the right settings.

What evidence can you provide that more often than not people are CPU limited in the game?

Sure next gen cards will be faster, so will next gen CPUs, I don't see your point here.

 

[BababooeyHTJ]

Depends what you change, simply changing the resolution will have negligable impact on the CPU but will have a massive effect on the GPU, I've already posted screenshots of my GPU at 99% and my CPU at 50% as an example of that.

It depends entirely on the setting you use, you can make it CPU or GPU limited by simply setting the right settings.

What evidence can you provide that more often than not people are CPU limited in the game?

Sure next gen cards will be faster, so will next gen CPUs, I don't see your point here.

Again, task manager isn't accurate at reporting cpu usage.

I'm just going by my past experience with this engine and the similarities that I have seen in this game. Also the benchmarks that I have seen and posted. You can believe what you want, I don't care. Even with just a 2500k and GTX570 when the framerate really dips its usually due to the cpu and not the video card, nevermind people running multi-gpu.

This engine is more cpu dependent than gpu dependent. Its no different in that regard than Oblivion, Fallout 3, or New Vegas.

 

[PrincessFrosty]

Again, task manager isn't accurate at reporting cpu usage.

Task manager may not be completely accurate but it's accurate enough to say it's not using anywhere near the resources of a single stock 2600k while clearly it's maxing out the GPU. If you want me to test with something "more accurate" then post what you feel is a accurate enough test and I'll endeavour to run it.

I'm just going by my past experience with this engine and the similarities that I have seen in this game. Also the benchmarks that I have seen and posted. You can believe what you want, I don't care. Even with just a 2500k and GTX570 when the framerate really dips its usually due to the cpu and not the video card, nevermind people running multi-gpu.

As I said, it depends entirely on the settings, I could take that hardware and run the game and tweak it so it was either/or depending on what settings I use, there's plenty to throw at the video card, high resolutions, high levels of super sampled anti-aliasing that will bring any card to it's knees and make the game GPU bottlenecked.

This engine is more cpu dependent than gpu dependent. Its no different in that regard than Oblivion, Fallout 3, or New Vegas.

You're comparing apples to oranges, CPUs and GPUs are different things and you can put different loads on both of those components depending on what settings you run the game in. So no, the engine is not "more cpu dependent than gpu dependent", it's dependent on what settings you pick and what hardware you have.

Not everyone is going to have the same balance of hardware some people have very old gpus and modern cpus, some people have the opposite, some people have tiny resolutions which limit the stress they can put on the graphics card. A lot of people haven't turned vsync off and might not be bottlenecked by either the CPU or GPU but rather hitting the frame rate limit of the game and have power to spare on both components.

 

[BababooeyHTJ]

Task manager may not be completely accurate but it's accurate enough to say it's not using anywhere near the resources of a single stock 2600k while clearly it's maxing out the GPU. If you want me to test with some "more accurate" then post what you feel is a accurate enough test and I'll endeavour to run it.

No, its not even accurate enough. If you listen to task manager this game uses four cores which it does not. There are a lot of things to take into account with a cpu. Task manager is not even accurate enough.

For example this engine really likes shared cache. You can see that in the Anandtech sandybridge review and in the hardware heaven sandy-e review. How does task manager account for shared L3 cache?

As I said, it depends entirely on the settings, I could take that hardware and run the game and tweak it so it was either/or depending on what settings I use, there's plenty to throw at the video card, high resolutions, high levels of super sampled anti-aliasing that will bring any card to it's knees and make the game GPU bottlenecked.

I'm sorry but you are making assumptions that you can't back up. Who cares if extreme levels of anti-aliasing that no one uses put more of a load on the video card? That goes without saying. Guess what, with this engine high resolutions can put more of a load onto the cpu just like in a lot of other open world games. That is especally true with eyefinity and NV surround since there is also more geometry to render due to the larger fov.

You're comparing apples to oranges, CPUs and GPUs are different things and you can put different loads on both of those components depending on what settings you run the game in. So no, the engine is not "more cpu dependent than gpu dependent", it's dependent on what settings you pick and what hardware you have.

Did you read what I said? With a typical high end video card that a lot of people here seem to have like a gtx570 you are most likely to be cpu bound when you see the big framerate dips in this game. Just like its been with the past three Bethsoft games on this engine.

Here is what I saw while walking around riverwood at 1920x1080 with the game maxed out on a GTX570 at stock speeds.

This is one of the areas where I noticed some slow downs. These are the areas where I care about if I am cpu bound or not. There are a lot of people using more powerful graphics cards, multiple cards, and gpu overclocks on this forum. On the otherhand no ones cpu is really going to be much faster. So yeah, chances are you are more likely to be cpu bound in this game and that gap is only going to grow with upcoming video cards.

2500k @ 4.6ghz
2011-11-13 18:56:52 - TESV
Frames: 1831 - Time: 36489ms - Avg: 50.180 - Min: 41 - Max: 69

2011-11-13 18:57:57 - TESV
Frames: 1234 - Time: 25334ms - Avg: 48.709 - Min: 38 - Max: 62

2011-11-13 18:58:57 - TESV
Frames: 1723 - Time: 34741ms - Avg: 49.596 - Min: 41 - Max: 64

2011-11-13 19:00:08 - TESV
Frames: 1852 - Time: 39905ms - Avg: 46.410 - Min: 39 - Max: 59

2500k @ stock

2011-11-13 19:20:54 - TESV
Frames: 1385 - Time: 35553ms - Avg: 38.956 - Min: 32 - Max: 49

2011-11-13 19:22:04 - TESV
Frames: 1304 - Time: 34398ms - Avg: 37.909 - Min: 30 - Max: 48

2011-11-13 19:23:07 - TESV
Frames: 1116 - Time: 28111ms - Avg: 39.700 - Min: 32 - Max: 57

2011-11-13 19:24:09 - TESV
Frames: 1122 - Time: 28158ms - Avg: 39.847 - Min: 33 - Max: 52

 

[IsaacMM]

I ordered a Noctua D-14 just because of this game, I hope I don't regret it.

 

[Stupid_Newbie]

I regret mine and haven't even used it much. Lol I want h2o

Now I don't pretend to know a tonne about how it all works, but wouldn't 50% CPU usage on a 4 core CPU, running a 2 core game, be 100% on those 2 cores? Or am I way off.

 

[PrincessFrosty]

No, its not even accurate enough. If you listen to task manager this game uses four cores which it does not. There are a lot of things to take into account with a cpu. Task manager is not even accurate enough.

For example this engine really likes shared cache. You can see that in the Anandtech sandybridge review and in the hardware heaven sandy-e review. How does task manager account for shared L3 cache?

Seriously, use your head. I'm showing you 99% usage on the GPU while simultaneously showing you 50% usage on the CPU. There's 2 lines of evidence to show that it's GPU limited here, first of all the GPU is maxed out to a degree you would expect with GPU bottlenecking and the CPU is clearly not maxed out.

In this particular circumstance (under the conditions I play the game) the game is CLEARLY GPU bottlenecked and quite frankly any attempt by you at this stage to discredit this is just complete rubbish. If you want me to run different CPU tools then I will, tell me what you want me to run and I'll run it (within reason), I asked you last post but you've refused to provide an answer to that. You'd also have to provide good reason to think afterburner incorrectly reports GPU usage to explain why the video card perfectly maxes out at ~99% usage.

I'm sorry but you are making assumptions that you can't back up. Who cares if extreme levels of anti-aliasing that no one uses put more of a load on the video card?

What specific assumption am I making? I'm saying that FACTUALLY you can balance the load either way to be CPU or GPU bottlenecked based on the settings the user picks. I'm not trying to say anyone should care, what I'm saying is that your claims that it's one way or the other are WRONG and that what happens in real life (bottleneck wise) on someone PC depends on both their specific hardware and also the settings they choose to run at.

That goes without saying. Guess what, with this engine high resolutions can put more of a load onto the cpu just like in a lot of other open world games. That is especally true with eyefinity and NV surround since there is also more geometry to render due to the larger fov.

First of all the game is Horz+ and you only get a greater hFOV when the aspect ratio becomes more wide. Aspect ratio != resolution, you can have multi-monitor configurations which are the same aspect ratio as single screens, so actually you can MASSIVELY increase the screen resolution without the FOV changing even slightly.

In any case FOV only serves to further prove my point and disprove your point. FOV is yet another user choice and can be set independently of screen resolution anyhow, which means that again the performace and CPU/GPU bias of peoples configurations change based on what settings they use.

Did you read what I said? With a typical high end video card that a lot of people here seem to have like a gtx570 you are most likely to be cpu bound when you see the big framerate dips in this game. Just like its been with the past three Bethsoft games on this engine.

Yes I understand what you're saying, and I'm saying whether or not that is true depends entirely on what settings you pick for the game. You can make it CPU limited or GPU limited based on what settings you pick.

There are a lot of people using more powerful graphics cards, multiple cards, and gpu overclocks on this forum. On the otherhand no ones cpu is really going to be much faster. So yeah, chances are you are more likely to be cpu bound in this game and that gap is only going to grow with upcoming video cards.

"Chances are"? What are the chances? Do you have detailed information or statistical data about the averages at what settings the average gamer uses for Skyrim?

 

[PrincessFrosty]

Heh right so I just ran some tests to actually drive this point home:

1) Normal settings for me, 2560x1600 4xAA, 16xAF, Ultra settings
2) Lowered settings, 1280x800 0xAA, 0xAF, lowest settings

Test run inside the guard barracks in Whiterun, vsync off so frame rate uncapped. Specs are in sig

1) ~99% GPU usage in afterburner, ~50% usage on 4 cores
2) ~40% GPU usage in afterburner, ~85% usage on 4 cores spiking up to about 95% usage

There we have it clearly both CPU and GPU limited with exactly the same hardware just different settings.

 

[BababooeyHTJ]

Heh right so I just ran some tests to actually drive this point home:

1) Normal settings for me, 2560x1600 4xAA, 16xAF, Ultra settings
2) Lowered settings, 1280x800 0xAA, 0xAF, lowest settings

Test run inside the guard barracks in Whiterun, vsync off so frame rate uncapped. Specs are in sig

1) ~99% GPU usage in afterburner, ~50% usage on 4 cores
2) ~40% GPU usage in afterburner, ~85% usage on 4 cores spiking up to about 95% usage

There we have it clearly both CPU and GPU limited with exactly the same hardware just different settings.

Indoors? Walk around Riverwood and tell me what you see.

I just read some of your last post. Do you really not understand why a larger fov would effect the cpu? Do you understand what the cpu does when gaming? I'm not going to continue arguing with you since you know it all. Yeah, this is going to be the next Crysis for testing video cards.

 

[PrincessFrosty]

Do you really not understand why a larger fov would effect the cpu?

I never said anything to the contrary, I said that the FOV doesn't necessarily increase with resolution which is the claim you were making, that's a FACT like it or not.

Yeah, this is going to be the next Crysis for testing video cards.

That's a strawman of my position, I never said that nor implied it.

 

[Stupid_Newbie]

Can you run one more test?
Your regular settings at stock CPU speeds, then again at say 4ghz

 

[PrincessFrosty]

I've not actually put a custom cooler on my CPU yet it's just the stock one, so I'm a bit hesitant to start overclocking it right now, but I might give that a shot tonight when I get home. I'd be surprised to see an FPS increase with the GPU running flat out at 99% but it would be nice to confirm that one way or another.

 

[Stupid_Newbie]

If you have a k, oc it!

 

[PrincessFrosty]

If you have a k, oc it!

It's only been like 3-4 days with the new rig, give me chance! Hehe. I think I manged to damage my 5970 from overclocking which has put me off pushing the hardware unless I need to.

A few hours and I'll be home and do some testing on this.

 

[BababooeyHTJ]

Can you run one more test?
Your regular settings at stock CPU speeds, then again at say 4ghz

I already did that with the game maxed out walking around Riverwood which was one of the areas where I saw an annoying fps drops. My findings seem to mirror the Tom's hardware benchmarks. Yes, you are going to be vga limited indoors and in dungeons. That goes without saying. Its the larger drops in the cities that most people are annoyed by which is generally cpu limited.

The fact is that Skyrim loves a strong cpu. Just like Oblivion, Fallout 3, and Fallout New Vegas do. There is no arguing that.

 

[PrincessFrosty]

Sigh.

Well here we go, I ran in "normal" settings for me which is 2560x1600 with 4xAA and 16xAF in all max settings, 95FOV and vsync off, I fast travelled to Whiterun and sat near the front gate, set the time to just after 9 where all the AI wakes from slumber and starts wandering around the city and I watched the female blacksmith do her thing at the forge. The frame rate was about 75fps give or take a little sway either way.

I took 3 screenshots of my desktop, one with the game running but before I alt tabbed back in so you can see the CPU speed set to 3.6Ghz on the Asus overclocking tool, actually it reads 1600Mhz because it's idle and speedstep is lowering it, but you can clearly see the multiplier set to 36.

2nd screen shot was after I alt tabbed into the game and had a prance about in the lowest district of the town and back to the front gate, then alt tabbed back out. It shows GPU usage at 99% solid and CPU at something like 50% across the 4 cores.

3rd screenshot was after setting the CPU to 40x multi and then alt tab back in to game and did the same thing pranced back and forth in the lower district, the frame rate stayed about the same near as I could tell, I wasn't logging fps numbers but it was obvious there was no difference between the numbers greater than the margin of error (2-3fps).

You can clearly see that CPU usage is still marginal and not any different from the first run, but more importantly the video card is still firmly locked at 99% GPU usage, clearly demonstrating that in Whiterun with my hardware and my settings the game is GPU limited.

http://img233.imageshack.us/img233/9908/beforeq.jpg
http://img684.imageshack.us/img684/3418/36ghz.jpg
http://img442.imageshack.us/img442/1304/40ghz.jpg

Whether you are CPU or GPU limited has ALWAYS depended on the relative speed of the hardare and the settings you pick in game. Some games "like" more CPU power relative to other games of their era, but you can't compare CPU and GPU usage inside one game because the usage on the CPU load is mostly static between 2 different test systems where as the GPU load can vary by several orders of magnitude depending on what settings you pick.

That's why professional hardware reviewers push CPUs as fast as they can when reviewing video cards to make sure the bottleneck is on the video card, and when reviewing CPUs they lower all the graphics settings so that game benchmarks are CPU limited. You can push most games either way depending on what hardware you use and what settings you use and Skyrim is no different in this regard.

I think I get what you're trying to say, the game "loves" CPUs in so far that the CPU requirements of the game in relation to the CPU requirements of other modern age games is higher, and GTA IV had the same problem, people with bad CPUs had issues running the game smoothly because they'd upgraded their video cards but neglected their CPUs. The same is true for GTA IV though, whether or not any 1 particular person was CPU or GPU limited was based entirely on what hardware they have and what settings they use.

 

[GoldenTiger]

Indoors? Walk around Riverwood and tell me what you see.

I just read some of your last post. Do you really not understand why a larger fov would effect the cpu? Do you understand what the cpu does when gaming? I'm not going to continue arguing with you since you know it all. Yeah, this is going to be the next Crysis for testing video cards.

The guy you're talking to doesn't seem to understand things at all... no point arguing with people who just can't do so. Your posts are accurate.

 

[PrincessFrosty]

The guy you're talking to doesn't seem to understand things at all... no point arguing with people who just can't do so. Your posts are accurate.

Argumentum ad populum.

'nuff said.

 

[PrincessFrosty]

I already did that with the game maxed out walking around Riverwood which was one of the areas where I saw an annoying fps drops. My findings seem to mirror the Tom's hardware benchmarks.

I've just repeated the tests above by running the entire length of riverwood down the main road during noon, other than the FPS being at about 60-65fps average (about 10fps slower) the results are the same, GPU is locked at ~99% usage and the CPU is no where near fully loaded.

The GPU usage is the most damning evidence, being locked at ~99% usage is a typical case of something being GPU limited, you can argue tasks managers accuracy all you like but untill the GPU usage spikes down below 99% usage then it's GPU limited.

 

[MavericK]

He had a perfectly valid point. Basing your results on an indoor scenario is totally absurd for this game.

Personally, I never get 99% GPU utilization outside in a city. More like 40% and yet the framerate is struggling.

 

[PrincessFrosty]

He had a perfectly valid point. Basing your results on an indoor scenario is totally absurd for this game.

Personally, I never get 99% GPU utilization outside in a city. More like 40% and yet the framerate is struggling.

No he's making sweeping generalizations about what the majority of players experience or use as their settings, without backing them claims up with evidence.

I've actually provided evidence to back my claims

1) I've used settings to shift bottleneck from CPU and GPU to prove that settings make a difference
2) I've used real world settings to prove I'm GPU limited, from inside buildings, inside a main city in the game and now from the wilderness, I've posted clear and concise evidence of this.

I've also made every effort to fill requests regarding overclocking to further provide more numbers on the issue.

All he's done is complain that task manager isn't accurate (not given a source to actually back that claim up) and then when I query him on 2 separate occasions what he would deem a more accurate tool to measure CPU performance I've not had a response.

 

[Stupid_Newbie]

I just through this game ran on 2 cores only so 50% of 4 cores would be 2 cores taxed. I just saw thg, where relatively high settings (yes 1920x1080 isn't bad, but ultra 16x and 4x are pretty hard on gpu), and even then the oc was good for 10fps. I'd think a 570 at those settings should be near 99%. I just have to get office setup and get my desk up so I can game. I didn't buy 950lbs of desk and a 2k machine to air damn it! Work gets in the way

 

[BababooeyHTJ]

Just saw this posted at Rage3d in regards to Skyrim.

i have it but don't oc much
but i did get about 10fps more going from my i7 930 to a 3930k both at stock with the same drivers at 1600p
i will test oc on it later on

There was another interesting link posted there. Check out this techreport Skyrim performance review.

We were surprised by two results on the graph above: the Core i7-3960X's 20% performance lead over the 2600K, and the AMD FX series' lousy showing. The FX-8150 and FX-6120 averaged just 50fps, making them just 1fps faster than the Phenom II X6 1100T, which was also sluggish. By comparison, the i5-750 averaged 56fps, the i3-2120 delivered 57fps and the old i7-920 managed a solid 59fps. Meanwhile, the more modern i5-2500K was sitting pretty with a respectable 67fps -- 1fps below the i7-2600K.

 

[MavericK]

Interesting that they are getting 60 FPS on a stock i7 920, whereas mine overclocked does not get 60 FPS all the time.

It would be nice if someone did a test that showed which settings had the most impact on performance. They claim shadow quality in that article, but no results to back that or see what the difference is in both performance and IQ.

 

[Nerdydave]

It just seems like we are getting off topick I thought he asked how CPU scales in skyrim. I have not tested skyrim but other games I have played OCing did a little bit new CPU did more and new video card always made the biggest difference.

Now I have had games the were little choppy and went smooth after a small over clock.

From personal experance I have never seen my average FPS improve more then 5 FPS.

 

[Nerdydave]

When I get home tonight I will. Check my FPS are with skyrim I am playing on one of my media center PCs and it plays smooth it is an AMD 5000+ 4gigs of DDR-2 memory and a GTX 550 TI

 

[MavericK]

From personal experance I have never seen my average FPS improve more then 5 FPS.

From CPU overclocking? Really depends on the game and the level of overclock. For instance, I got a huge FPS boost in ArmA 2 by overclocking my CPU.

 

[PrincessFrosty]

Interesting that they are getting 60 FPS on a stock i7 920, whereas mine overclocked does not get 60 FPS all the time.

It would be nice if someone did a test that showed which settings had the most impact on performance. They claim shadow quality in that article, but no results to back that or see what the difference is in both performance and IQ.

The article they test in 1680x1050, your sig says you're using a 1920x1080 monitor, having a higher resolution causes a higher load on the GPU with a similar load on the CPU and so tends the game towards being more GPU limited.

The more GPU limited you become the less important changes to CPU speed become when you finally become completely GPU limited changing the CPU speed makes next to no difference.

 

[Blkout]

Skyrim scales with GPU and CPU increases which is not typical of most games. Most games only scale well with one or the other. That being said, if you're playing at higher resolutions like 1920x1080, the game scales better with more GPU power than CPU power s do most games. There's a much larger increase in frames per second by adding a second video card than there is by overclocking your CPU from 3GHz to 4GHz.

 

[karri0n]

No he's making sweeping generalizations about what the majority of players experience or use as their settings, without backing them claims up with evidence.

I've actually provided evidence to back my claims

1) I've used settings to shift bottleneck from CPU and GPU to prove that settings make a difference
2) I've used real world settings to prove I'm GPU limited, from inside buildings, inside a main city in the game and now from the wilderness, I've posted clear and concise evidence of this.

I've also made every effort to fill requests regarding overclocking to further provide more numbers on the issue.

All he's done is complain that task manager isn't accurate (not given a source to actually back that claim up) and then when I query him on 2 separate occasions what he would deem a more accurate tool to measure CPU performance I've not had a response.

I'm curious,

Under any config did you get CPU utilization to drop below 50%? It almost seems(to me) that if it's locked hard at 50%, you might be maxing out both cpu and gpu. It's obvious however, that if your gpu is at 99%, you are being gpu limited. Trying to make an argument contrary to this is rather silly. That doesn't mean that both aren't holding you back.

On a side note, I'm running this game on intel integrated graphics, with a Pentium G620@2.6 ghz . I am running with all settings on ultra, vsync off, fov 90, no AA, no anisotropic, and 800 x 600. I don't have an actual fps rating, but just by looking I can tell it's at least 30 fps with no stutter. Outdoor performance seems a bit better than indoor. A processor way lower on the chain than yours, at a much slower clock speed, who is also devoting some of his power to being a gpu, is able to handle rendering all the actors and other outdoor effects. If, as you say, resolution will not affect cpu(if you have a separate gpu anyway), this puts a bit of a hole in the theory that it's a matter of the rendering of all the outdoor actors being too cpu intensive.