I completely agree with that, the magic numbers seems to be about a 3.4Ghz clocked sandybridge or a 3.8Ghz wolfdale (I've got both so I've tried with them out of curiousity). One of those will see 100% usage on one core, clocking higher seems to just use smaller percentages of that first core, my testing involved clocking this quadcore machine down to 2.6, 3, 3.4, 4 and where I normally play at 4.6 and at 3.4 and up there were no FPS gains on the 2600K. With the wolfdale I tried at its default 3.2 then 3.6, 3.8 and 4.2, 3.8+ the FPS wasn't gaining and the first core was no longer consuming 100%.

Conspiracy theory time: What's the chances that Skyrim's CPU utilization was tuned to what the default clockrate of what was then the strongest desktop CPU?

ED: Oh and to ensure consistency and all that I was testing using very low graphics settings just to ensure the FPS was relying on the CPU's side of the house as much as possible.

Obviously with vsync off ?

If the GPU isn't maxed, vsync is off, and increasing CPU clock speed doesn't increase performance - what's the bottleneck?! Memory bandwidth? doubtful.

As to the tuning; I'd wager it was driven more by the capabilities of a certain Microsoft console

All I know is my stock i5 2500k eats the game nomnomnomnom no problemo. Love this processor

3.4? but I only has 3.3 should I OC my ASUS BIOS makes it like so easy(or so I have read like a million times). Also anyone know a good guide or reference for Overclocking as far as what I can bring the clock speed up to with just a stock fan on my processor inside a 932 HAF case that keeps everything pretty cool at stock speed.

Aye vsync off for sure. With those conditions though its not like the CPU needs to reach 100% and ideally if nothing important is going on an app shouldn't be gunning to max out the CPU for no good reason. There's a reason your web browser don't consume 100%, it doesn't have to. Same with the CPU side of Skyrim, to run the AI and all the rest in that dragonreach->tree view for instance its only got 'x' things it needs to accomplish and at some point there's just nothing more for it to be doing so if your processor can manage that without capping out then that's it, that's the highest need and it'll only consume what it needs to reach that point. The only way to make your CPU work harder is to introduce more things for the engine to be doing, such as my stress test of loading a few dozen NPCs which in turn generate more work for the engine through indeed their pathing needs and what not. Oh and screwing with the sgtm value can play havok with CPU needs too, try making eveyone running around 10x normal speed and you'll see the proverbial manure vs fan

Mind you, as Tesval and Skyboost have both proven: there is much left to be desired with Skyrim's usage of the CPU in terms of efficiency but that just means less CPU time devoted once that apex is reached.

Nope, you have 3.7Ghz

I played around with the clockrates of my 2600K using the generic fan and my after market heatsink and I can say you can pretty safely OC to about 4Ghz without resorting to higher then stock voltage increases and it'll stay within about 70c under full 8 vcore load (transcoding video for instance). You can experiment with the asus suite tool that likely also came with your board (very likely same prog that came with my P8P67 WS Revolution). It'll be in your tray if you've got it loaded, just open the Suite II, under tools go to Turbo VEvo, the TurboV tab and 'CPU Ratio' and just bump it up to 40x and give that a whirl. With this family of CPUs it seems to be north of 4Ghz that you need to toy around with voltage settings to maintain stability.

ED: Oh and north of 4Ghz is also where the voltage + clockrate generates heat in quantites that the stock heatsink gets into 'danger danger' land.

I was safely able to hit 4.5Ghz on "Auto" voltage. Above there I had to manually enter in voltage tweaks. Though I never tried overclocking with the stock hs/f. Its still sitting in a closet unused.

Yah that auto voltage does some crazy stupid things so I only go manually at this point, caught its decision to make the base voltage 1.6 once...wat! Btw you've gunning at 1.5v or higher eh? With this heatsink (thermalright U120) I don't bring it over this 4.6Ghz@1.45v very often and when I do I make sure it won't use all 8 cores else yah it starts reaching into the 80c+ fast. That Noctua is a monster, in hind sight I really should have just picked one of those up instead of recycling the same HS I'd used with my wolfdale (at the time the 120U was best rated in a roundup vs OC'ed Q6600s so figured it could take it). That thing is also shorter and probably would have let me use the 140mm fan properly on this case's side panel too.

Wot?!?
When vsync is off you can always render more frames.
It's not like the game operates at a fixed timestep; it uses elapsed time per frame & interpolation, more frames = finer grain interpolation.

Granted the physics engine is broken when per frame time delta gets too small, and rendering at framerates beyond vsync is not normally desirable.

However, the point is, the framerate should always increase when the bottlenecking factor is eliminated.
If you are saying that your GPU isn't the bottleneck, and increasing your CPUs clockspeed doesn't improve performance either, then the bottleneck must be elsewhere.
Whether it's PCIe bus bandwidth, memory bandwidth, blocking disc I/O, etc; there must be something else that is preventing further framerate increases.

Oh, it was the GPU's domain that was then the bottle neck for sure at that point even with the super low settings. We are talking closing in on 300fps at the dragonreach->tree view here and about 1200FPS if the game's paused. At this point and again bringing up Skyboost: I have a feeling the real bottle neck is the game engine's lack of optimizations itself, if two guys and a few memory remaps can get that much out of the CPU, gotta imagine what could be done by some genuinely capable of people on the video engine's side too. I have a feeling there's a lot left to be desired on that front, since agreeably there should be no real high point to FPS without at least one of the components getting 100% usage and clearly in that setup neither of my components were approaching anywhere close to saturation, in fact quite the opposite.

Eh? CPUID and My System Screen say 3.3 where you thinking of the i7 2600k?
http://i39.tinypic.com/2d5mxt.jpg

Thanks just wanna bump it up a little nothing crazy...as far as I read this processor is very OC friendly and I have even seen a few videos of people getting 5Ghz with extreme cooling setups.

Look again. What's the actual core speed shown in CPUZ (it's in Mhz, but I'm sure you're capable of working that out)

See that line in CPUid that says 'core speed'? That's your real clockrate, this family of CPUs have a 'turbo mode' which is triggered by anything such as running that video, a 2600k turbos to 3.8ghz and I guess a 2500k turbos to 3.7ghz. Yah I know, Intel really could have done a better job of presenting what the default setups actually do. So that app I mentioned earlier in the thread as well: You can use that to bump the turbo up to 4ghz and try it.

ED: but don't get me wrong, the turbo concept is a great one and I'm glad to see them going back to this sort of thing for desktops. My CPU typically clocks itself at 1600mhz when I'm farting around forums and what not but when I launch something it'll automagically go up to its turbo mode 4.6Ghz.

Sweet thanks guys

now to find out if my p67 has that BIOS not saving changes issue....haven't touched it yet.

Wrong.... as stated numerous other factors included..... it's the same failure that occured when people stated that a faster frequency single core cpu was better than a slower dual core.... Initially this is true.. long term how ever is NOT...


It's the i5 Sandybridge... references currently available modern technologies.. not older sockets or anything...


Turbo boost from my testing doesn't kick in with Skyrim.

Using the currently available dual core and quad core from the same manufacturer with nearly identical cpu die's aside from the obvious differences, Actually further testing, Larger cache has very little to do with skyrims performance levels. From my own testing... you could have a i7 3960K... (i do)... and make the game run exclusively at almost identical frequencies as the smaller cache version of the i7's... There are obvously other factors.. but cache is not entirely a huge factor in skyrim.

Your not going to find many dual core cpu's that have the same cache as quad core....


This is irrelevant to the discussion when the OS/memory/GPUs are all using the EXACT Same things...


If you must know, testing on an Asus P8H67-M Evo rev 3.0 with an ATI Radeon 6950 2gb running as a 6970 2gb..... coupled with 2x4gb DDR3 PC16000 (2000mhz) running on both systems at 1333mhz identical timings and settings.

OS is windows 7 x64 sp1 fully updated using Seagate Momentus XT 7200RPM 32mb cache Hybrid drive
Outputting 1080p display
Everything is EXACTLY the same.


confirmed and tested


confirmed and tested


Excluding the OCTO-Core cpu...... confirmed and tested


Also confirmed and tested ... VERY little difference all within errors of margin


Exactly

Of course other factors contribute, like the CPU architecture. However the game uses two cores well, that's it. However of the same CPU architecture, a 3.6GHz dual core processor will run Skyrim better than a 2.9GHz quad core.

Yes. Exactly. That proves that you're wrong. If you disable two of the cores on a quad core processor you will get virtually the exact same performance as with the extra cores enabled. Or in other words the extra cores aren't helping at all.

Until you do something more then stand in the same spot watching the clouds go by. I've tested this, Skyrim loves it some CPU and can use 8 vcores if it needs to and garaunteed if you tried the same stress test as I posted the results of earlier in this thread with any dualcore (any dualcore, any family and any real speed) it would have been much uglier. But you know, continue to argue that a computer with half the ability isn't any less capable then one that isn't hamstrung.

Exactly in that while disabling the 2 extra cores.. there is very little differences between it and a dual core running essentially the same speeds and so forth.

When you bump the dual core up to a higher speed and then enable the other 2 cores on a slower speed quad core..... the quad core will take a bit of a bump up and surpass the "faster" dual core.

I was not saying that dual core vs quad core with EVERYTHING else exact has no differences.

Sorry... but a Quad core will win at a lower frequency than a dual core at a higher frequency EVEN if it has hyper threading.

Really there shouldn't be an arguement.

This is like argueing over a 2 cylinder engine having as much capabililities and power as a 4 cylinder engine.... Provide with the same load in which both a 2 and 4 cylinder engine will give you the "speed" you want... however accelerating to that level is another matter entirely... top it off sustaining is another matter too... the 4 cylinder engine is going to maintain MUCH better.... add a little extra load and the 4 cylinder will be pretty much untouched by that load.. where as the 2 cylinder will exponetially fail to keep up to the point where it will start to stall consistantly or just be unable to do the task efficiently at all.

Sure you may get a little more mileage on that dual core initially, but once you have the throttle opened up and maxed out, the quad core will blow by in efficiency fairly quickly.

There really is no arguement other than cost to lifespan benefits.... can you afford a quad core? if yes.. buy it.... if no? look at the best dual core you can manage but suffer the shorter lifespan and capabilities of the cheaper price you pay.

Shorter lifespan? What? Dual cores don't have shorter lifespans than quad cores Likewise in Skyrim a higher frequency dual core with HT will be better than a lower frequency quad, all other things being equal (cache, architecture etc.). In other games it might be different, but for Skyrim it's quite clear.

Like mentioned, that's windows trying to split the load equally between cores. That's not the game doing it.

Keep reading to the next screen shot and explain that one then, like that guy I bet you'll just forget you ever said that.

I've seen both screenshots, and the replies. I fail to see how that guy "just forgot he ever said that".