Jump to content

Need Advice Plz, Upgrading


38 replies to this topic

#21 Barbaric Soul

    Member

  • PipPipPipPipPipPipPip
  • 887 posts

Posted 31 January 2013 - 05:11 AM

View PostNirilus, on 30 January 2013 - 11:26 PM, said:

This game is very CPU dependent for...a game. And you will get a performance hit with less than a strong dual core...but my results above, without a doubt, that the video card makes a HUGE difference.

No matter what processor you have IT WILL NOT PLAY WELL ON REASONABLE SETTINGS WITH LESS THAN A GTX400 SERIES.



Ohhhh, so now your saying the game IS infact CPU dependent. Look, the GPU is going to play a major roll in a computer's performance in gaming, no matter what game it is. We know that. It's pretty common knowledge. But that does not mean this game does not require a powerful CPU to be able to max out it's settings. This is not CoD where you can get a bazillion FPS with a C2D E2220 paired with a GTX680(just a example, I'm fully aware of the bottleneck that would happen with that setup).

Oh, and I never said anything about reasonable setting, as that would vary depending on who your talking to. To me, reasonable settings is 60fps all high settings at 1080p. To someone else, it might be a mix of medium/high settings at 1080p. And to someone else, it might mean all medium settings at 1080p. What would be unacceptable to me might be perfectly fine for the next guy. That all depends mostly on how willing a person is to keep thier PC upgraded and running at the high-end of performance. But I have yet to hear of any dual core getting better than 25fps reguardless of which video card it had. That makes this game CPU dependent.

Oh and when DX11 is released, I'm willing to bet anyone with less than a GTX560ti is going to be really dissapointed and mad at PGI.

#22 Catamount

    Member

  • PipPipPipPipPipPipPipPipPip
  • LIEUTENANT, JUNIOR GRADE
  • 3,305 posts
  • LocationBoone, NC

Posted 31 January 2013 - 07:10 AM

One thing I want to emphasize is that MWO is unable to be run smoothly on most CPUs at ALL settings, not just high. With my 7970OC and 3570k, my fps drops to 50 or so, whether I'm on very high or low. 50 is fine, sure, but again, that's with the fastest quad core CPU on the market.

MWO *can* be GPU bound if you have a crappy GPU or turn settings too high, but if one wants 60fps, the game is always CPU bound. That 60 fps number will drop sharply with CPUs further and further blow the $200 mark. The average gaming CPU won't even net 40, no matter what GPU one uses, or what settings the game is run on.


Generally speaking, the GPU is more important than the CPU in a gaming machine, and should get more money (perhaps 1.5-2x more, to be judged on a case-by-case basis), but the CPU is still important, and claiming that MWO isn't CPU bound is incorrect, plain and simple. Unless you're on a stupidly lopsided system, like a 3570 using only its integrated GPU, the game will choke on the CPU the moment it's running at settings that allow the GPU to smoothly run it.

Nirilus, so far you've made 3 patently incorrect statements here. That's okay, we all make mistakes, but incorrect statements in a tech support forum confuse those we're trying to help and undermine the purpose of the forum. Please try to be more careful, and if you are corrected on a wrong statement, by multiple people, don't stubbornly stick to insisting you're correct after almost every one if the most frequent contributors come in to correct you.

Edited by Catamount, 31 January 2013 - 07:13 AM.


#23 Oderint dum Metuant

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 4,758 posts
  • LocationUnited Kingdom

Posted 31 January 2013 - 10:12 AM

View PostVulpesveritas, on 30 January 2013 - 07:45 PM, said:

It pretty much boils down to the following CPU's for this game if you want ideal performance, though it will do 'okay' with lower end quad core CPU's, to get 45fps+ min, you need the following;
Intel:
Sandy Bridge Series i7 desktop
Sandy Bridge-E series i7 desktop
Ivy Bridge Series i5 desktop
Ivy Bridge Series i7 Desktop / high end laptop

AMD:
FX-83xx series, preferably overclocked.


Add Sandy bridge I5's to that.

#24 Nirilus

    Member

  • PipPip
  • Philanthropist
  • 44 posts

Posted 01 February 2013 - 07:38 PM

View PostCatamount, on 31 January 2013 - 05:06 AM, said:

My Radeon HD 5850, which is at least a real gaming GPU (unlike the HD4000) was easily bottlenecked, even by my 3570k, on a combination of medium and high settings, and a 5850 is not an especially fast GPU. I was running a GPU worth 1/3 what my CPU is worth, and the CPU was still the bottleneck. That's with a top of the line CPU. A more average gaming CPU, like a Phenom II X4 or an FX6100, won't run this game at anywhere near 60fps, with any GPU, on any settings.


Are you saying your game is bottlenecked by a 3750k i5? That's what I'm running currently. Definitely not a bottleneck. As I indicated earlier - when I run my cpu monitor, it never runs more than 50% of my processor.

The fact that my system does not bottleneck when I have the same CPU indicates that the difference is GPU based.

"You're implying that the game is either CPU or GPU intensive; in truth, it's both, in abundance. The CPU-dependency is just more important."
No, if you read through my comments, I've implied that it's both. But where we disagree is that I'm pretty certain that a lot of processors can handle the CPU side with some overhead. The pool of video cards that can run the game well, especially at higher resolutions/settings is considerably smaller than the pool of CPUs that can run the calculations side of it.

#25 Nirilus

    Member

  • PipPip
  • Philanthropist
  • 44 posts

Posted 01 February 2013 - 08:29 PM

Let's put up or shut up. Please post a picture of your processor utilization with your processor name clearly visible. (I uploaded this one through http://imgur.com so I could post it in.)

Looky here. I played the game at highest settings @1650x1050 on a 3750k i5. (I moved the window so I could get the screenshot) With 5 firefox tabs and several other programs, like photoshop running in the background...

As you can see, processor utilization is 30% (Full disclosure, it gets up to 50% in the heat of battle). An equivalent dual core could handle this as well. This would include most ivy bridge dual cores (i5 and i7) as well as many i5 and i7 SB dual cores as well. Remember, many dual cores have higher clock speeds than their quad core counterparts.

I would expect you to suffer if you're using a core2duo, as I've already said, but if you have a sufficient dual core, or almost any quad from the last 5 years, then the biggest thing you need to worry about is the vid card.

I would also note that there's no benefit for anything over 4GB of RAM in your system - at least to play MWO.

Posted Image

Edited by Nirilus, 02 February 2013 - 02:37 AM.


#26 Barbaric Soul

    Member

  • PipPipPipPipPipPipPip
  • 887 posts

Posted 02 February 2013 - 06:21 AM

This is what Catamount and myself were talking about with dual cores-

View PostTheFlayedman, on 02 February 2013 - 04:34 AM, said:


I was until my recent upgrade playing on a wolfdale E8600 at stock 3.33Ghz 8GB DDR2 and a GTX480 while it was playable just the performance was bad 15-25 fps in fights at 1900x1200 res all set to low.

I upgraded to a core i5 3570k overclocked to 4.4GHZ 8GB DDR3 with the same GTX480 and now in fights I get 40 - 55 fps in fights sometimes over 60fps but not often with 1900x1200 res everything set to high it looks and plays like a different game almost.


Quote from this thread- http://mwomercs.com/...-e8500-and-mwo/

Me posting that SS is kinda pointless as my CPU is more powerful than your is (2600k/8gigsRAM/single7970), but here you go-
Posted Image

#27 Catamount

    Member

  • PipPipPipPipPipPipPipPipPip
  • LIEUTENANT, JUNIOR GRADE
  • 3,305 posts
  • LocationBoone, NC

Posted 02 February 2013 - 02:15 PM

Nirilus, its obvious that you really don't understand what a CPU bottleneck is.

A CPU bottleneck doesn't mean that your CPU gets loaded to 100%, and the hardware can't physically execute anymore instructions, of any sort. I means that the software has loaded up as many cores as it can, to the maximum of its parallelization, and those cores cannot execute anymore instructions, because, and this is the part you don't seem to understand, the game is waiting for existing instructions to be processed. This will not necessarily show up at 100% load, although it will show up in the Windows task manager as all cores being roughly even.

There are plenty of games that will bottleneck on a quad core CPU and still only load up that CPU to 25%. All that's required is a single core game. Supreme Commander was a big culprit for this a few years back, because it was single-threaded and massively CPU intensive for the Conroe CPUs of the era. How did it show up while bottlenecking on the CPU? With all four cores of a quad core at 25% (if you had a dual core, both would load to 50%, while a single core CPU would load to close to 100%). Why didn't it take advantage of the leftover CPU power? Because there is a limit to parallelization in software, as described by Amdahl's Law. On an octathreaded CPU, a game with very non-parallelized code would show all 8 core at 12.5% (give or take, of course). Note that there are limitations to that; you can see poor threading to an extent. Your own CPU graphs show unequal CPU loading (just like Barbaric's above, though his CPU shows it far more than my own 8 threaded CPU in the other machine, oddly enough).

MWO is not single threaded, so it does a much better job with this. It makes very good use of dual core CPUs, and a fair bit of use of quad core CPUs, and then it just stops being able to use anymore resources from your CPU, even though they're free, because it hits the limits of its software parallelization. Put in laymans terms, the software can't give the CPU anymore instructions until the CPU finishes the instructions it's already carrying out, so adding more cores does no good, because the only way to speed up game execution is to speed up each instruction, which means per-core performance, because ultimately, there is a serial nature to much of how the game is processed (the CPU can't start on instruction B until A is done, or C until B is done; even if your chip has free resources to execute B and C it can't because the game is still waiting on A).

Right now, there is no CPU, anywhere, that is capable of executing MWO's instructions quickly enough to keep up with the demand of the program fast enough to enable 60fps. That is the definition of a CPU bottleneck. It doesn't matter what Taskmgr shows; all that matters is that there is no CPU, in existence, that enabled 60fps at stock clocks (at around 4.4ish the 3570k pretty much gets there, from what I've seen with my CPU). THAT is the bottleneck.


So when you say

View PostNirilus, on 30 January 2013 - 03:13 PM, said:

The majority of cpus out there have more than enough power to handle the CPU dependant portions of the game.


you are patently incorrect. Right now, there is no CPU, anywhere that can execute this game smoothly, let alone any CPU with "more than enough" power to execute the game smoothly, let alone a majorityof CPUs that can execute this game smoothly. The 3750k is the only chip that even comes close, and even it can't maintain 60fps with any amount of action, regardless of GPU, and regardless of setting. That's the CPU bottleneck. My 7970 can render the game at max settings at 90+fps, and does regularly, until anything starts happening, and then the CPU holds it back to 45-50, hence the CPU is bottleneck that GPU. If I turn the game down to its lowest settings, those minimum framerates do not change, demonstrating that the CPU is, in fact, the bottleneck.

The telltale sign of any CPU bottleneck in a game is the same minimum framerate at all graphical settings, showing that the GPU is rendering the same amount regardless of how hard it's worked. This is exactly what we see in MWO so please, unless you have some actual evidence to the contrary, that's it.

This concept is really nothing new to any gamer who's even been around for the last decade, back when we were having the single core vs dual core, and later dual core vs quad core debate. Back that few short years ago, people easily understood that a game could max out its ability to run on a CPU, and be bottleneck without actually loading up that CPU to anything close to 100%. Maybe you just aren't old enough to remember, but that was at the core of the multithreading debates. It was widely understood that just because the task manager didn't report 100% CPU load, didn't mean the game wasn't bottlenecking; it just meant it was bottlenecking because it was poorly threaded. This is why, for several years, people stuck with higher clocked dual cores over lower clocked quad core iterations of the same architecture, because we knew the game wanted per-core performance, not multicore performance.

I'm sorry that was has been an obvious and ubiquitously understood concept is going over your head. I really don't know how many more ways I can explain what's been obvious to PC gamers for a decade.

Edited by Catamount, 02 February 2013 - 03:21 PM.


#28 Catamount

    Member

  • PipPipPipPipPipPipPipPipPip
  • LIEUTENANT, JUNIOR GRADE
  • 3,305 posts
  • LocationBoone, NC

Posted 02 February 2013 - 03:07 PM

And just to show, once and for all, that MWO is, in fact, CPU-bottlenecked on a 3570k (using my spare 5850; the 7970 is out on a warrantly claim)

Here's a screenshot of the game, at straight medium settings, at 1920x1200


Posted Image

Framerates get as high as 55ish on these settings, and at this resolution, but as soon as the team starts moving out, it drops to 45ish. Is the CPU causing it to drop that low, or the GPU?

Well, here's the game running at all low settings, at 1024x768

Posted Image

At these settings, the GPU will start churning out over 200fps, but again, once action starts, even when it's just the team moving out, the minimum fps drops right back down the mid 40s. Since the GPU can get better framerates than this at just under three times the resolution, and with much higher graphical settings, clearly that's not why this scene -which isn't even having anything particularlylintensive be rendered- is dropping down to the mid 40s again.

The GPU is not the bottleneck here, because it's not being taxed to draw this scene at 46fps, ergo, the CPU is holding it back to these framerates.


If anything, the real mystery is why I'm having to show this. We've been talking about how MWO bottlenecks on high end CPUs for months now.

Edited by Catamount, 02 February 2013 - 03:09 PM.


#29 Nirilus

    Member

  • PipPip
  • Philanthropist
  • 44 posts

Posted 02 February 2013 - 05:39 PM

I'm seeing different results. Clearly, my shots have a lot more going on than your screenshots, so if I am processor limited, I should get lower FPS than you.

As shown by barbaric's shot, with 3 idle cores, his cores are definitely not overthreaded in any way shape or form. by the fact that they do not max percentage, then they definitely have headroom for each individual thread. I know apparantly you think differently than that, but we'll get to that later.

See my screenshots below on my computer, a 3750k. If I was processor limited, I would get identical (or worse because my shots have more pew pew pew,) results than you. Please note that my system is currently limited to 60/75 FPS max (system vsynced) which is why you won't see obscene framerates (I really just don't feel like digging around in my settings and unlocking it again), but 55 fps is the lowest I've seen on lower settings.

After we get through this issue, I will respond to your ideas about the task manager, and such. But first things first. Data.
Posted Image

Posted Image

Edited by Nirilus, 02 February 2013 - 05:54 PM.


#30 Flapdrol

    Member

  • PipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 1,986 posts

Posted 02 February 2013 - 05:58 PM

That's just 2 screenshots.



This is an entire game, I recorded on low res and 1 fps to not load up the cpu. Really low gpu load halfway, one of the cpu cores is always loaded very high.

i5-750 @ 4.2 Ghz and a gtx470.

the most annoying thing is minimum fps isn't really effected by settings, not even the model/terrian detail stuff. Average fps goes up enormously but minimum fps is dominated by how many mechs are in view. With 8v8 I can stay above 30 but if 12v12 is introduced it'll be an absoluted disaster.

ontopic:
single core 3 gig intel (over clocked)
Gforce 210 pci express video (512mb )

both are not at all usable for running this game. So both need an upgrade.

Edited by Flapdrol, 02 February 2013 - 06:02 PM.


#31 Catamount

    Member

  • PipPipPipPipPipPipPipPipPip
  • LIEUTENANT, JUNIOR GRADE
  • 3,305 posts
  • LocationBoone, NC

Posted 02 February 2013 - 06:02 PM

Nirilus, your logic doesn't even remotely follow. First off, CPU intensity is dictated by more than simply what can be seen on screen (your screenshot shows a lot fewer mechs moving around you than were in my screenshots). Rendering overhead is not the only source of CPU load. Secondly, you're assuming that two systems will perform exactly the same, which is a rather wild assumption given gross performance inconsistencies we've seen in MWO.

I have already demonstrated, amply, that the 3570k bottlenecks the system through a framerate comparison that shows the game dropping to low framerates in non-CPU bound conditions. Maybe you think your 3570k isn't bottlenecking the game, and who knows, maybe it's a mystical 3570k that isn't constrained to the normal physical limitations of a stock clocked chip of that model, but the rest of us have been seeing reports of CPU bottlenecking when trying to hit 60fps since long before you ever joined these forums. Everyone on these forums already knows MWO bottlenecks on a 3570k, and we've been discussing it for ages, especially since it was shown by Russian reviewers (but they simply confirmed what we already knew long ago).


If you want to be the one guy who obstinately refuses to accept that, then so be it; it's only one of many incorrect things you've asserted on this thread, and I'm not going to indulge you in your clear desire to continue arguing for the sake of arguing. As long as everyone else on this forum understands how easily MWO becomes CPU-bound on most systems, I'm satisfied with that. You can go off and believe whatever else you please, based on whatever non sequitur you please. The the regular contributors here will continue advising based on actual fact.

The only way you'd ever present anything compelling would be a fraps benchmark run, through a full battle, with minimum fps at 60fps or significantly higher (with a stock clocked 3570k), because so far, that's something no one else here has reported being able to attain.

Edited by Catamount, 02 February 2013 - 06:09 PM.


#32 Catamount

    Member

  • PipPipPipPipPipPipPipPipPip
  • LIEUTENANT, JUNIOR GRADE
  • 3,305 posts
  • LocationBoone, NC

Posted 02 February 2013 - 06:33 PM

View PostFlapdrol, on 02 February 2013 - 05:58 PM, said:

That's just 2 screenshots.



This is an entire game, I recorded on low res and 1 fps to not load up the cpu. Really low gpu load halfway, one of the cpu cores is always loaded very high.

i5-750 @ 4.2 Ghz and a gtx470.

the most annoying thing is minimum fps isn't really effected by settings, not even the model/terrian detail stuff. Average fps goes up enormously but minimum fps is dominated by how many mechs are in view. With 8v8 I can stay above 30 but if 12v12 is introduced it'll be an absoluted disaster.

ontopic:
single core 3 gig intel (over clocked)
Gforce 210 pci express video (512mb )

both are not at all usable for running this game. So both need an upgrade.


You seem to be getting a lot of what the rest of us are. Quad core CPUs shows one really loaded core, and then three modestly loaded cores. Unfortunately, while an i5-750 is still an okay gaming CPU, MWO is the exception there.

I'm not surprised that it can't keep up with a 470, hence why turning settings up and down doesn't help your FPS (because it's your CPU, rather than your GPU, choking). This bodes poorly for a lot of users because it means that Phenom II X4s, and similarly performing budget chips also aren't going to cut it, which is exactly what I saw with my Phenom II X4 965 (which used to be a competitor for the i5-750).


That's why I bought an i5-3570k. It may not keep up with even my 5850 much of the time, let alone my 7970, but at least 45-50 minimum fps in really intense furballs is adequately smooth. My Phenom II went down to below 30 fps sometimes, and my laptops i7-720qm goes to 20fps. Having a CPU that can almost keep the game completely smooth is at least better than not. OTOH, when DX11 comes in, it will massively alleviate CPU strain, if Crysis 2 is any indication, so that system would eventually net decent performance (the 470 really isn't a bad card at all, even these days), but if you're already looking to upgrade anyways, and don't want to wait for some unspecified far off time for MWO to be smooth, then I'd start with the CPU first, then replace the GPU, since the former is holding you back far more.

Edited by Catamount, 02 February 2013 - 06:37 PM.


#33 Nirilus

    Member

  • PipPip
  • Philanthropist
  • 44 posts

Posted 02 February 2013 - 09:29 PM

Catamount, I have 3 mechs fully in view in one screenshot. Yours does not. I have more mechs in my screenshots. You can also see in the rader in that screenshot, that there are lots of mechs in the vincinity offscreen as well. A video doesn't change that.

The point is, it doesn't matter how much proof I offer, you won't accept it. I have the same processor and I'm getting ~20% better performance with a better vid card than you are getting with a reasonably decent vid card.

IF we are experiencing the same weak link with the same identical part, then that limiting factor will be the same. But the fact is my computer has a better GPU, and likely better throughput. However, none of that helps my processor process things faster. It Eliminates OTHER bottlenecks.

"Maybe you think your 3570k isn't bottlenecking the game, and who knows, maybe it's a mystical 3570k that isn't constrained to the normal physical limitations of a stock clocked chip of that model'

I'm pretty sure that the difference we see with two identical stock processors indicates that you haven't proved your point. No amount of condecention will change that.

"The only way you'd ever present anything compelling would be a fraps benchmark run, through a full battle, with minimum fps at 60fps or significantly higher (with a stock clocked 3570k), because so far, that's something no one else here has reported being able to attain."
I clearly stated that it can drop to ~55fps. I certainly believe that this is a graphically intensive game, and that SLI'd high end video cards would be able to do 60fps. As far as me, I'm tired of screwing around with you on this. I'm just not invested enough to go start installing programs and setting them up so I can prove you wrong beyond a shadow of a doubt.

------
You have clearly shown that you do not understand how hyperthreading/multicores and how task manager core usage works, your comment that an octocore would show 12.5% on each of 8 cores if only one core is loaded is completely ridiculous (You looked at barbaric's screenshot, right)
-----
"You seem to be getting a lot of what the rest of us are. Quad core CPUs shows one really loaded core, and then three modestly loaded cores. Unfortunately, while an i5-750 is still an okay gaming CPU, MWO is the exception there."

And this is the kernal of truth that I agree with you on. The single thread performance and total performance are the two factors here. The original poster was choosing between very old processors (Core2duo and core2quad) - neither of which would mark very well on either account (I recommended the quad in that scenario.) but there are newer hyperthreaded dual cores that should be able to hit both marks. I got the 3570k i5 because it is a good balance at a good price.

It may have been different before the bug patches, but even in firefights, I am not overloading a single core.

------
I have also played on a phenom (I don't know which one) with a gtx460 (My brother's desktop.) It played acceptably, and certainly got higher than 30fps at mid settings and resolutions.

I've maintained that the load on my cpu is high FOR A GAME - 50%(80% on 1 core) utilization on a 3750k is pretty high - but I have yet to see someone with less that a gtx400 get acceptable framerates on anything other than minimum settings.

You underestimate how much load is on the GPU.

Edited by Nirilus, 02 February 2013 - 09:30 PM.


#34 T0rmented

    Member

  • PipPipPipPipPipPip
  • Knight Errant
  • Knight Errant
  • 317 posts
  • LocationEngland

Posted 02 February 2013 - 10:31 PM

i get >60 fps with an i5 2500k (oc to 4.4gHz) 2xfactory oc gtx660ti and 8GB 2133MHZ ramm. I know only one gfx card is in use even with forced sli, but once they decide to allow those who spend on their systems the gift of their own sli i wont ever have to stoop below 55fps (typically my low fps) ever again.
I should point out my eye meltingly fast fps is achieved with all settings as high as they can go @ 1920x1080.
I'd also like to point out that ive always had better than 50fps average since closed beta and i swapped out my old 460s since then.

Edited by T0rmented, 02 February 2013 - 10:38 PM.


#35 Flapdrol

    Member

  • PipPipPipPipPipPipPipPip
  • The 1 Percent
  • The 1 Percent
  • 1,986 posts

Posted 03 February 2013 - 01:50 AM

sli wont help with a cpu bottleneck. You could run with more antialiasing or at a higher res with it though.

#36 Oderint dum Metuant

    Member

  • PipPipPipPipPipPipPipPipPip
  • Ace Of Spades
  • Ace Of Spades
  • 4,758 posts
  • LocationUnited Kingdom

Posted 03 February 2013 - 03:32 AM

View PostNirilus, on 02 February 2013 - 09:29 PM, said:

Catamount, I have 3 mechs fully in view in one screenshot. Yours does not. I have more mechs in my screenshots. You can also see in the rader in that screenshot, that there are lots of mechs in the vincinity offscreen as well. A video doesn't change that.


You should probably look at his screenshots again, and use fingers to count (specifically the river city one).

The game is highly CPU dependent, PGI acknowledge its CPU optimization is poor. The game runs on DX9 also CPU orientated which means that modern GPU's are not being used to anywhere near the level they should be, DX 11 will change that.

Until DX11's implementation that will be the case, the game is CPU bottlenecked at this time period.

#37 TheFlayedman

    Member

  • PipPipPip
  • Survivor
  • 76 posts

Posted 03 February 2013 - 07:08 AM

My cpu utilisation is at 100% on 4 cores during gameplay with a i5 3570k cpu. Not sure what my gpu utilisation is thou although i do know it gets heated up to 88C and the fan ramps up to 4k rpm :/ so it must be fairly high

Edited by TheFlayedman, 03 February 2013 - 07:10 AM.


#38 Catamount

    Member

  • PipPipPipPipPipPipPipPipPip
  • LIEUTENANT, JUNIOR GRADE
  • 3,305 posts
  • LocationBoone, NC

Posted 03 February 2013 - 07:32 AM

DV, Nirilus is going to argue this, for the sake of arguing it, until hell freezes over. Even when I completely remove the GPU from the equation by running at 1024x768 on minimum settings, with the GPU being able to churn out over 200fps, he won't accept that it's the CPU dragging the game back down into the 40s and 50s, despite the fact that similar FPS across different graphical settings being a CPU bottleneck is basically computing 101.

The rest of us know the game is CPU-bottlenecked in most cases (and that no stock PU can drive the game at 60fps, even when the GPU can), and are advising as such on this forum, and that's enough.

View PostTheFlayedman, on 03 February 2013 - 07:08 AM, said:

My cpu utilisation is at 100% on 4 cores during gameplay with a i5 3570k cpu. Not sure what my gpu utilisation is thou although i do know it gets heated up to 88C and the fan ramps up to 4k rpm :/ so it must be fairly high


I've noticed this game can definitely get GPUs very hot, which can depend as much on how the game is loading up the GPU as the actual GPU load. Kerbal Space Program, 3DMark11 and Furmark all get my GPU up to 100% across all four cores, but there is a wide, wide disparity in the temperatures each produces ;)

This game's hot-running tendancies were actually what allowed me to spot the need for a warranty claim on my 7970, which completely destabilized, at almost any GPU load, above a little over 60C (hard-to-spot problem initially, because I live 4200 feet up, and it's rather cold in here right now even at the best of times; MWO was the first game to pump out enough heat long enough to where the case's ambient temperature and GPU load combined were enough to get the card warm).

I haven't looked at CPU utilization in MWO yet, but again, the game is wildly inconsistent (another thing we've been discussing on these forums forever). I'll give it a lookover later tonight, and when my replacement 7970 comes in.

Edited by Catamount, 03 February 2013 - 07:57 AM.


#39 Az0r

    Member

  • PipPipPipPipPipPip
  • 343 posts

Posted 03 February 2013 - 07:51 AM

Just a quick FYI, a game doesnt have to completely max one core to 100% to be "bottlenecked" by the cpu. If increasing the compute power of the GPU doesn't result in higher FPS but increasing the per clock cycle compute power of the CPU OR increasing the frequency of the CPU does, then the game IS cpu "bottlenecked" Just like in MWO.

Edited by Az0r, 03 February 2013 - 07:52 AM.






1 user(s) are reading this topic

0 members, 1 guests, 0 anonymous users