Login

russian armor

CoH2 GPU/CPU usage

28 Jan 2016, 11:25 AM
#61
avatar of kamk
Donator 11

Posts: 764



Ok thanks for your input guys, I guess I'll just stick with the rig I have then with a GPU that doesn't max out, with a processor that ranks fairly highly for this type of game (single thread) and already using NVIDIA which seems to work better than AMD.

Overclocking and turning off physics then I guess if I want to try something, and maybe trying AA off. What does the setting physics do?

Oh, don't get me wrong here, you will get fancier graphics with a fancier GPU.
My point was mainly about your upgrade choice. Just some food for thought: too much money spent, for such little improvement.
i.e. 960 4GB ~230+, 970 "4"GB ~330+ (- ~30 for RotR atm)
- should have elaborated there, sorry for my multi-tasking fail.

It's AMD CPUs who suffer with COH2, not the GPUs.

Physics setting controls stuff like ragdoll effects etc. (not all physics per se). Some see decent improvements - depends on setup, as usual.
Edit: maybe some dev can answer you this question *looks to someone called Cuddletronic*
28 Jan 2016, 14:52 PM
#62
avatar of Trubbbel

Posts: 721

Ok, yes that was indeed what I was after: how much improvement for money do I get? Is it worth it? And the costs seems to outweigh the improvements.
29 Jan 2016, 01:24 AM
#63
avatar of Cyanara

Posts: 769 | Subs: 1

Ok, yes that was indeed what I was after: how much improvement for money do I get? Is it worth it? And the costs seems to outweigh the improvements.


You are unlikely to see any significant improvements in this game. It would be a large waste of money.

Almost any other game though, it would probably help a lot.
13 Jan 2017, 09:25 AM
#64
avatar of Cyanara

Posts: 769 | Subs: 1

Question: Has anyone ever found a way to increase their CPU utilisation? For example, my Core i7 960 seems to max out at 60% utilisation these days. It used to be 40%, so maybe a patch improved something? Still a lot being wasted though.

Interestingly, I just tried a new i5 6600k system and that was using a much more reasonable 80% of the CPU. However it seemed to produce the same frame rates as my i7, do I'm kind of confused about that.

My thoughts are along the lines of whether maybe better RAM throughput can feed the CPU better, DDR4 in particular.
14 Jan 2017, 04:56 AM
#65
avatar of kamk
Donator 11

Posts: 764

Are you sure the 60% apply to COH2?
I see the same 2.x cores being used, with the usual 30-40'ish% overall for Quadcore+HT.

How did you test the two?
You could just set different Ram profiles in BIOS, and test with the 6600k
14 Jan 2017, 07:17 AM
#66
avatar of Cyanara

Posts: 769 | Subs: 1

jump backJump back to quoted post14 Jan 2017, 04:56 AMkamk
Are you sure the 60% apply to COH2?
I see the same 2.x cores being used, with the usual 30-40'ish% overall for Quadcore+HT.

How did you test the two?
You could just set different Ram profiles in BIOS, and test with the 6600k


If you use the processes tab in Task Manager or Process Explorer you can sort them by CPU usage. It will show you both the total CPU usage and how much is being used by different processes.
14 Jan 2017, 10:59 AM
#67
avatar of kamk
Donator 11

Posts: 764

jump backJump back to quoted post14 Jan 2017, 07:17 AMCyanara

If you use the processes tab in Task Manager or Process Explorer you can sort them by CPU usage.

You don't say :snfPeter:

But seriously, same 30-40% range for me, as mentioned above. I guess HT was ON as well? How long was your test?
14 Jan 2017, 11:14 AM
#68
avatar of Cyanara

Posts: 769 | Subs: 1

jump backJump back to quoted post14 Jan 2017, 10:59 AMkamk

You don't say :snfPeter:

But seriously, same 30-40% range for me, as mentioned above. I guess HT was ON as well? How long was your test?


Sorry, was/am very tired. I actually turned HT off. Yet to try again with it on. Might be relevant.

I didn't benchmark. I just loaded the same maps and observed the starting area FPS in most cases. Graphics and other settings all on lowest to avoid bottlenecks.
14 Jan 2017, 11:25 AM
#69
avatar of kamk
Donator 11

Posts: 764

jump backJump back to quoted post14 Jan 2017, 11:14 AMCyanara


Sorry, was/am very tired. I actually turned HT off. Yet to try again with it on. Might be relevant.

I didn't benchmark. I just loaded the same maps and observed the starting area FPS in most cases. Graphics and other settings all on lowest to avoid bottlenecks.

It is. Without HT your percentage will only use the remaining 4c/4t, and therefore will show higher load.

The integrated benchmark is quite decent to test worst case scenarios. Did you OC the i7?
14 Jan 2017, 11:38 AM
#70
avatar of Zyswen

Posts: 149 | Subs: 1

coh2 just bad optimized.
17 Jan 2017, 17:30 PM
#71
avatar of kamk
Donator 11

Posts: 764

jump backJump back to quoted post14 Jan 2017, 11:38 AMZyswen
coh2 just bad optimized.

Well, that part we definitely know. Question is: how can one battle / improve it?
17 Jan 2017, 18:19 PM
#72
avatar of Zyswen

Posts: 149 | Subs: 1

jump backJump back to quoted post17 Jan 2017, 17:30 PMkamk

Well, that part we definitely know. Question is: how can one battle / improve it?

soliton is relic fix, or buying titan X. Kappa
17 Jan 2017, 18:22 PM
#73
avatar of kamk
Donator 11

Posts: 764

Titan X won't help, we already know that :P
17 Jan 2017, 19:09 PM
#74
avatar of ferwiner
Donator 11

Posts: 2885

If it is going at 40% CPU and 30% GPU you are bottlenecked on CPU. The 40% which you are talking about is 40% of power of all of your (I assume) 4 cores. This is one thread going at full speed on one core additionally delegating some minor work to other ones. So actually it is using 100% of what your processor can do for that particular aplication.

The reason is that it is almost impossible to create a multithreaded simulation that can be reproduced. It will bottleneck to one thread sooner or later, whatever you do. This is the case with all RTS games and most other genres.

Also, for some strange reason, windows tends to show more than 25% percent for single threaded programs. When I run a single threaded program that actually does same thing for a long time, windows always shows 30-40% of CPU usage on that program for me if all other cores are free.

As semi related note - intel introduced new i3-7350K unlocked double core processor. It is aimed to be bought by games becouse even though for last 10 years almost all gamer setups featured quad core processors game developers still didn't came up with a way to use more than 2 efficiently. And even though functional programming is becoming more and more popular it is not a subject to change for quite some time becouse of the problems I described before.

Summing up, if you are aiming to improve your fps don't neceserily blame coh2, even though it could be better optimised for sure, but overclock your CPU, even if that would mean you need to turn some cores off.
18 Jan 2017, 13:53 PM
#75
avatar of kamk
Donator 11

Posts: 764


Summing up, if you are aiming to improve your fps don't neceserily blame coh2, even though it could be better optimised for sure, but overclock your CPU, even if that would mean you need to turn some cores off.

Thing is, plenty of other (older) games may not use more than two cores efficiently, but they rarely ever have to, nor struggle as much as COH2 does.

So, the question here is: does it make any sense to need way above average single-thread performance, just to get a fluid and stable 60+ FPS experience? Of course, if you run some low IPC potato, COH2 may not be blamed, but even high end CPUs struggle with this game.
18 Jan 2017, 20:01 PM
#76
avatar of ferwiner
Donator 11

Posts: 2885

jump backJump back to quoted post18 Jan 2017, 13:53 PMkamk

Thing is, plenty of other (older) games may not use more than two cores efficiently, but they rarely ever have to, nor struggle as much as COH2 does.

So, the question here is: does it make any sense to need way above average single-thread performance, just to get a fluid and stable 60+ FPS experience? Of course, if you run some low IPC potato, COH2 may not be blamed, but even high end CPUs struggle with this game.


It is true that there must be some feature that is done inefficiently or simply shlouldn't be there in the first place. I would recommend playing with graphical settings a bit, becouse in my experience the problem only exists for people who buy new machine and automatically turn everything on max, which would seem reasonable for a 4 year old game but actually is not.

One thing that comes to my mind is antialiasing. Theoretically it should be done only by GPU but in my experience it hits performance really bad in coh2 no matter the setup, much more than in other games. They probably just screw it up somewhere there. All we can do is avoid it.
18 Jan 2017, 20:41 PM
#77
avatar of Dangerous-Cloth

Posts: 2066

I noticed that the game looks better with low AA then it does with high AA. I also noticed that on low AA my CPU and GPU both clock higher than when I run the high AA. Sometimes my GPU doesn't even clock above 900MHz, and gives me like 40 fps. This only happens on CoH2..

#Relic
20 Jan 2017, 06:56 AM
#78
avatar of kamk
Donator 11

Posts: 764

-snip-
One thing that comes to my mind is antialiasing. Theoretically it should be done only by GPU but in my experience it hits performance really bad in coh2 no matter the setup, much more than in other games. They probably just screw it up somewhere there. All we can do is avoid it.

Yup, that's why i suggested to turn it off ingame, and force it via driver (if needed / wanted). Physics is another easy FPS boost.

Dangerous-Cloth, similar things happen in other CPU limited games as well. I.e. CSGO player X only reaching 200 instead of 300 FPS - if your GPU is decent enough, it will just idle around. The drops in COH2 just hurt way more.

So, Intels Kabylake i3 7350k does have at least one market. :snfPeter:
20 Jan 2017, 10:50 AM
#79
avatar of ferwiner
Donator 11

Posts: 2885

jump backJump back to quoted post20 Jan 2017, 06:56 AMkamk

Yup, that's why i suggested to turn it off ingame, and force it via driver (if needed / wanted).


That's what I do for quite some time. The only drawback is that if forced in settings, the GPU does the full scale AA always when the coh2 window is present. This basically means that AA is done not only on the 3D generated content, but also on fonts. This of course couses fonts to blur. In game AA simply "knows" which part of window should be modified, while the driver one has no idea.
4 Apr 2017, 11:58 AM
#80
avatar of Huwmungous

Posts: 1

jump backJump back to quoted post13 Jan 2016, 09:39 AMPlaguer
So yesterday I got an idea to check how much CoH2 uses my GPU, CPU etc..
So the thing I noticed is that CoH2 uses around 30% of my GPU and around 40% of my CPU


I have been plagued with slowdowns/lag recently so I have been looking into this.

My problem turned out to be temperature issues; I installed another fan and am now running quite happily on an old AMD FX-8150 Eight-Core system.



The graph shows temperature above the line, GPU in blue, CPU in green and load per-cent below the line - GPU in orange, CPU in brown.

You can see the moment my game ended (seven minutes out), shortly followed by exiting to Windows.

Anyone experiencing issues should definitely consider poor cooling as a cause.



1 user is browsing this thread: 1 guest

Ladders Top 10

  • #
    Steam Alias
    W
    L
    %
    Streak
Data provided by Relic Relic Entertainment

Replay highlight

VS
  • U.S. Forces flag cblanco ★
  • The British Forces flag 보드카 중대
  • Oberkommando West flag VonManteuffel
  • Ostheer flag Heartless Jäger
uploaded by XXxxHeartlessxxXX

Board Info

422 users are online: 1 member and 421 guests
Crecer13
15 posts in the last 24h
41 posts in the last week
95 posts in the last month
Registered members: 44643
Welcome our newest member, Leiliqu96
Most online: 2043 users on 29 Oct 2023, 01:04 AM