[SOLVED] CPU cores are not fully utilized in games ?

Dec 8, 2022
5
1
10
Hello everyone.

My system has i5 12400f and 3070ti with 16gb ram. Based on that system i don't think that i'm getting the fps that i should get. Some games that i tested are;
Spider-man Miles Morales: RTX on ultra settings, 1080p: 30-65fps; RTX off low settings 720p: 50-80fps
Cyberpunk 2077: RTX on ultra settings, DLSS-performance, 1080p: 35-60fps; RTX off low settings, DLSS-off ,720p : 60-90fps
I think my system is experiencing a CPU bottleneck, in games when i speed up with a car or something game starts to stutter to load the npc's and other objects but i5 12400f is enough to feed 3070ti at least for just gaming purpose as far as i know. So i thought there is something wrong with my system either about hardware or software.

All my drivers including CPU chipset are up to date.

Here is an image of performance metrics from Cyberpunk 2077 with ultra settings. Here is shown that only one core of CPU is being used to nearly fullest others are not even used at all. Is this normal?
View: https://imgur.com/a/yUq44Vt
 
Last edited:
Solution
Your single thread cpu-z score is nominal, considering that you are running single channel.
Good dual channel ram might get you to the 702 level.
I don't know how much that would gain you.
If your two sticks of ram came from a single matched kit, they should be able to run in dual channel mode.
Put one stick in each channel as your motherboard manual instructs.

And... FWIW user benchmark is considered as junk science.
Not 100% certain....but this looks like a CPU bottleneck on the Game Main Thread. Looks very much like CPU Main Thread limit in Flight Simulator 2020. GPU % is only 65% and one CPU thread sits at over 85%. That screams CPU bottleneck.

Cyberpunk is very demanding on CPU. Each game has a main thread to run the game. Some tasks can spread out to other threads, but there is still a main thread. Looks like its kinda on the limit. In Flight Sim (which is very main thread limited) I always have one limiting thread sitting around like in the 80s%

Just do a test and lower some of the very CPU demanding settings and see how that changes up things. In particular: lower crowd size
 
  • Like
Reactions: Mecha351
Many games are dependent on the performance of the single master thread.
Not much you can do about it excepting perhaps game settings.
12400 is nicely balanced with a 3070 class graphics card.
If you want to do better, upgrade to a cpu with stronger single thread performance.
Run cpu-z bench and look at the single thread rating.
You should get a number like 702 which is normally quite good:
https://valid.x86.fr/bench/0gt9vb/1
 
  • Like
Reactions: Mecha351
Many games are dependent on the performance of the single master thread.
Not much you can do about it excepting perhaps game settings.
12400 is nicely balanced with a 3070 class graphics card.
If you want to do better, upgrade to a cpu with stronger single thread performance.
Run cpu-z bench and look at the single thread rating.
You should get a number like 702 which is normally quite good:
https://valid.x86.fr/bench/0gt9vb/1
Here is the result; https://valid.x86.fr/bench/qnut7v/1
I get 678 single thread score but i need to add that at the moment my computer is using single channel ram as 2x8gb in 'A' channel. I know this really effects performance but i borrowed my friend's RAMs to enable dual channel since my RAMs are not compatible to work with dual channel and the only difference that i get was like 10fps increase. Maybe this is what you get from 3070ti and 12400f but when look at the videos on youtube people seem to run CP2077 on ultra settings at like 90-100fps. I just want to make sure if i'm getting the most out of my components.
 
Your single thread cpu-z score is nominal, considering that you are running single channel.
Good dual channel ram might get you to the 702 level.
I don't know how much that would gain you.
If your two sticks of ram came from a single matched kit, they should be able to run in dual channel mode.
Put one stick in each channel as your motherboard manual instructs.

And... FWIW user benchmark is considered as junk science.
 
  • Like
Reactions: Mecha351
Solution
Here is the result; https://valid.x86.fr/bench/qnut7v/1
I get 678 single thread score but i need to add that at the moment my computer is using single channel ram as 2x8gb in 'A' channel. I know this really effects performance but i borrowed my friend's RAMs to enable dual channel since my RAMs are not compatible to work with dual channel and the only difference that i get was like 10fps increase. Maybe this is what you get from 3070ti and 12400f but when look at the videos on youtube people seem to run CP2077 on ultra settings at like 90-100fps. I just want to make sure if i'm getting the most out of my components.
Piece at a time.

First piece fix the ram with a 2x8GB kit fitted in the proper slots.
 
  • Like
Reactions: Mecha351
Hello everyone.

My system has i5 12400f and 3070ti with 16gb ram. Based on that system i don't think that i'm getting the fps that i should get. Some games that i tested are;
Spider-man Miles Morales: RTX on ultra settings, 1080p: 30-65fps; RTX off low settings 720p: 50-80fps
Cyberpunk 2077: RTX on ultra settings, DLSS-performance, 1080p: 35-60fps; RTX off low settings, DLSS-off ,720p : 60-90fps
I think my system is experiencing a CPU bottleneck, in games when i speed up with a car or something game starts to stutter to load the npc's and other objects but i5 12400f is enough to feed 3070ti at least for just gaming purpose as far as i know. So i thought there is something wrong with my system either about hardware or software.

All my drivers including CPU chipset are up to date.

Here is an image of performance metrics from Cyberpunk 2077 with ultra settings. Here is shown that only one core of CPU is being used to nearly fullest others are not even used at all. Is this normal?
View: https://imgur.com/a/yUq44Vt
Alright everyone. I bougth a new pair of RAM's and made them run on dual channel and the problem seems to be solved now in cp2077 and spider-man i get %99 gpu utilization most of the time. Thank you all for the answers. I think this thread may be useful for people who are upgrading their system and chose to neglect RAM just as i did.
 
  • Like
Reactions: Karadjgne
RAM was only running @2100MHz, with the old kit 2400 should have been possible
3200MHz is the standard frequency for your CPU

And... FWIW user benchmark is considered as junk science.
If you are looking only at the result numbers, then yes. It´s a really huge database of tested systems, which are very easy to compare. Also you can check compatibility for motherboard, RAM, GPU, CPU, BIOS version.
Because you can find even older systems and can have a look if e.g. a GPU will work with this or this motherboard/CPU/RAM,...

It´s the easiest and fastest tool I know to check for basic errors, like BIOS, firmware versions, exact information about used hardware and drivers. Max. frequencies used in the bench. And in this case showing the RAM speed not properly set to at least 2400MHz (2100MHz shown).
Also the really slow test result indicates a problem with the RAM, which was like said only running at 2133MHz. The next question would have been, check memory for errors with memtest and if the DIMMs are installed correctly as dual channel.