Question PC not fulfilling my expectation: How to find the source of the problem?

[theAExample]

Hey! So I bought and built this PC in July 2022 and ever since then, Ive not had one moment where I thought "Wow! This PC is soooo good". It runs games like CS2, Valorant (so games that do not need much to run) perfectly but as soon as it comes to heavier games like COD, Elden Ring (I know optimization is not the best), and other games, it doesnt perform all too well, even if I change my settings.
Now Ive tried fixing this issue multiple times over those 2 years but no success. Ofc XMP is enabled, tried different graphic settings, deleted and installed newest drivers for GPU. There are so many things you can try and do but I get overwhelmed with the options.
And Im not the biggest fan of having to do a clean install atm.

As in specs:
Using Win11 and playing on 1440p
CPU: i5 12600K
CPU cooler: Corsair H150i Elite Capellix
GPU: Msi 3080 Suprim X
Mainboard: ROG STRIX B660 A Gaming
RAM: Vengeance RGB Pro SL (2x16)
Storage: 1TB M.2, 2TB M.2
PSU: Corsair RM750x

In terms of temps it never really gets high and at what percentage the CPU/GPU work. Its like almost always below 50%, 40% and around 30%, even while playing, its almost concerningly low sometimes.
The following 2 pictures are Time Spy benchmarks, the higher one was with energy plan "Ultimate" and lower was "High Performance". Apparently I used Ultimate for 2 years, but thats beside the point. And 2 days ago I also made a Time Spy benchmark and that was at like 16.7k points?

Id really appreciate your advice, if you need me to do more benchmarks please let me know. Sadly Im completely lost when it comes to reading benchmarks and knowing which ones to do.

https://ibb.co/rpk0L0D

https://ibb.co/ZfzwH7B

 

[geofelt]

What are your expectations?
Your cpu/gpu are quite good.

Run the cpu-z bench on your i5-12600K
You should get a single thread score about 773:
https://valid.x86.fr/bench/bhgfiz/1

 

[LinuxDevice]

Just a thought on something you said, that temps never go up too high. This is also an indication that the software itself is not driving the hardware to actually use the hardware limits. I can't say for sure about anything in particular, but it is possible that the games themselves are contributing to this issue on the ones you say are "heavier games".

For a GPU to go faster one must feed it more data, which in turn is CPU. The CPU of course might need to be fed from RAM, or the m.2, or from a network. This brings to mind two things to check.

First are GPU and CPU both running at around the same percentage of their full capability? If the CPU is running at 50% when the GPU is at 100%, then the GPU is bottlenecking. If both are running at something like 30% or 50%, then you can consider them "balanced" (which is good). If consumption is not bottlenecked by CPU or GPU, then often it is a case of feeding more data.

Along the second possibility, feeding it data, sometimes RAM is not using XMP (or I think something like DOCP on some RAM). I would check in your BIOS settings if any high speed profiles are missing. Do note that I am not speaking of overclocking, I'm simply talking about ordinary intended higher clocks which may not be set by default for some reason.

If for some reason the RAM is acting as single channel (you said you have two sticks), then it might be either (A) not using the correct MOBO slots, or (B) the RAM sticks are not matched timings (in which case you would have likely installed from different packages rather than purchasing them as a matched set in a single package).

I'll also suggest that 32 GB of RAM is likely enough to run without swap. Even if it is only as a test, I will suggest you disable or remove any swap from your system. Force it to not swap out, and see if anything changes. If the help from this is significant, then it might be the o/s swapping out is the problem (which could also wear out your m.2s a bit faster).

 

[theAExample]

Just a thought on something you said, that temps never go up too high. This is also an indication that the software itself is not driving the hardware to actually use the hardware limits. I can't say for sure about anything in particular, but it is possible that the games themselves are contributing to this issue on the ones you say are "heavier games".

For a GPU to go faster one must feed it more data, which in turn is CPU. The CPU of course might need to be fed from RAM, or the m.2, or from a network. This brings to mind two things to check.

First are GPU and CPU both running at around the same percentage of their full capability? If the CPU is running at 50% when the GPU is at 100%, then the GPU is bottlenecking. If both are running at something like 30% or 50%, then you can consider them "balanced" (which is good). If consumption is not bottlenecked by CPU or GPU, then often it is a case of feeding more data.

Along the second possibility, feeding it data, sometimes RAM is not using XMP (or I think something like DOCP on some RAM). I would check in your BIOS settings if any high speed profiles are missing. Do note that I am not speaking of overclocking, I'm simply talking about ordinary intended higher clocks which may not be set by default for some reason.

If for some reason the RAM is acting as single channel (you said you have two sticks), then it might be either (A) not using the correct MOBO slots, or (B) the RAM sticks are not matched timings (in which case you would have likely installed from different packages rather than purchasing them as a matched set in a single package).

I'll also suggest that 32 GB of RAM is likely enough to run without swap. Even if it is only as a test, I will suggest you disable or remove any swap from your system. Force it to not swap out, and see if anything changes. If the help from this is significant, then it might be the o/s swapping out is the problem (which could also wear out your m.2s a bit faster).

Hi, thanks for you answer!

Id have to do more tests to know 100% if one of those 2 are bottlenecking but Im leaning towards it being balanced. Would you happen to know how I check if one is bottlenecking the other? Or how to check if they are actually working with their full capability.

I will check the high speed profiles in BIOS soon like u advised me to! Im pretty sure the MOBO slots are correct, like not pretty but 100% and idk if I understood u correctly but both sticks came from the same package.

Id also have to read something about "swap", I have never heard about swapping
Thanks!

 

[theAExample]

What are your expectations?
Your cpu/gpu are quite good.

Run the cpu-z bench on your i5-12600K
You should get a single thread score about 773:
https://valid.x86.fr/bench/bhgfiz/1

My expectations were/are that the PC is able to run easily above 100FPS no matter what graphic settings. Ofc some very demanding games might have problems with above 100 FPS on max settings but at least run it smoothly with no lags etc. An afterthough: I just want my PC to be able to run games easily with no problems. I dont really remember why my expectations were. And its disappointing for me if it couldnt run games easily with high-er settings really good yet

I just ran the cpu-z bench, was a bit confused on where to find it but here you go:
https://valid.x86.fr/e0qhhv
It doesnt look too good to me. Am I supposed to close all applications except CPU Z or does it not matter?

 

[geofelt]

The all thread number, if I read it right is 6324 while the population of others score higher, about 7220.
I think many submitting results have overclocked their systems which is why yours look lower in comparison.
I was actually more interested in the single thread rating which is what gamers need most.

The impact of settings, and resolution fall mostly on the graphics card.
Try this experiment:
Run YOUR games, but lower your resolution and eye candy.
This makes the graphics card loaf a bit.
If your FPS increases, it indicates that your cpu is strong enough to drive a better graphics configuration.
If your FPS stays the same, you are likely more cpu limited.

 

[theAExample]

The all thread number, if I read it right is 6324 while the population of others score higher, about 7220.
I think many submitting results have overclocked their systems which is why yours look lower in comparison.
I was actually more interested in the single thread rating which is what gamers need most.

The impact of settings, and resolution fall mostly on the graphics card.
Try this experiment:
Run YOUR games, but lower your resolution and eye candy.
This makes the graphics card loaf a bit.
If your FPS increases, it indicates that your cpu is strong enough to drive a better graphics configuration.
If your FPS stays the same, you are likely more cpu limited.

I see, Ill try that as soon as I can!

But if its the case that the FPS increases, that could theoretically mean that Id have to play on a lowered resolution which would be all but convenient and nice to look at.
Also, when I built my PC I made sure not to bottleneck my GPU with my CPU. And the 12600K was supposed to have little to no, so only a bit of bottleneck. Im just confused how such a relatively new PC with good parts is not running extremely well by itself.

A thing I also look at is my friends new PC that I helped him build. Roughly the same price as mine back then, and yes, of course new PC parts will perform better than the ones 2 years ago but I never wouldve thought that his RX 7900 XTX + 7800X3D will perform about 3 times better than my PC

 

[Zerk2012]

My expectations were/are that the PC is able to run easily above 100FPS no matter what graphic settings. Ofc some very demanding games might have problems with above 100 FPS on max settings but at least run it smoothly with no lags etc. An afterthough: I just want my PC to be able to run games easily with no problems. I dont really remember why my expectations were. And its disappointing for me if it couldnt run games easily with high-er settings really good yet

I just ran the cpu-z bench, was a bit confused on where to find it but here you go:
https://valid.x86.fr/e0qhhv
It doesnt look too good to me. Am I supposed to close all applications except CPU Z or does it not matter?

What monitor resolution are you using.

 

[thestryker]

Based on just the Time Spy results it seems like your CPU might be underperforming some. I'm guessing this is a DDR4 system which can certainly have an impact on certain games (latencies also can come into play). While the B660 platform prevents CPU overclocking it should be running Intel's standard turbo which it doesn't look like it is. The Time Spy results are showing what looks to be ~4.5Ghz which is definitely slower than it should be.

 

[Bob.B]

Hey! So I bought and built this PC in July 2022 and ever since then, Ive not had one moment where I thought "Wow! This PC is soooo good". It runs games like CS2, Valorant (so games that do not need much to run) perfectly but as soon as it comes to heavier games like COD, Elden Ring (I know optimization is not the best), and other games, it doesnt perform all too well, even if I change my settings.
Now Ive tried fixing this issue multiple times over those 2 years but no success. Ofc XMP is enabled, tried different graphic settings, deleted and installed newest drivers for GPU. There are so many things you can try and do but I get overwhelmed with the options.
And Im not the biggest fan of having to do a clean install atm.

As in specs:
Using Win11 and playing on 1440p
CPU: i5 12600K
CPU cooler: Corsair H150i Elite Capellix
GPU: Msi 3080 Suprim X
Mainboard: ROG STRIX B660 A Gaming
RAM: Vengeance RGB Pro SL (2x16)
Storage: 1TB M.2, 2TB M.2
PSU: Corsair RM750x

In terms of temps it never really gets high and at what percentage the CPU/GPU work. Its like almost always below 50%, 40% and around 30%, even while playing, its almost concerningly low sometimes.
The following 2 pictures are Time Spy benchmarks, the higher one was with energy plan "Ultimate" and lower was "High Performance". Apparently I used Ultimate for 2 years, but thats beside the point. And 2 days ago I also made a Time Spy benchmark and that was at like 16.7k points?

Id really appreciate your advice, if you need me to do more benchmarks please let me know. Sadly Im completely lost when it comes to reading benchmarks and knowing which ones to do.

https://ibb.co/rpk0L0D

https://ibb.co/ZfzwH7B

See if you can run this.

Please click the green human to continue - UserBenchmark

If you can... run it with the browser closed and when it finishes post a link to the results page.

 

[theAExample]

What monitor resolution are you using.

I mean I already said 1440p, isnt that what are you asking for?
So 2560x1440 written out. Correct me if Im wrong!

 

[theAExample]

See if you can run this.

Please click the green human to continue - UserBenchmark

If you can... run it with the browser closed and when it finishes post a link to the results page.

Sadly userbenchmark just keeps on not letting me use the benchmark

 

[Bob.B]

Yup you have to get lucky perhaps try again later.

 

[theAExample]

Yup you have to get lucky perhaps try again later.

Once I try, should I click on all things to get benchmarked or only one?

 

[theAExample]

Based on just the Time Spy results it seems like your CPU might be underperforming some. I'm guessing this is a DDR4 system which can certainly have an impact on certain games (latencies also can come into play). While the B660 platform prevents CPU overclocking it should be running Intel's standard turbo which it doesn't look like it is. The Time Spy results are showing what looks to be ~4.5Ghz which is definitely slower than it should be.

Hmm.. I see. At the moment Im thinking maybe BIOS needs some tweaking but Ive never done it before since I only changed XMP and thats it. Someone advised me to check if XMP is actually enabled and it is. But other than that Im clueless. Trying to watch videos to know what I could change but no luck so far.

Also I read somewhere that using Armoury Crate to update drivers is not good since ASUS uses old drivers. I only used Armoury Crate for the first time a few days ago and updated some drivers since it gave me the option but now Im considering to reinstall all (CPU) drivers, but dont know what the smartest way is to go about it yet.

 

[Bob.B]

Once I try, should I click on all things to get benchmarked or only one?

Select all.

 

[theAExample]

Select all.

Interesting, I actually managed to get in after continously installing and starting the program again and again lmao!
But sadly it doesnt recognize my GPU(?) on both runs..

This benchmark is where I only selected CPU because I didnt expect it to actually benchmark something:
https://www.userbenchmark.com/UserRun/68933951

And this one is where I selected all:
https://www.userbenchmark.com/UserRun/68933974

 

[Bob.B]

Interesting, I actually managed to get in after continously installing and starting the program again and again lmao!
But sadly it doesnt recognize my GPU(?) on both runs..

This benchmark is where I only selected CPU because I didnt expect it to actually benchmark something:
https://www.userbenchmark.com/UserRun/68933951

And this one is where I selected all:
https://www.userbenchmark.com/UserRun/68933974

You might want to update the bios your way behind and also the chipset driver if it's old.

 

[LinuxDevice]

Hi, thanks for you answer!

Id have to do more tests to know 100% if one of those 2 are bottlenecking but Im leaning towards it being balanced. Would you happen to know how I check if one is bottlenecking the other? Or how to check if they are actually working with their full capability.

I will check the high speed profiles in BIOS soon like u advised me to! Im pretty sure the MOBO slots are correct, like not pretty but 100% and idk if I understood u correctly but both sticks came from the same package.

Id also have to read something about "swap", I have never heard about swapping
Thanks!

Just replying in order, I did not yet read the next post.

I think you've found benchmarking apps already, but one of the most basic tools you can use is the task manager (assuming Windows). You can switch to viewing CPU activity or GPU activity. If you just get an idea of percentage use while in one of those higher load scenarios, see if it is roughly at some non-100% level; then check the other. I've not tried to run two task managers at once, maybe that is possible. Having a second monitor to see this while in your app really helps.

Someone else might have suggestions for viewing loads in the middle of a game if you have only one monitor. The trick is to see CPU and GPU activity at the same time.

I don't see if you are using Windows 10 or 11. But here is a Windows 10 example Google search:
"windows 10" disable swap

The point is that you are disabling any swap file. Swap has existed since I think the 80286 chip. The old 8080 and 8088 probably did not have swap. Swap came about because of how expensive RAM was in the early days; most computers choke in some way if they run out of RAM, but there are "Jedi mind tricks" to use disk storage to store RAM content temporarily, use the RAM for some other purpose, and then when you need that in RAM again, restore it from disk. This is on by default, and the scheduler in the system kernel decides when something is not of high enough priority, and puts it in swap memory instead so actual RAM is freed up. This is a lot slower than RAM, and if you have enough RAM, only serves as a "safety" in case something momentarily uses more RAM than available. 32 GB of RAM is easily enough for many uses, including most games, but if you were to have other content running at the same time, then it might be that you need swap (e.g., if you have 25 browser pages open while gaming). Just look at the Google search and test, then enable swap again when done if it doesn't help. This skips an enormous contributor to slowing if swap is the reason.

 

[LinuxDevice]

Interesting, I actually managed to get in after continously installing and starting the program again and again lmao!
But sadly it doesnt recognize my GPU(?) on both runs..

This benchmark is where I only selected CPU because I didnt expect it to actually benchmark something:
https://www.userbenchmark.com/UserRun/68933951

And this one is where I selected all:
https://www.userbenchmark.com/UserRun/68933974

I don't know anything about those benchmarks. I would be careful though about ones you download and install if they are not from reputable servers (there is some security risk). Someone else might know of a good benchmark for you, but you likely want to mention if it is Windows 10 or 11 (or something older).