CPU usage difference between systems and optimization

System 1 (desktop):
RX 580 8GB (1380 GHz boost)
i5 4670k 3.6 GHz
8 GB DDR3 1600 MHz
2x SSD + 2 TB HDD (game on 7200rpm HDD)

System 2 (laptop GL702ZC Asus):
RX 580 4GB (1077 boost)
Ryzen 5 1600 3.4 GHz
8 GB DDR4 2400 MHz
SSD + HDD (game on 5600 HDD)

Windows and stuff like that is pretty much the same. Settings are also the same. High Image, High textures, Low snow, off Physics, off Vsync and AA..

On my father's desktop the CPU usage when playing a 3v3 game with/vs Expert AI (don't know if Easy/expert uses more CPU power) is constantly in the range of 60-80, which is great and the FPS starts at 100-110 and steadily decreases to 40-50 during the later stages of the game as more stuff in happening.

On my workstation laptop the CPU usage is constantly in the range of 20-40 and the FPS starts at 70-90 but as the game progresses, it decreases to 20-35 which is much lower than the (CPU/RAM weaker) desktop. Why is the utilization of Ryzen so much lower than the Intel i5? Is it because the game was made before ryzen were out and as such the optimization for that processor is poor? The 20% utilization is really really low. I'm used to playing this game with 25 FPS but those moments when I visit my parents every now and then, this game is one of the few that runs better on a weaker system.

Can anyone explain this phenomenon?

Also does anyone know if the dual channel would help the laptop? I've ordered another (identical) RAM stick of 8GB for the laptop to put it in dual channel mode. Anyone have any experience with dual/single differences?

System 1 (desktop):
RX 580 8GB (1380 GHz boost)
i5 4670k 3.6 GHz
8 GB DDR3 1600 MHz
2x SSD + 2 TB HDD (game on 7200rpm HDD)

System 2 (laptop GL702ZC Asus):
RX 580 4GB (1077 boost)
Ryzen 5 1600 3.4 GHz
8 GB DDR4 2400 MHz
SSD + HDD (game on 5600 HDD)

Windows and stuff like that is pretty much the same. Settings are also the same. High Image, High textures, Low snow, off Physics, off Vsync and AA..

On my father's desktop the CPU usage when playing a 3v3 game with/vs Expert AI (don't know if Easy/expert uses more CPU power) is constantly in the range of 60-80, which is great and the FPS starts at 100-110 and steadily decreases to 40-50 during the later stages of the game as more stuff in happening.

On my workstation laptop the CPU usage is constantly in the range of 20-40 and the FPS starts at 70-90 but as the game progresses, it decreases to 20-35 which is much lower than the (CPU/RAM weaker) desktop. Why is the utilization of Ryzen so much lower than the Intel i5? Is it because the game was made before ryzen were out and as such the optimization for that processor is poor? The 20% utilization is really really low. I'm used to playing this game with 25 FPS but those moments when I visit my parents every now and then, this game is one of the few that runs better on a weaker system.

Can anyone explain this phenomenon?

Also does anyone know if the dual channel would help the laptop? I've ordered another (identical) RAM stick of 8GB for the laptop to put it in dual channel mode. Anyone have any experience with dual/single differences?

Comparing laptop performance to desktop performance?

Comparing laptop performance to desktop performance?

Given the fact that the laptop is stronger, yes.
It's a desktop Ryzen chip and RX 580 desktop chip with only difference being it's downclocked cause of the power/temperature. Still, those 300 MHz is GPU bound and reflects in the difference of FPS in most GPU intensive games. No matter the settings, the CPU on i5 4670k (4 core, 4 thread, much less cache than ryzen) is always used effectively while the ryzen is not.

Also, it's not throttling in any way. Laptop temps are 65C/63C CPU/GPU and desktop 73C/55C CPU/GPU. Even the temps are better in the laptop.

Final argument. Laptop has better benchmarks in physics parts of 3Dmark since the GPU is a bit stronger in the desktop. Still, laptop RX 580 is more than enough to handle COH2.

Given all that. Yes, I am comparing laptop to desktop

Can anyone explain this phenomenon?

Yep. This game doesn't make effective use of hyper-threading (aka SMT on Ryzen), only real CPU cores. This means that Windows gets confused about how report overall CPU usage. Dividing cores into two threads often has performance gains for many applications, but it is definitely not the same as having two actual cores, so overall CPU usage is not accurate in this case.

Your dad's CPU is from when Intel didn't include hyper-threading on their mid and low-range CPUs (since they had no competition from AMD at the time). This means Windows is reporting CPU usage accurately as far as this game is concerned.

You can right-click on the CPU graph in Task Manager to show all of your CPU threads and see how loaded they are individually (Change graph to -> Logical Processors).

If you turn off SMT in your BIOS, you should see similar (and more accurate) CPU usage to the Core i5. However, even though Windows will report your CPU as being used much more, you shouldn't expect much, if anything, in the way of increased performance. Can't hurt to try, but you will likely lose performance in almost every other game if you don't re-enable it.

Ultimately, the COH2 engine is just terribly made. Troubled and rushed development aside, I've heard it was apparently made on an Intel compiler and really favours their CPUs. But that said, AMD literally didn't have a good gaming CPU until they very recently launched the Ryzen 3000 series nearly a decade later. Also, as you've noticed, even with an Intel CPU it can't use it all and only gets worse as each game goes on, regardless of GPU. I did some tests on a 9900K and performance was really not much better.

I run a Ryzen 1600 on desktop (with a GTX 1080) but I probably only get as low as 30-40fps towards the end of an artillery intensive 4v4 match. Your performance does seem unusually poor. I would have suggested thermal throttling, but I guess not based on your data.

I've never really played against a bunch of AI players though. I have to assume that would use more CPU than human players. Maybe try a couple of 3v3 automatch games and compare the end game frame rates?

Given the fact that the laptop is stronger, yes.

No, your laptop is not stronger. It has the same GPU but 300 Mhz less GPU clock and that is a huge difference.

Also the i5 4670k has a slightly better single and dual core performance than your Ryzen 5 1600 and because the game can't utilize more than 1-2 cores and doesn't make use of hyperthreading at all it will likely perform better. The reason why the R5 1600 runs the game on a lower CPU usage % is because most of its 6 cores and 12 threads are idle while the i5 4670k only has 4 cores.

This is the reason why you have less FPS on the laptop compared to your desktop.

COH2 is optimised so poorly that you could plug in the latest i9 or Ryzen 9 and it wouldn't make a significant difference due to how badly the engine is designed for modern hardware.

No, your laptop is not stronger. It has the same GPU but 300 Mhz less GPU clock and that is a huge difference.

Also the i5 4670k has a slightly better single and dual core performance than your Ryzen 5 1600 and because the game can't utilize more than 1-2 cores and doesn't make use of hyperthreading at all it will likely perform better. The reason why the R5 1600 runs the game on a lower CPU usage % is because most of its 6 cores and 12 threads are idle while the i5 4670k only has 4 cores.

This is the reason why you have less FPS on the laptop compared to your desktop.

COH2 is optimised so poorly that you could plug in the latest i9 or Ryzen 9 and it wouldn't make a significant difference due to how badly the engine is designed for modern hardware.

GPU is weaker and it translates into 10 FPS less in Graphics test in Firestrike (which is expected for 300 MHz less core clock), but 8k more in phyiscs (7096 vs 15132) and combined test is only 4 FPS less on laptop. Overall scores are the same (10650 for laptop, 10800 for desktop).
Overall, yes, the GPU is weaker but the laptop by itself is stronger. A bit weaker gaming but much stronger workstation.

Anyway, thanks for clarifiying that the engine is s*** and ineffective.

snip

Thanks for the great clarification. So in other words, nothing I can do about it. Do you think that going dual channel would help with the FPS?
2v2 on most maps is steady 50+ FPS
3v3 dips from 90 FPS to average 25-30
4v4 dips a bit more with average 20 FPS in long long arty spamfests.
1v1 runs fluently for the whole game.
I tested those only with AI since I can't really multitask and steadily check the FPS while playing vs people and MSI afterburner for some reason won't log AVG FPS, nor minimums.

It's a shame that it was rushed like that. Being able to play Witcher 3, high, fluently at 60 and not COH2 at custom low/high setting is a bummer.

Thanks for the extensive reply. Much appriciated.

Do you think that going dual channel would help with the FPS?

Oh, I didn't pay attention to your RAM.

RAM clock speed and latency makes a substantial difference to performance for all Ryzen CPUs due to it controlling the Infinity Fabric speed that links the CCXs. Your CPU would have 2 CCXs containing 4 cores each, and those CCXs trying to communicate to each other would be the greatest source of latency (and therefore bottlenecking) in your CPU.

The optimal RAM speed for Ryzen CPUs before you hit diminishing returns is "3200Mhz" and higher. At "2400Mhz" for your current RAM there is definitely room for improvement in your CPU's performance.

Show Spoiler

Latency also matters with RAM. The 'CL' number is the most important, but it is completely relative to the clock speed. So you should try and buy RAM with the lowest CL number you can reasonably afford at a given clock speed. CL16 is a good one to aim for with 3200Mhz DDR4.

Show Spoiler

As for dual channel, I don't have data for COH2, but this JayzTwoCents video makes for a pretty compelling argument to always use it: https://youtu.be/bm4U_S_5dSo?t=643

I would also just generally recommend having 16GB (2x8GB) of RAM for gaming in general these days. If you have your settings maxed out, then long games on large maps could quite possibly be causing your 8GB of RAM to fill up and page to your hard drive, which could definitely tank performance. I know your dad's also has 8GB, but there may be other stuff on your laptop also using RAM. Close any unnecessary background programs, and then try turning textures right down and seeing if it makes a difference. If it helps, then bump up textures by one level at a time.

So all in all, I would first try lowering textures and closing background programs (especially browsers) to see if that helps. But regardless, if you find you have the money for something similar to 2x8GB of 3200Mhz CL16 RAM, you are likely to see a noticeable improvement in performance for all your games. There are an abundance of tech videos on Youtube showing real-world benefits in the double digit range. The biggest problem I'm seeing is that most laptop RAM seems to be 2666Mhz or lower, so if you look at upgrading you may have to shop around to find something better. Good luck.

Show Spoiler

Oh, I didn't pay attention to your RAM.

RAM clock speed and latency makes a substantial difference to performance for all Ryzen CPUs due to it controlling the Infinity Fabric speed that links the CCXs. Your CPU would have 2 CCXs containing 4 cores each, and those CCXs trying to communicate to each other would be the greatest source of latency (and therefore bottlenecking) in your CPU.

The optimal RAM speed for Ryzen CPUs before you hit diminishing returns is "3200Mhz" and higher. At "2400Mhz" for your current RAM there is definitely room for improvement in your CPU's performance.

Show Spoiler

Latency also matters with RAM. The 'CL' number is the most important, but it is completely relative to the clock speed. So you should try and buy RAM with the lowest CL number you can reasonably afford at a given clock speed. CL16 is a good one to aim for with 3200Mhz DDR4.

Show Spoiler

As for dual channel, I don't have data for COH2, but this JayzTwoCents video makes for a pretty compelling argument to always use it: https://youtu.be/bm4U_S_5dSo?t=643

I would also just generally recommend having 16GB (2x8GB) of RAM for gaming in general these days. If you have your settings maxed out, then long games on large maps could quite possibly be causing your 8GB of RAM to fill up and page to your hard drive, which could definitely tank performance. I know your dad's also has 8GB, but there may be other stuff on your laptop also using RAM. Close any unnecessary background programs, and then try turning textures right down and seeing if it makes a difference. If it helps, then bump up textures by one level at a time.

So all in all, I would first try lowering textures and closing background programs (especially browsers) to see if that helps. But regardless, if you find you have the money for something similar to 2x8GB of 3200Mhz CL16 RAM, you are likely to see a noticeable improvement in performance for all your games. There are an abundance of tech videos on Youtube showing real-world benefits in the double digit range. The biggest problem I'm seeing is that most laptop RAM seems to be 2666Mhz or lower, so if you look at upgrading you may have to shop around to find something better. Good luck.

Show Spoiler

Tyvm. Much appreicated. The memory stick I ordered for dual channel is this one Samsung M471A1J43CB1. It's identical to the one inside and this Asus laptop (3 years old), won't support speeds above 2400 MHz without some serious custom BIOS tweaking which I won't do for one game since all other games run perfectly fine 60 FPS.

Thanks again for an extensive research.

Thanks again

No worries. Best of luck, and I hope it helps.

Ultimately though, we're all just hoping for COH3 to be released on a much better engine

Thanks for the great clarification. So in other words, nothing I can do about it. Do you think that going dual channel would help with the FPS?
2v2 on most maps is steady 50+ FPS
3v3 dips from 90 FPS to average 25-30
4v4 dips a bit more with average 20 FPS in long long arty spamfests.
1v1 runs fluently for the whole game.
I tested those only with AI since I can't really multitask and steadily check the FPS while playing vs people and MSI afterburner for some reason won't log AVG FPS, nor minimums.

It's a shame that it was rushed like that. Being able to play Witcher 3, high, fluently at 60 and not COH2 at custom low/high setting is a bummer.

Thanks for the extensive reply. Much appriciated.

I used to run an i5 6600k and a RX 480 and have way more FPS. When I start a 2v2 I have 120+ FPS and it never goes below 50-60 FPS in late game arty fests either. My settings are mostly ultra with AA on low because the higher AA levels don't make the game look better at all IMO and only cost FPS.

I don't play 4v4 but in 3v3 I never have dips to 25-30 either.

A few weeks ago I upgraded to Ryzen 7 2700X because MSFS 2020 runs like shit on quadcore CPUs with HT and the performance is basically still the same for COH2 as I expected.

There are random games every now and then that are super laggy for no apparent reason but normally your set-up should be more than enough for COH2.

I used to run an i5 6600k and a RX 480 and have way more FPS. When I start a 2v2 I have 120+ FPS and it never goes below 50-60 FPS in late game arty fests either. My settings are mostly ultra with AA on low because the higher AA levels don't make the game look better at all IMO and only cost FPS.

I don't play 4v4 but in 3v3 I never have dips to 25-30 either.

A few weeks ago I upgraded to Ryzen 7 2700X because MSFS 2020 runs like shit on quadcore CPUs with HT and the performance is basically still the same for COH2 as I expected.

There are random games every now and then that are super laggy for no apparent reason but normally your set-up should be more than enough for COH2.

I also think that the setup is more than enough. I'm getting another RAM stick to see if the increased bandwidth will help in this CPU heavy game.

I'm positive that there is no throttling as the temps are good and Firestrike gives consistent results. + the core/memory/cpu clock line is as flat as it can get in MSI during the whole game.
1077 for GPU core
2000 for GPU mem
3400 for CPU

and I highly doubt that on lower settings than blvckdream's RX 480 is fighting, that 200 MHz higher core clock on boost delivers such massive difference. Especially since COH2 is not GPU heavy and in GPU heavy games the growth is linear when it comes to core clock on GPU.

EDIT: Btw. thanks for sharing your own experience.

I'm getting another RAM stick to see if the increased bandwidth will help in this CPU heavy game.

I don't know if it is a "CPU heavy game". Would be interesting if someone wanted to spend the time to monitor CPU load during a 4v4 and look if a core gets maxed out at the end of the game.

No worries. Best of luck, and I hope it helps.

Ultimately though, we're all just hoping for COH3 to be released on a much better engine

I just tested Fields of Winnekendonk, 3v3 (again all expert AI):

Lowest possible settings: Starts at 100 FPS, 10 minutes after: Average FPS goes to 24
Highest possible settings: Starts at 60, 10 minutes after: Average FPS goes to 23

So no matter the settings, the FPS will drop to that average. My God the optimization is horrible