CoH2 on 2020 Hardware

Finally got my parts in and it's time to benchmark everyone's favorite game from 2013. I was curious to how much more performance could be squeezed from coh2, it's always chugged when you pan the camera down. Normally in gameplay this is never an issue but if you ever want to sit back and look parallel to the battlefield.... Not great. I will be testing with my new 5800x and 3080 today.


Game is tested with all settings on max at 1440p using the in-game benchmark, with an FPS limit set to 142hz.

Original hardware specs:
Ryzen R7 1700 (overclock disabled)
1070
3000Mhz ddr4



New cpu, motherboard, and ram
Ryzen R7 5800x
1070
3600Mhz ddr4


Slightly faster average frame time, about what I expected. 8% more performance from the cpu and MOBO upgrade.

Final build:
Ryzen R7 5800x
3080 FE
3600Mhz ddr4


Show Spoiler
Double the average FPS and 130% higher minimum FPS. I have a framerate limiter set at 142 FPS so presumably the max and average would be slightly higher. It turns out you can just brute force the Essence Engine to spit out as many FPS as possible. Crazy to think I used to play this game on a 650ti Boost.

Nice specs and it will get even better after the patch.

Nice. How did you manage to set the fps cap to 142? I thought you could only set it to 120 max.

This shows though that the game is absolutely bottlenecked by the graphics card, CPU changes don't seem to do anything.
But the game already looks very pretty on second highest settings. Even my 4 year old medium rig can handle it decently,

Wasn't it kinda expected? The fact that the game has poor CPU optimization equals no great gains when upgrading. Some people have said that the game greatly favors Intel architecture, so there's that. And last but not least, GPU is the main driving force so you can pretty much brute force every good looking game by getting a more powerful card.

The important question is: How to reduce FPS drops in teamgames.
1v1 has no fps drops throughout the 1h game. 2v2 has some, once both sides are at 100 cap. 3v3 starts getting stuttery at 15-20 minute mark. 4v4 is a general shit***w unless you have a brutal system.

I'm hoping the 64bit people clean up the code and that it actually runs smoother. 30 FPS is not that bad and it's more than playable, but constant drops to 10 when scrolling the map is eye wrecking.

Nice. How did you manage to set the fps cap to 142? I thought you could only set it to 120 max.

+1

The important question is: How to reduce FPS drops in teamgames.
1v1 has no fps drops throughout the 1h game. 2v2 has some, once both sides are at 100 cap. 3v3 starts getting stuttery at 15-20 minute mark. 4v4 is a general shit***w unless you have a brutal system.

My guess is that larger game modes increase the two largest time killers: FOW and Pathing.

vCOH calc'ed FOW like once every second. COH2 seems to do it almost every frame. And it takes a bit of time to do if done on the CPU as opposed to the GPU. Dont know where Relic does it. Maybe they moved to GPU for Coh2.

Pathing just takes time in general. You have to guess all the possible pathways and find the shortest difference. So bigger modes = more units pathing, more units blocking pathing, larger maps so you need to do this calc a lot more, etc. This is probably all CPU based for 2013 code? Units also seem to favor red cover so vehicles get a speed boost on roads. But vCoh fans remember units that had to path long distances would just stop halfway there because they could not figure out a path anymore and their stored path got blocked.

Larger game modes also have more artillery/tanks/mortars/AT going which means adjusting terrain height, terrain cover, terrain impass(pathing), terrain precise map(pathing unit width), terrain texturing, etc all on the fly while things are blowing up.

I am always amazed the game runs at all. So kudos to Relic for the performance they got out of it. My point is dont get your hopes up too high.

+1

Oh, I actually found out how to set the fps limit on the game. On steam, you go to game properties>set launch options> and write "-refresh 144" without commas.

My guess is that larger game modes increase the two largest time killers: FOW and Pathing.

I do want to point out in 4v4s there are two factors for lagging:

* Increased server side lag - due to Relic servers
* Increased client side lag - due to GPU

I would be surprised if pathing is calculated client-side, so more pathing should make the server itself lag not the player.

Fog of War on the other hand is calculated client-side, hence it is so susceptible to being hacked w/ map hacks. This is why Valve made fog of war server-side with DotA.

Larger game modes simply have more pathing and AI behavior calculations which can create server side lag.

Larger game modes also have more terrain-modifying explosion holes (which are known to be unoptimized and is an applied texture on the terrain). So as the game advances there are more and more shaders the GPU needs to process. In addition to just more units and thus more textures.

Nice. How did you manage to set the fps cap to 142? I thought you could only set it to 120 max.

I set the fps cap using rivatuner, I forgot about the cap in game tbh. Maybe the in game cap is disabled during the benchmark?

The important question is: How to reduce FPS drops in teamgames.
1v1 has no fps drops throughout the 1h game. 2v2 has some, once both sides are at 100 cap. 3v3 starts getting stuttery at 15-20 minute mark. 4v4 is a general shit***w unless you have a brutal system.

I have absolutely no frame drops in 3v3.