What's new
Embers Adrift

Register a free account today to Ignite your Adventure! Once signed in, you'll be able to participate with the Embers Adrift community. Your active account will also be the same account used to purchase, download, and login to the game.

Benchmarking the Resolution Scaling feature

Aeldar

Active Member
Below is a summary of benchmark results comparing three systems with the resolution scaling feature. This was testing using the code on the weekend 19-21st, will test again with the new performance setting in the recent patch (yesterday).

- MacBook Air M1 (8 GB unified memory, 8 GPU version)
- Desktop PC with Nvidia 3060 Ti (16+8 GB)
- Laptop PC with Nvidia 3050 Ti (16+4GB)

It seems that the lower amount of memory in the system the greater the benefit (makes sense). Particularly the 8 GB Macbook Air is getting a huge benefit.

Note regarding charts 2a/2b and the 3060 Ti result: In that test, the frame rate was 75 regardless of the changes I made. Even with no frame rate limiter and the monitor refresh rate is 100Hz.

1661510150907.png

1661510192286.png

Tests were done having the Options window opened (i.e. didn't close it after each setting).
In-game location: NE Newhaven Valley, out in the woods with no players. STANDING STILL! Didn't record moving FPS. Horizon at ca 50% of the screen

Benchmark settings version 1:
All settings on medium
Resolution 1440x900
SMMA anti aliasing
15 View distance
0% shadow distance
100% grass distance
All checkboxes checked in the lower section


Settings "real life 3060 Ti"
All settings on medium
Resolution 3440x1440
No anti aliasing
77% View distance
80% Shadow distance
100% Grass distance
All checkboxes except "Grass shadows" checked in the lower section


Settings "real world M1" (8GB, 8 GPU version)
All settings on low
Resolution 1440x900
No anti-aliasing
15 View distance
0% shadow distance
100% grass distance
No checkboxes checked in lower section
 

Aeldar

Active Member
Updates after the weekly patch, tests done on PC:

1) The setting "Balanced" gives slightly higher performance compared to "Performant".
i.e. Potato > Balanced > Performant > High Fidelity
GPU is at 100%, CPU is ca 40-50%

2) I've seen no screen tearing at all (Vsync is off, no FPS limit). I read somewhere that you noticed the overall performance was a bit lower this patch. Could it be that Vsync is "always on" regardless of the setting? I see no change in FPS when enabling/disabling Vsync
 

Undone

Codemaster
Stormhaven Studios
Updates after the weekly patch, tests done on PC:

1) The setting "Balanced" gives slightly higher performance compared to "Performant".
i.e. Potato > Balanced > Performant > High Fidelity
GPU is at 100%, CPU is ca 40-50%
Hrm, I don't see any reason why balanced would be giving better performance than performant - possible that your measurements are within the margin or error with one another?

2) I've seen no screen tearing at all (Vsync is off, no FPS limit). I read somewhere that you noticed the overall performance was a bit lower this patch. Could it be that Vsync is "always on" regardless of the setting? I see no change in FPS when enabling/disabling Vsync
What is the refresh rate of your monitor? If it is a higher refresh rate then you are likely hitting the framerate cap of your rig before vsync.
 

Aeldar

Active Member
Hrm, I don't see any reason why balanced would be giving better performance than performant - possible that your measurements are within the margin or error with one another?


What is the refresh rate of your monitor? If it is a higher refresh rate then you are likely hitting the framerate cap of your rig before vsync.
I was also surprised. I did several measurements in Newhaven Valley and the Balanced FPS was consistently higher (53,9 FPS vs 49,3 FPS on average with all tests summarized):
- Stand-still test using different resolution scaling
- Average FPS when running a certain track (did the same track twice on each setting)

However when I did a stand-still test now in Northreach (NE area) I got 37 FPS for Balanced and 38 FPS for Performant (average FPS during ca 10 seconds).

But there are so many things that affect the measurement, could be a coincidence...

About Vsync: I get same FPS using "No limit", "Max 60" and "Vsync" and get no screen tearing in either setting. Usually the FPS does down a bit when enabling Vsync. It appears to me that Vsync is always enabled regardless of the "Frame rate limiter" setting.
Screen refresh rate is 100Hz. The recorded FPS was in the range of 22-66.
 

Undone

Codemaster
Stormhaven Studios
About Vsync: I get same FPS using "No limit", "Max 60" and "Vsync" and get no screen tearing in either setting. Usually the FPS does down a bit when enabling Vsync. It appears to me that Vsync is always enabled regardless of the "Frame rate limiter" setting.
Screen refresh rate is 100Hz. The recorded FPS was in the range of 22-66.
If your refresh rate is 100hz when V-Sync is enabled the game will try to render at 100fps - which it will unlikely ever reach. In other words, V-Sync in your case is the same as a framerate limiter of 100fps - if you never reach that target then it's no surprise you are not noticing a difference.
 

Aeldar

Active Member
I thought Vsync would prevent partial frames at the cost of FPS?

Atleast in previous tests I got screen tearing when not using Vsync. When I enabled Vsync the screen tearing disappeared but resulting in slightly lower FPS.
 

Undone

Codemaster
Stormhaven Studios
I thought Vsync would prevent partial frames at the cost of FPS?
But what does that mean? It means it will limit your application's frame rate to that of your monitor's refresh rate. It doesn't have any impact if the application can't reach that frame rate in the first place.
 

Aeldar

Active Member
You are right, that's what it does. Hmmm I must have mixed it up with something else. I thought it was a generic fix for anything that displays a partial frame (even if below max frame rate).

I recall now that some time back I found that my Windows refresh rate was set to 60Hz and not 100Hz on my new monitor. Maybe that was when I saw screen tearing.