Slow performance on Steam Deck

[rhossack]

Any tips on settings that could improve performance? It runs fine on my desktop, but at about 30 fps on a Steam Deck. I emulate Mushihimesama Futari via Xenia on the same system and it runs fine. Looking at the F4 menu settings, nothing jumps out at me as an option I could toggle to get it to run better.

[eandis]

are you running the windows build? potentially could be because of that. the next releases will also have a linux build bundled

[rhossack]

Ah, that's probably why. I just checked, and it seems I'm using Linux xenia. I was using ge-proton11 for the sdoj recomp. I'll give it a shot with Linux when it's ready!

[SidonCryerd]

are you running the windows build? potentially could be because of that. the next releases will also have a linux build bundled

I really want to run it on the Steam Deck; I hope a Linux port comes out.

[eandis]

https://github.com/eandis/SDOJ-Recomp/releases/tag/prerelease

included a linux build in the latest build. test it out and let me know how it runs

[SaltSudoku]

included a linux build in the latest build. test it out and let me know how it runs

Tested out the Linux build on my end. No issues using the default/provided instructions. (Don't have a Steam Deck to test it out on that device)

[rhossack]

https://github.com/eandis/SDOJ-Recomp/releases/tag/prerelease

included a linux build in the latest build. test it out and let me know how it runs

Linux version has the same Steam Deck performance issues as running Windows in proton, but additionally has the message "Capturing Vulkan. Frame: . 35 ms (27 FPS)
F12, PrtScrn to capture. 0 captures saved." stuck at the top of the screen

[SidonCryerd]

Steam deck

The game didn't work; I even created the extraction folders, but it wouldn't open.

[eandis]

@rhossack will look into performance more, it might be because of one of the latency patches to the game, but i haven't measured the performance difference between vanilla/patched yet. In the meanwhile, can you try running it on xenia and tell me if the game runs full speed there?

@SidonCryerd are you launching through the launch.sh file? does the game crash at runtime or does it not launch at all? if a log file was produced please send it here as well

[rhossack]

@rhossack will look into performance more, it might be because of one of the latency patches to the game, but i haven't measured the performance difference between vanilla/patched yet. In the meanwhile, can you try running it on xenia and tell me if the game runs full speed there?

I found out that something in some update, possibly proton, broke the version of xenia canary I had. So I can't test anything with xenia currently. I can hear the game, but all I see is a black screen after the intro video. But Mushihimesama Futari ver1.5 ran full speed with no issues a few months ago.

Edit: I updated to a new, linux version of xenia, and it runs the sdoj iso full speed.

[SidonCryerd]

@rhossack will look into performance more, it might be because of one of the latency patches to the game, but i haven't measured the performance difference between vanilla/patched yet. In the meanwhile, can you try running it on xenia and tell me if the game runs full speed there?

@SidonCryerd are you launching through the launch.sh file? does the game crash at runtime or does it not launch at all? if a log file was produced please send it here as well

It seems like no logs were generated; is there a specific folder?

[eandis]

@rhossack will look into performance more, it might be because of one of the latency patches to the game, but i haven't measured the performance difference between vanilla/patched yet. In the meanwhile, can you try running it on xenia and tell me if the game runs full speed there?
@SidonCryerd are you launching through the launch.sh file? does the game crash at runtime or does it not launch at all? if a log file was produced please send it here as well

It seems like no logs were generated; is there a specific folder?

in the launch.sh file:
edit --user_data_root="$ROOT/user_data" to --user_data_root="$ROOT/user_data" \
and on the next line add --log_file="log.txt"

@rhossack will look into performance more, it might be because of one of the latency patches to the game, but i haven't measured the performance difference between vanilla/patched yet. In the meanwhile, can you try running it on xenia and tell me if the game runs full speed there?

I found out that something in some update, possibly proton, broke the version of xenia canary I had. So I can't test anything with xenia currently. I can hear the game, but all I see is a black screen after the intro video. But Mushihimesama Futari ver1.5 ran full speed with no issues a few months ago.

Edit: I updated to a new, linux version of xenia, and it runs the sdoj iso full speed.

thank you for testing. i will see what i can do

[SidonCryerd]

No Logs

[eandis]

weird. add a \ after the --log_file line and on the next line add --log_level="debug"

[eandis]

Hello! added an option to turn off the input lag fixes via launch arguments. Try running the game with one or both patches disabled, and see if it improves performance.

https://github.com/eandis/SDOJ-Recomp/releases/tag/prerelease2

[rhossack]

Hello! added an option to turn off the input lag fixes via launch arguments. Try running the game with one or both patches disabled, and see if it improves performance.

https://github.com/eandis/SDOJ-Recomp/releases/tag/prerelease2

On steam deck, I'm not sure what changed but it seemed to run a bit better than the previous version. Aside from that, running with both args was better than with neither, but both clearly had more slowdown than was intended. 2 questions:

1. Is there a way to turn off the text at the top of the screen about vulkan frames and such?

2. Exactly how is the slowdown meant to be presented? When intended slowdown occurs, should the frame rate of the image being rendered remain at 60 fps, or is it expected for the framerate to match whatever intentional slowdown is happening?

Either way, I know the slowdown is unintentional because it's heavy and noticeable even when only a few bullets (<10) are on screen, but it feels almost intentional because the more stuff is happening on screen, the (significantly) worse the slowdown gets. It's a lot worse without the params than with, but it's so slow even with the params that I didn't bother trying just one or just the other.