[RPi 3] Optimized lr-snes9x using PGO

Griever

@darksavior As @hhromic said, links to patches are included in the OP. As for the profiling data, I'm not sure how reusable it is and wouldn't want to promote reusing data for older versions of snes9x

Do be warned that compilation can take 15-20 minutes and profiling will be very, very slow.

edit: Going by every other project I see that uses PGO, I doubt the data is reusable. Automating profiling isn't really feasible without macros and savestates and that's just too much effort for me.

hhromic

Right, profiling data is not entirely reusable when updating the binary. You are right, I overlooked that.
Indeed automating the profiling is not trivial to do :(

MapleStory

Pardon my ignorance but how does one go about building lr-snes9x from source with the optimized patches?

Darksavior

@Griever Yes, if it's not too much trouble, a brief tutorial would be nice on how to use the patches. I know enough to dive in the .sh installer file and change stuff.

Darksavior

With snes9x 1.57 out I finally decided to take a stab at this and manually edit the changes since I have no idea how to use patches. Seems to be working.

MapleStory

@Darksavior

Did you do anything other than copy Griever's RetroPie-Setup changes? I replaced my system.sh and lr-snes9x.sh files to match his, updated lr-snes9x from source and despite now being on 1.57, the core runs worse than the binary he posted in the original post of this thread.

Darksavior

@MapleStory I had that problem at first until I read his first post which says

"Note that you need to comment out CXXFLAGS, LDFLAGS and uncomment the other CXXFLAGS variable when generating the optimized binary."

Do that and you'll be good to go.

#CXXFLAGS+=" -fprofile-dir=/home/pi/pgo/out -fprofile-generate=/home/pi/pgo/out"

CXXFLAGS+=" -fprofile-dir=/home/pi/pgo/out -fprofile-use=/home/pi/pgo/out"

#LDFLAGS+=" -lgcov"

MapleStory

@Darksavior

Thanks, made the changes and it compiled successfully but Kirby's Dreamland 3 cat/hamster room fluctuates between 58.7-59fps. Doesn't get to a clear 60fps. Griever's binary in the first post, however, maintains a stable 60fps throughout. Other than compiling lr-snes9x from source after editing the scripts, was there any other updates I should've done?

Darksavior

@MapleStory I only tested it against the official build on a pi3b non-plus at 1300 and it was an improvement over that one. It's still not as fast as the Griever's build. I wouldn't know what else would be required to improve it. A pi3b+ user should be fullspeed on 1.57 unless you use the slowdown fix hack. This is my first time editing building options so I'm rather new to this.

mitu

@MapleStory The final binary use profiling(performance) data - i.e. you run the emulator a few times, in the games you're interested on improving performance, and some performance data is generated. Then you re-compile the binary again, this time the compiler reads the performance data and produces another binary that takes into account this perf data, with some new optimisations added based on the data.

That's why @Griever said it's difficult to automate the build of such compiler, it's practically non-automated if you want to have meaningful results.

pjft

So, I've been meaning to ask in case anyone had tried it out: if I use the binary above on a pre-RA 1.57 setup, will it still be an improvement/will it at least run, or are the improvements dependent on updating RetroArch as well? I'm not keen on updating it as I have a stable, working setup and I'm under the impression I read a few things changed on the performance front.

Darksavior

@pjft Try it. Be adventurous and just go for it. Back up your build, of course.

mitu

@pjft You can try the core provided, but for sure the save states wouldn't work and there have been some modifications in the recent core version to accomodate the run-ahead feature introduced with RA 1.7.2 (?), so they might be imcompatibilities.

pjft

@mitu @Darksavior Sounds good. I'll create a separate emulator entry and try it out.

Thanks.

MapleStory

The RetroPie-setup script updated today and now I can't update lr-snes9x from source with the PGO changes. Anyone else having this issue?

EDIT: Found the problem. So it seems like there's a limit to how many games you can profile, otherwise it will crap out during compiling the optimized core. I probably profiled 15-20 games, so I think it's better off to just go with 5-10.