big_guitar Posted February 11 Share Posted February 11 (edited) 18 hours ago, GB_Baker said: Maybe I'll compile Stella 6.7.1 next. Would that be useful? I meant mostly from the 'beyond Mame2003' perspective, but I think most of those would be too much for the GSP to handle. In most cases you'd think a native build would be better (such as with Stella), but in Windows I've seen better mednafen performance under Windows-RA than with the stand alone, perhaps because they are parsed out into multiple library cores instead of a single core library. When I tried those other Jungle releases beyond 2003 and what I found online for Mame, it seemed to take too long to load from the GSP hardware (although mame2003-plus loads just fine on GSP!). Then again, it could have been the myarcade version of RA not able to handle it. I look forward to testing your 1.17 build! It will be nice to use a keyboard for tweaking as needed (amongst other things). Edited February 11 by big_guitar Quote Link to comment Share on other sites More sharing options...
GB_Baker Posted February 11 Share Posted February 11 5 hours ago, fluxit said: Are the assets required? I have your binary and cfgs in /sdcard/retroarch, and your runme.sh in /sdcard/agsp. Cores I've added are in /sdcard/retroarch/cores. My 'Xbox 360 pad' works great in the menu. I can "Load a core," and the selected core's info is displayed in the lower left as expected. When I then "Load content," I'm able to navigate to my Games directory and select a rom. Retroarch comes back with "No cores available." I see the same behavior, regardless of the selected core. These are cores and roms that I've previously tested successfully with the GsP's included RetroArch. My "Romoral K808" macro keypad with scrollwheel also works great in the menu, including using the "Enter" key to select options. I did download the databases so when I load a ROM I am usually presented with option to select a core. Check my config for all of the locations. Maybe you are missing some directories like core-configs. 1 Quote Link to comment Share on other sites More sharing options...
FatherOfJanitor Posted February 11 Share Posted February 11 I am new to the GSP and have updated the firmware. I am trying to add games. I have downloaded the Atari Age bundle but whenever i try copying to the sd card I get invalid file name or length. What am i doing wrong Quote Link to comment Share on other sites More sharing options...
Vic20Ian Posted February 11 Share Posted February 11 On 2/10/2024 at 7:21 PM, Vic20Ian said: Custom firmware experiment using Brad_from_the_80's install. I played around with TOP RACER 2 which has Hard set to 4. I changed it to varous values including 5 and nothing changes with the controls or gameplay. It remains A Brake, B, Accelerate, C Nitrous. Prhaps Hard determines if the game goes on the Paddles menu and the launch for those games. I can't see it doing much else at the moment. I ran Warlords for 2600 and Arcade set to Hard 0 today instead of 5. No difference detected. I think the Hard field in the database has maybe been superseded by the Gametype. Quote Link to comment Share on other sites More sharing options...
fluxit Posted February 12 Share Posted February 12 9 hours ago, GB_Baker said: I did download the databases so when I load a ROM I am usually presented with option to select a core. Check my config for all of the locations. Maybe you are missing some directories like core-configs. Hmm.... I'm still not able to load a core, then select a game. I do, never the less- have a suggestion. I don't recall what input_menu_toggle_gamepad_combo = "2" represents, but it isn't intuitive, IMO. input_menu_toggle = "home" is probably better. This uses the home button on the console(gpio) to bring up the RetroArch menu when a core is running. While in the RetroArch menu, the button still acts like "Home" on a keyboard, IOW it takes you to the top of the list of options. Another suggestion, is to add the env variables: export XDG_CONFIG_HOME=/sdcard export XDG_RUNTIME_DIR=/sdcard to runme.sh. Otherwise, RetroArch will create it's directory structure in the default usable location- the root of the ext2 data partition. A partial set of overrides in the .cfg file will not keep this from happening, in my experience. I'll go back over my .cfgs later, but I'm stumped as to why I can't actually load a game. I've even tried "installing" my cores from the download directory. RetroArch indicated success, but there was no change in it's behavior. I load a core, and when I try to load a rom I still see "No cores available." It really doesn't make sense. I know that the cores are loading, because the file filter is working. 1 Quote Link to comment Share on other sites More sharing options...
fluxit Posted February 12 Share Posted February 12 RetroArch produced the attached interesting, but not useful logfile, with logging enabled. retroarch__1970_01_01__00_00_29.log Quote Link to comment Share on other sites More sharing options...
GB_Baker Posted February 12 Share Posted February 12 27 minutes ago, fluxit said: Hmm.... I'm still not able to load a core, then select a game. I do, never the less- have a suggestion. I don't recall what input_menu_toggle_gamepad_combo = "2" represents, but it isn't intuitive, IMO. input_menu_toggle = "home" is probably better. This uses the home button on the console(gpio) to bring up the RetroArch menu when a core is running. While in the RetroArch menu, the button still acts like "Home" on a keyboard, IOW it takes you to the top of the list of options. Another suggestion, is to add the env variables: export XDG_CONFIG_HOME=/sdcard export XDG_RUNTIME_DIR=/sdcard to runme.sh. Otherwise, RetroArch will create it's directory structure in the default usable location- the root of the ext2 data partition. A partial set of overrides in the .cfg file will not keep this from happening, in my experience. I'll go back over my .cfgs later, but I'm stumped as to why I can't actually load a game. I've even tried "installing" my cores from the download directory. RetroArch indicated success, but there was no change in it's behavior. I load a core, and when I try to load a rom I still see "No cores available." It really doesn't make sense. I know that the cores are loading, because the file filter is working. Good tip on the env variables. Does the runtime directory override the default path to the internal retroarch binary? Maybe unnecessarily, I created the sym link in /tmp. I've never seen it create a directory structure there. The debug console does flash a bunch of lines saying that it tried to write to a location that was read only while trying to create folders. The menu combo that I use is L3+R3 (pressing the thumbsticks on the xbox controller). In mame 2003 it is L3 to get to the game menu. I have found that when I reload a different version of the same core that it will crash the first time. Second load always works. I think it has to clear some database with old information. I have also found the regular armhf cores seem to work pretty well. http://buildbot.libretro.com/nightly/linux/armhf/latest/ I am in the process of testing cores that I have compiled. The libretro-build.sh script for building cores is not super clear on neon and hardfloat options. I am trying to force it with -mfpu=neon-vfpv4 -mfloat-abi=hard in the arm-linux-gnueabihf-gcc commands. So far mame2003 seems to be working fine. This is what I have cooking now: prboom ecwolf mame2003_plus virtualjaguar stella2014 atari800 a5200 quicknes mgba fbneo genesis_plus_gx snes9x2010 mupen64plus_next parallext parallel_n64 gearcoleco. Any specific core requests? The ones on Zoltans site seem super buggy to me. Quote Link to comment Share on other sites More sharing options...
GB_Baker Posted February 12 Share Posted February 12 1 hour ago, fluxit said: RetroArch produced the attached interesting, but not useful logfile, with logging enabled. retroarch__1970_01_01__00_00_29.log 8.32 kB · 1 download Here's what one of mine looks like with all logging turned on. Lots of testing Doom. Had to dial down the FPS from 60 to 35 in the game otherwise it was too fast. In the core setting resolution of 640x400 was best for Prboom. retroarch__1970_01_01__00_03_19.log Quote Link to comment Share on other sites More sharing options...
GB_Baker Posted February 12 Share Posted February 12 A few home brew cores. I think I compiled with neon and hardfloat. Not sure how to tell if I was successful. prboom ecwolf mame2003 mame2003_plus virtualjaguar stella2014 atari800 a5200 quicknes mgba fbneo genesis_plus_gx snes9x2010 mupen64plus_next gearcoleco mupen64plus_next for sure does not work. I was surprised it compiled. cores.tar.gz 2 Quote Link to comment Share on other sites More sharing options...
fluxit Posted February 12 Share Posted February 12 (edited) 6 hours ago, GB_Baker said: Good tip on the env variables. Does the runtime directory override the default path to the internal retroarch binary? Maybe unnecessarily, I created the sym link in /tmp. I've never seen it create a directory structure there. The debug console does flash a bunch of lines saying that it tried to write to a location that was read only while trying to create folders. The menu combo that I use is L3+R3 (pressing the thumbsticks on the xbox controller). In mame 2003 it is L3 to get to the game menu. I have found that when I reload a different version of the same core that it will crash the first time. Second load always works. I think it has to clear some database with old information. I have also found the regular armhf cores seem to work pretty well. http://buildbot.libretro.com/nightly/linux/armhf/latest/ I am in the process of testing cores that I have compiled. The libretro-build.sh script for building cores is not super clear on neon and hardfloat options. I am trying to force it with -mfpu=neon-vfpv4 -mfloat-abi=hard in the arm-linux-gnueabihf-gcc commands. So far mame2003 seems to be working fine. This is what I have cooking now: prboom ecwolf mame2003_plus virtualjaguar stella2014 atari800 a5200 quicknes mgba fbneo genesis_plus_gx snes9x2010 mupen64plus_next parallext parallel_n64 gearcoleco. Any specific core requests? The ones on Zoltans site seem super buggy to me. I've not tested to answer your first question. You could just call the binary using its full path though? Not that it matters as what you came up with works fine. Ah yes, the switches under the sticks. I forget that they are there most of the time. The problem is sorted. This version of RetroArch apparently requires the core info, and only the core info AFAICT, in order to function 'manually,' using the menu. Unfortunately, retroarch/info is one of the directories that was not automatically created by the program. Once I unzipped info.zip into /sdcard/retroarch/info, your build of RetroArch is working well for me. Edited February 12 by fluxit 1 Quote Link to comment Share on other sites More sharing options...
big_guitar Posted February 12 Share Posted February 12 (edited) 16 hours ago, fluxit said: RetroArch produced the attached interesting, but not useful logfile, with logging enabled. retroarch__1970_01_01__00_00_29.log 8.32 kB · 2 downloads Do you enable the verbosity flag? and/or pass "-v" ? I've found it very useful to export as much log info as possible (but not debug mode yet) but I'm also exporting to my own file... elif [ $trap == "2" ]; then $RA117 -v --log-file "/sdcard/jlog/retrolog_v.txt"\ -c $activecfg\ --subsystem "cv"\ -L "$libPath" "$rom1" >$retrolog 2>&1; Also you said "This version of RetroArch apparently requires the core info" This makes sense, as my secondary file reports: /sdcard/RA117/retroarch: /usr/lib/libasound.so.2: no version information available (required by /sdcard/RA117/retroarch) My primary issue is it's not picking up my OZONE fonts yet for some reason, but the icons are showing, still have the "???" for the text... I have "regular.ttf" and "bold.ttf" listed in my ozone directory... getting close... mostly using my pre-existing CFG's [with mostly menu-specific changes], but not sure yet if the non-300 range numbers will work in the config file passed, but the controller CFG is recognized. I will say that, the Space Fury under FBNEO is not flickering anymore, so it definitely had something to do with that funky MA version of RA. Nevermind, it seems that Space Fury does still flicker on this hardware with FBNEO... Edited February 12 by big_guitar Quote Link to comment Share on other sites More sharing options...
GB_Baker Posted February 12 Share Posted February 12 51 minutes ago, big_guitar said: Do you enable the verbosity flag? and/or pass "-v" ? I've found it very useful to export as much log info as possible (but not debug mode yet) but I'm also exporting to my own file... elif [ $trap == "2" ]; then $RA117 -v --log-file "/sdcard/jlog/retrolog_v.txt"\ -c $activecfg\ --subsystem "cv"\ -L "$libPath" "$rom1" >$retrolog 2>&1; Also you said "This version of RetroArch apparently requires the core info" This makes sense, as my secondary file reports: /sdcard/RA117/retroarch: /usr/lib/libasound.so.2: no version information available (required by /sdcard/RA117/retroarch) My primary issue is it's not picking up my OZONE fonts yet for some reason, but the icons are showing, still have the "???" for the text... I have "regular.ttf" and "bold.ttf" listed in my ozone directory... getting close... mostly using my pre-existing CFG's [with mostly menu-specific changes], but not sure yet if the non-300 range numbers will work in the config file passed, but the controller CFG is recognized. I will say that, the Space Fury under FBNEO is not flickering anymore, so it definitely had something to do with that funky MA version of RA. Nevermind, it seems that Space Fury does still flicker on this hardware with FBNEO... Did you try with my home brew fbneo? Quote Link to comment Share on other sites More sharing options...
GB_Baker Posted February 12 Share Posted February 12 I think the next thing I'm going to compile for arm is the evtest utility that monitors udev device events. It should capture all of the button numbers for a USB controller. Quote Link to comment Share on other sites More sharing options...
big_guitar Posted February 12 Share Posted February 12 (edited) 28 minutes ago, GB_Baker said: Did you try with my home brew fbneo? Not yet, but I will! Need to figure out the font thing... Although the guy who built the version I tested said he can't duplicate the issue. 21 minutes ago, GB_Baker said: I think the next thing I'm going to compile for arm is the evtest utility that monitors udev device events. It should capture all of the button numbers for a USB controller. There is 'evtest' on the stock GSP which I ran for mine, but I don't think there is jstest... Of course there could be a newer evtest. Edited February 12 by big_guitar Quote Link to comment Share on other sites More sharing options...
big_guitar Posted February 12 Share Posted February 12 (edited) 17 hours ago, fluxit said: interesting, but not useful logfile There must be some setting somewhere I'm missing, when I compare the first 4 lines of our log files... Could be impacting my ozone fonts perhaps... This was reporting "Capabilities" with my previous stuff, so I should be able to figure this out... Edited February 12 by big_guitar Quote Link to comment Share on other sites More sharing options...
GB_Baker Posted February 12 Share Posted February 12 2 minutes ago, big_guitar said: There must be some setting somewhere I'm missing, when I compare the first 4 lines of our log files... Could be impacting my ozone fonts perhaps... Did you pull the most recent assets from http://buildbot.libretro.com ? Winmerge your cfg and mine. Quote Link to comment Share on other sites More sharing options...
big_guitar Posted February 12 Share Posted February 12 (edited) 5 minutes ago, GB_Baker said: Did you pull the most recent assets from http://buildbot.libretro.com ? Winmerge your cfg and mine. Unfortunately, the LINUX-ARMHF builds on that site date back to 2020. There have been many fixes since then. I have been comparing lots of files with winmerge and/or kdiff3, of course I have a lot of custom folders & such... but I'm still looking! Edited February 12 by big_guitar Quote Link to comment Share on other sites More sharing options...
GB_Baker Posted February 12 Share Posted February 12 3 minutes ago, big_guitar said: Unfortunately, the LINUX-ARMHF builds on that site date back to 2020. There have been many fixes since then. I have been comparing lots of files with winmerge and/or kdiff3, of course I have a lot of custom folders & such... but I'm still looking! https://buildbot.libretro.com/assets/frontend/ Quote Link to comment Share on other sites More sharing options...
big_guitar Posted February 12 Share Posted February 12 (edited) 5 minutes ago, GB_Baker said: https://buildbot.libretro.com/assets/frontend/ Thank you, sorry I missed the 'assets' reference, although the 'ozone' folder from the assets.zip matches the folder I have for ozone, but I'll see what else I could be missing or if I'm pointing something wrong elsewhere... Edited February 12 by big_guitar Quote Link to comment Share on other sites More sharing options...
GB_Baker Posted February 12 Share Posted February 12 2 minutes ago, big_guitar said: Thank you, sorry I missed the 'assets' reference, although the 'ozone' folder from the assets.zip matches the folder I have for ozone, but I'll see what else I could be missing or if I'm point something wrong elsewhere... Overlays maybe? Quote Link to comment Share on other sites More sharing options...
big_guitar Posted February 12 Share Posted February 12 35 minutes ago, GB_Baker said: Overlays maybe? Oddly, it got worse when I set input_overlay_enable to false, and set input_overlay to an empty string, and that made the icons go away... so strange... I unzipped the frontend assets.zip to a different folder and pointed assets_directory there, and that didn't make any difference.. probably missing something obvious 😒 Quote Link to comment Share on other sites More sharing options...
fluxit Posted February 13 Share Posted February 13 3 hours ago, big_guitar said: There must be some setting somewhere I'm missing, when I compare the first 4 lines of our log files... Could be impacting my ozone fonts perhaps... This was reporting "Capabilities" with my previous stuff, so I should be able to figure this out... Do you have user_language = "0" in your retroarch.cfg? Quote Link to comment Share on other sites More sharing options...
big_guitar Posted February 13 Share Posted February 13 (edited) 2 hours ago, fluxit said: user_language = "0" Yes indeed, I found this out searching online where people were wondering why CHINESE wasn't displaying properly! 😆 The language setting on all the MyArcade config files! Knew I had missed something basic. Will deal with the controller issues next once I get a chance. I hadn't even thought of it since the RGUI menu displayed fine... Edited February 13 by big_guitar 1 Quote Link to comment Share on other sites More sharing options...
GB_Baker Posted February 13 Share Posted February 13 1 hour ago, big_guitar said: Yes indeed, I found this out searching online where people were wondering why CHINESE wasn't displaying properly! 😆 The language setting on all the MyArcade config files! Knew I had missed something basic. Will deal with the controller issues next once I get a chance. I hadn't even thought of it since the RGUI menu displayed fine... I started with a clean Non-GSP cfg file. Who knows what additional items have been added since 1.7.4. Make sure audio is set to alsathreads and video is threaded. Quote Link to comment Share on other sites More sharing options...
big_guitar Posted February 13 Share Posted February 13 (edited) @GB_Baker Unfortunately experiencing the same screen shakiness with FBNEO+SpaceFury with your build of fbneo as with zoltanvb's last build I tested, could be something just on GSP... (This is a game-specific thing, I have not encountered this with any other games on GSP so far). The voice audio on that game is fine though under FBNeo, not sped up, but display not ideal. I recall MAME2003-plus performing the best with that game, although I believe the voice is sped up there, will try that again with your build. Then again, it's not a big deal to me, I have just been curious about issues encountered related to this game. -controller setup all good now... not sure why the myarcade RA version required those evtest equivalent event codes in the 300s -also interesting that RA 1.17.0 created a 'retroarch' folder off the root of my sdcard... perhaps due to those XDG environment variables? Edited February 13 by big_guitar 1 Quote Link to comment Share on other sites More sharing options...
.thumb.jpg.83a2761de6b278e27919f99f53c87389.jpg)
Like us on Facebook
Follow us on Twitter
AtariAge on Instagram
Watch us on YouTube
Chat on Discord
Recommended Posts
Join the conversation
You can post now and register later. If you have an account, sign in now to post with your account.
Note: Your post will require moderator approval before it will be visible.