@Weedy_Weed_Smoker I tested it, here is the result:
- ReARMed interpreter version “r20 from December 2014” / works perfect!!! Unloaded from RAM, loads save from snapshots, and loads those games that previously freezes!
- ReARMed GPU set as peops / application crashes.
We must try to fix the code and compile with dynarec/neon (it seems we are one step away from success).
The fix code is here!
@dagazik I compiled another version from December 2013, the fix you linked to is not affecting iOS but generic arm platforms…
Similar behavior(r19 version from December 2013) - application crashes. 
Application to crash from version ReARMed March 2016 and older(we tested 2016 - 2013).
@dagazik From the bug report you linked, it seems to me PCSX ReARMed with dynarec never worked on iOS previous to 6.1.2…
Can you find evidence it ever worked with dynarec on iOS 5?
In the meantime, I think I will put the “_interpreter” on the buildbot, and just compile the latest version…
@Weedy_Weed_Smoker Of course!)) Here is my video: https://www.youtube.com/watch?v=W3yJ9tax5-8
Here work with dynarec(otherwise it would have a low FPS on the processor A4). We started with what works but random freezes.
@dagazik Yeah, I meant “worked flawlessly”… (or at least without freezes and crashes…)
I’ve put both latest versions of the interpreter and dynarec core on the buildbot for now…
@Weedy_Weed_Smoker In new versions ReARMed (2021 - 2018) on:
In other words, the error occurred in the same places, but with different behavior.
Then when we rolled back to version ReARMed r22 from 2016: @AfonsoH application worked well, but my application crash(except for the version “_interpreter”).
It looks like we are dealing with two errors at once:
Please try to remove the space after “USE_DYNAREC = 1” in “Makefile.libretro” version r20 from December 2014. And compile one last time. 
Need exactly the version r20 from December 2014 because I have already checked the work of _interpreter.
Version r20 from December 2014 doesn’t have a space after USE_DYNAREC = 1 in Makefile.libretro, but I think it only happened with armv platform, never for the ios platform…
The commit you linked had the spaces, but removes them:
# iOS
else ifneq (,$(findstring ios,$(platform)))
ARCH := arm
USE_DYNAREC ?= 1<- no spaces
…
ifneq (,$(findstring armv,$(platform)))
…
USE_DYNAREC = 1 <- here are the 3 spaces@Weedy_Weed_Smoker Thank you. And it’s a pity of course that it didn’t work out.
There can be no proof of flawless performance. The Russian forum in 2015 discussed the possibility of compiling RetroArch for iOS 5.1.1, but the participants considered it impossible.
You did not know this and just compiled the version with iOS5 support. 
Thanks to this, Ipad 1st gen managed to launch ps1 games - no one did this before (there was a psx4all emulator, but it was intended for an iPhone screen and worked very bad).
@Weedy_Weed_Smoker Here’s what I thought: there is a small chance that errors are not related, in which case there may be a version between March 2016 (application crashes / that definitely this is due to the dynarec on iOS5) and jul 2018 (application randomly freezes / reason is unknown).
In theory version ReARMed 2017 - which can works perfect.
Try renaming the bios to UPPERCASE on the older crashing versions… (“SCPH101”, “SCPH7001”, “SCPH5501”, “SCPH1001”)
Versions prior to 8 Jun 2018 can only find the bios if it’s uppercase… 
changed
*bios[] = { "SCPH101", "SCPH7001", "SCPH5501", "SCPH1001" };
to
*bios[] = {
"SCPH101", "SCPH7001", "SCPH5501", "SCPH1001",
"scph101", "scph7001", "scph5501", "scph1001"
};@Weedy_Weed_Smoker Did not work out. I checked your versions 2013, 2014, 2015, 2016 (neon edition) - used all the recommended bios from the code.
@dagazik I fixed that in the iOS9 version from 2016 that works and I tried once again to compile the same version for iOS with this fix included, but I guess it didn’t fix your problem…
You can still try to update the core from within RetroArch to check if it’s still not working… Try to delete the configuration specific to PCSX ReARMed in the config file(s)!
@Weedy_Weed_Smoker Version from the repository also did not work.
I remove this before every test: /User/Documents/RetroArch/config/PCSX-ReARMed
Have you tried with different BIOS? I remember a while back ago that I had different luck with different BIOS (some would not load content, others would but crash immediately, others would take more time to crash). Maybe trying one at a time (just one BIOS file in the directory)?
@AfonsoH Yes, I tried different combinations - core crashes on any game just when you try to start it.
I understand the dependence of the quality of emulation and bios, but in this case it is not him. In addition, the non-accelerated version of the core work good.
I opened *.ipa packages of versions RetroArch 1.6.7, 1.5.0, 1.3.6 to extract the compiled core from them(using 7-zip): result of all versions of “_interpreter”(from 2017) is the same(but this time the behavior has changed): doesn’t unload from the RAM, doesn’t load snapshots, games freezes(earlier versions without acceleration worked perfectly).
Now it became clear how theory this is possible to solve: using the basic code of from ReARMed 2013-2016, and the acceleration code from ReARMed 2017-2021.
I think that If application crashes on iOS9 and freezing on iOS5 are cause of same code, then there is a hope that the developers fix it in new versions. But if the reason is different code, then It is doubtful that someone will be engaged in this, since the iOS5 platform has long been not target, and it seems that a simple selection of the version cannot fix this bug.
@Weedy_Weed_Smoker Some new data abaut bug ReARMed on ios5: I tried to load PicoDrive x32 games and they same cause the application to crash. Now i think maybe ReARMed in dynarec mode and PicoDrive in x32 mode use the same component apps/system. 
To eliminate the possibility of a problem is in hacking code i reset iPad to factory settings and jailbreak him via Absinthe 2.0.4 - but this did not solve the problem.
Can this information somehow help in solving?
@Weedy_Weed_Smoker I had a little time and I decided to return to this(want to accurately identify the version with which the core ReARMed is no longer unloaded from RAM to iOS5). But I need help because for iOS I compile the software for the first time I don’t quite understand what parameters need to be passed, as well as what libraries are missing in the system to successfully complete the compilation.
Here’s what I use: VirtualBox 6.1 + MacOS 10.8.5 (clover as a bootloader for disguising smbios) + XCode 4.6.3 + apple command line tools + MacPort package manager.
As a test of the functionality of this set: successfully compiled FCEUmm using the libretro-super script and tested it on a real device with positive results! 
But I can’t get PCSX ReARMed to compile regardless of whether I compile it directly using the ./configure tool or using the libretro-super script.
Since you have already managed to compile this core, I just want to ask: What parameters did you set to ./configure ReARMed for IOS5.1.1 + SoC A4?
@Weedy_Weed_Smoker This is the command I use for configuration build:
CC=“cc -arch armv7 -marm -miphoneos-version-min=5.0 -isysroot /Applications/Xcode.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS6.1.sdk” CXX=“c++ -arch armv7 -marm -miphoneos-version-min=5.0 -isysroot /Applications/Xcode.app/Contents/Developer/Platforms/iPhoneOS.platform/Developer/SDKs/iPhoneOS6.1.sdk” ./configure
Based on the config.log, I need a following libraries(libpng i compiled for armv7 from source and list has decreased a little ):
I don’t understand what it is: <sys/soundcard.h> and <alsa/asoundlib.h>
Do I really need to compile each library from source? 
I’m talking about something that is possible, somewhere there is a ready-made package for armv7 (that I could not find).
