Each application is publicly hosted on GitHub. Here's a website containing links to all the source code: https://github.com/Nun-z/SourceCode
I create the packages myself from the source code linked above. I have an entire infrastructure dedicated to making these packages and automatically uploading them to the Microsoft Store, which you can see here: https://github.com/Nun-z/AzureXboxScripts and here: https://github.com/Nun-z/NodeAzureUpptime.
Here's a guide to doing this manually: https://github.com/Nun-z/XboxRetailModeAppGuide Note that the process is difficult at first and can be a struggle to complete. In the Pricing and Availability settings, set visibility to "public audience" and set discoverability to "direct link only."
As far as I can tell, all available options are more or less the same. You can use any device with a web browser to install these apps to your Xbox (rather than going to the website on your Xbox's browser). Right now I am doing extensive testing to determine if there are significant differences in performance across the different versions.
Initial results show a difference of 0.2 seconds when recording the amount of time it takes each version (gamr13 and Nun-z) to progress through the same cutscene. Over a 36-second cutscene, this is equivalent to a difference of approximately one half of one percent. Here are the raw recordings for proof: https://youtube.com/playlist?list=PLJjTcel_snDinKrK4yzc99SdNStKXpm-6 . The last 5 videos will be uploaded when the daily YouTube limit resets. My methodology was to record 10 videos each of RetroReveal and gamr13's offerings, all while the intro cutscene to Shadow of the Colossus was playing. I then cropped the videos so they all started at the same time and all ended just after the hawk flew past the main character. Finally, I found the average video length of each version's cutscene and compared them. The average difference was 0.2 seconds. Here's a "side-by-side" comparison, with the fifth gamr13 video on top and the fifth RetroReveal video on the bottom: https://youtu.be/wTKnvcVGaqE. To address another objection, you can see that the gamr13 video is on top by comparing it to the fifth gamr13 video (available here: https://youtu.be/hfswoZY_UcA ) and the fifth RetroReveal video (available here: https://youtu.be/lDP1bQI6Xoo ). For something to look for, notice in the "side-by-side" video that the FPS for the top video begins at 56.60 FPS and the FPS for the bottom video begins at 56.19 FPS. Go to the fifth gamr13 video I linked to and go to 1:24. You will see that the FPS is listed as 56.60. Now go to the fifth RetroReveal video I linked and go to 1:17. You will see that the FPS is listed as 56.19.
Certain people believe that this is achieved through frameskipping. It's very impressive that some are so perceptive that they can see things that aren't actually there. In each video, you can see that frameskipping is turned off in the quick menu. For example, take a look at 0:43 of this video: https://youtu.be/iWs3_6c0QME. The frameskipping is clearly set to off. This is why I went through the quick menu in each video. Also, I am using identical self-dumped SCPH-7701 BIOS for all PCSX2 testing.
If you still have doubts, send me an email and tell me what proof you want.
I will next be testing Dolphin.