OP here. Happy to answer any technical questions about the implementation, especially around the static analysis of the call-graph (DAG) or the BSL 1.1 license choice.
(I also have a GIF demo of it blocking a malicious tool in real-time if anyone is interested, but didn't want to clutter the main post).
Impressive work, especially getting the GPU drivers stable enough for Steam. The PS5's custom I/O complex (that hardware decompression unit) must have been a headache to work around — did you end up bypassing it entirely or did you manage to get any of that throughput exposed to Linux?
I'm curious about thermal behavior under sustained load. The PS5's cooling is tuned tightly to the stock firmware's power envelope; running arbitrary Linux workloads probably hits different thermal profiles. Did you see throttling during long gaming sessions, or is the headroom sufficient when you're not pushing the RSX equivalent?
BC-250 had a standard set of SuperIO and Southbridge, and the PS5 has a duo consisting of a custom NVMe controller and custom SuperIO. If what is true from the PS4 is applicable to the PS5, there had to have been some patches to the kernel in order to add platform support for those custom chips.
As far as I can tell the Twitter poster hasn’t answered any of the hard questions either. Just the screenshots and some comments, but I also don’t have time to go through the hundreds of comments in that thread.
