0:00 / 0:00
PS3 emulation is fast on ARM now
Whatcookie
28K subscribers • 2 videos on Hallway
- Recommended by 1 person: 1 on Lobsters.
Advanced. Deep dive for people who know the field.
302K views2 months ago1 recommendation
Systems programmingGraphics and audioPerformance and optimizationCompilers and interpretersCPUs and computer architecture
Whatcookie explains the low-level assembly, SVE vector extensions, and compiler optimizations used to close the performance gap for RPCS3 on ARM.
In this video, I cover 6 months of work done on RPCS3 to close the gap between ARM and x86 performance. I've taken a seriously comprehensive approach with this video, so grab some food and get comfortable. You're in for an entire hour of optimizations, x86 and ARM assembly, and glorious framerate improvements. Intro Busy wait shenanigans Don't use Yield in place of Pause! What was LLVM doing on ARM? How we optimized SHUFB How we optimized the rest of the instructions The most optimized way to compare data on ARM How to play wow optimally SVE are the *special* vector extensions Optimizing RPCS3 with SVE ONE MORE FINAL: I need you Optimizing via hardware wait instructions Near miss and lightning round optimizations Half a Steam DeckWhy it's here
1 person recommends thisWhat people say. Recommended as an exceptionally detailed and educational deep dive into low-level ARM assembly and emulation optimization.