FEX-2604 released
The FEX-2604 update delivers significant memory savings by enabling Dynamic L1 lookup caches and optimizing Transparent Huge Page usage, which helps systems with limited RAM like those with 8GB or 16GB. Performance sees a major boost for games using x87 transcendental operations as SIN, COS, and TAN functions are now inlined within the JIT for an average speed increase of 3.7 times. Several critical bugs have been resolved including a workaround for Docker seccomp filter crashes and a fix for a pseudo-memory leak in the thread-pool allocator that previously ballooned resource usage during heavy threading. Advanced users can now utilize a new FEXGetConfig option to visualize memory alignment faults on ARM hardware, providing better insight into emulation performance gaps compared to native x86 behavior.
FEX-2604 released
The FEX-2604 update introduces significant memory optimizations, particularly beneficial for systems with limited RAM, such as those with 8GB or 16GB. Performance improvements have been made for games utilizing x87 transcendental operations, yielding an average speed increase of 3.7 times by inlining SIN, COS, and TAN functions within the JIT. Several critical bugs have been addressed, including a workaround for Docker seccomp filter crashes and a fix for a pseudo-memory leak in the thread-pool allocator that previously increased resource usage under heavy threading. Advanced users can now visualize memory alignment faults on ARM hardware through a new FEXGetConfig option, enhancing insight into emulation performance compared to native x86 behavior
