@sjw from RPCS3 blog: "To put it simply, TSX allows you to touch a piece of memory and then proceed with a task, but if any other thread touches that exact piece of memory, it aborts all the speculative work done. In the absence of TSX, the thread would have to use a mutex to lock the memory for its sole use, which makes it significantly slower than TSX."

@matrix @sjw To make matters worse to use it you have to disable "microcode updates" because it kept being disabled for your own good of course.
https://en.wikipedia.org/wiki/Transactional_Synchronization_Extensions

I heard they had to really optimize it because of the removal + AMD cpus lacking it.

@PhenomX6 @sjw Well, branch prediction tends to have a ton of vulnerabilities.
Yeah, they did a ton of work around and now it isn't a must and the performance difference isn't big.

@matrix yee that's the main thing that speeds it up because it can better emulate the SPEs and PPE talking to each other over the EIB.
Cell is a great example of why throwing more IPC/higher clocks, more cores, etc. doesn't always work for emulation.