PCSX2 Documentation/Threading VU1: Difference between revisions

''Originally written by cottonvibes''

Well if you've kept up with pcsx2's SVN, you'll notice we recently added a MTVU (Multi-Threaded microVU1) option, which runs VU1 on its own thread.

So now you may be wondering, “If threading has all those problems how can you get a speedup?” Well we had to be smart about it, and figure out a way to make threading work based on the properties of the hardware. One thing we know is that for a component to even be worth-while to thread, it needs to have relatively low interaction with other components (which will limit the amount of syncing needed with other components; and therefore increase the thread throughput), and it also needs to be a component that is used frequently and is computationally expensive to emulate (or else there’s no point in threading it). Luckily the GS fit that bill very well (which gave us MTGS), and to a lesser extent VU1 does (giving us MTVU).

[[File:ps2flow.png]]

By now you should hopefully have a better understanding of how pcsx2’s VU1 and GS threading basically works, but you may still have questions on why it’s taken so long for us to do it. Well as shown above there were a lot of problems with threading VU1 which needed to be solved, and we weren’t sure about the best ways to handle them. Another huge reason is pcsx2 was not in a good state to make VU1 threading a reality until recently. We didn’t have a centralized GIF Unit like the ps2 does; instead we had code that was all over the place and running the GIF paths without proper scheduling.  The new GIF Unit rewrite solved this problem for us. Another problem was that Super VU1, the old VU1 recompiler, was not thread safe (it combines stuff for VU0 and VU1 emulation together), so a new thread safe recompiler was needed, which was one of the goals for the microVU recompiler I wrote. Another problem was the sloppiness of a lot of old pcsx2 code which had needless inter dependencies between various other code modules; we needed to isolate the related code with code refactoring and rewrites to clearly separate the code and make it thread safe. We also needed a thread-safe code emitter which allowed us to run multiple dynamic-recompilers in parallel; we solved this by a code emitter rewrite and using thread local storage for the emitter’s global data. Lastly we needed someone who knew about all these various components and was bored enough to try making something like this work  Very Happy

{{PCSX2 Documentation Navbox}}