Emulator

Well, I think we found a game that had hidden bounding box usage. Were there any effects broken before that point?

Thanks for the quick response.

Broken effects? Do you have anything specific in mind that I should look for?

Interesting. I'll check it out.

Bounding Box is used for a lot of interesting effects... I couldn't think of how Resident Evil uses it though.

Your computer is pretty slow (no offense, if it was running it fine before, then we don't want to have a performance regression anyway) and bounding box is very expensive. We enabled bounding box by default thinking there were only 4 games that used it, and those games needed it to function. We didn't expect any other games to be affected... but I believe someone put forth the possibility that other games could enable it and do nothing with it... which is a possibility here.

Ok I understand but isn't there a way to enable the Bounding Box with game ini only for those few games that actually need it? Also I believe that killing Resident Evil in such a way would be rather disappointing for many people. Of course it is up to you what the best approach should be.

The reason we enabled Bounding Box by default was to see if it affected any other games. Because of this, I /think/ that this will cause us to put the option back in unless there's a game breaking issue it fixes.

I'd rather actually see if there are more games. Is there still a way to manually override Bounding Box in the INIs?

No, I removed the setting altogether. I want to test the game to figure out what values are being spewed out by bbox, perhaps there is something different with or without bbox in use.

Also 'The Sky Crawlers: Innocent Aces' huge speed down.....
4.0-3657 -> 60fps
4.0-3659 -> 5fps

If we were to bring the option back, it would be preferable if it were a "speed hack" type option. i.e. the option should Disable BBox emulation.

I've been checking and for some reason, Resident Evil is clearing the bbox regs (thus turning on bbox calculation), but then it never uses them. This makes things very slow for no graphical changes at all, the speed hit is roughly 50% in my computer.

Can you think of a way to detect this and automatically disable BBox calculation?

This is the exact thing we feared happening. I guess we have to have an option for disabling the bounding box emulation for this particular case. What a shame.

Skidau there is no way to know if the bbox regs are going to actually be used when they are first cleared.

The best I can do is to add a check to determine if bbox regs were used the last time they were cleared and if they were not, bbox calculations wouldn't be enabled the next time the regs were cleared.

That sounds like it would work in most cases.

I think so too. I'll try it out and see what happens ;)

After a fast check, it seems to be working. RE is fast again and Paper Mario is still showing the effects.

I'll test it a little bit more tomorrow and if all is well I'll open a PR with the code change.

Hi, is there any progress with this issue? I'm asking because I saw the monthly Dolphin progress report and there the current issue is mentioned as already solved although it still isn't.

Any update on this? If we can't find out a good way to auto-detect, re-adding the option is fine.

Apparently the issue got fixed recently by 4.0-4219 - OGL: implement bounding box support with ssbo.
Thank you degasus!

I would love to do for DirectX what degasus did for OpenGL, but it's way beyond my current knowledge.

I'm having no luck with the viewport fix (see my pull request) so this issue is still pending :\

@crudelios Don't worry, it's also way beyond my d3d knowledge. :/

But it won't be as hard as my patch for someone who has already worked a bit with d3d. All VideoCommon related parts (eg async operation between threads) are already done.

I know for sure I don't have enough D3D knowledge. What would be the equivalent type to SSBO?

This is fixed now on D3D.

It was said by one of the super devs(degasus) that there should still be a change to disable BBox like phire's suggestion on your PR Crudelios.

So I don't know if the status should change here and a new issue report be made??
Because I'm pretty sure any performance impact that OP wanted fixed is now done.

Fixed by 4.0-4555 (ff4526b4a9b7dbcbdbaff35a657739e272aea637)