Dolphin, the GameCube and Wii emulator - Forums

So there will be new API's coming out that will help reduce the overhead for the CPU and Dolphin relies heavily on the CPU for the performance. So are you guys planning on supporting the new API's ?

It's less of a "we need it" and more of a "it will be there when someone wants to work on it".

IMO, literally the only thing Vulkan would be good for is improving performance for AMD Linux users.

(05-22-2015, 01:16 AM)Anti-Ultimate Wrote: [ -> ]It's less of a "we need it" and more of a "it will be there when someone wants to work on it".

IMO, literally the only thing Vulkan would be good for is improving performance for AMD Linux users.

You are wrong sir

Sonicadvance1: I'm also doubtful about our benefit of vulkan. It's designed to reuse modern "display lists", but we can't do so. So in the end, we'll still ask the driver to do all of those stuff all the time. We also can't build them up threaded easyly (maybe with a second level fifo....) In my opinion, it's not that unlikely that vulkan will be slower than a good OpenGL implementation.

Apparently the next Mac OS will be stuck with OpenGL 4.1 since Apple is moving to Metal. I hope that the Mac build of Dolphin will be able to switch to Metal, since I noticed a few new features that need 4.3 or 4.5.

(06-11-2015, 01:21 PM)wdarkk Wrote: [ -> ]Apparently the next Mac OS will be stuck with OpenGL 4.1 since Apple is moving to Metal. I hope that the Mac build of Dolphin will be able to switch to Metal, since I noticed a few new features that need 4.3 or 4.5.

I think they should wait and see if Vulkan will be available for OSx. We don't need more closed and single-platform APIs.

In theory Dolphin is the best usage case for DX12, its heavily CPU bottlenecked. Its more complicated than that however:

1. Emulation is not just rendering the game on a PC. Its a total clone of the architecture implemented in software, which requires CPU time. Alleviating the calls between the CPU and GPU would help performance but not by the miraculous numbers that are being claimed for modern games. Those games might see those improvements because they make a lot of draw calls due to fancy lighting engines and high polygon models, something which Wii and Gamecube never were intended for.

2. Moving to another API is a long, complicated process. With something as complicated as Dolphin there would almost certainly be problems cropping up months or even years later. In my opinion the performance is already very good for people with modern hardware. Developer time would be better spent perfecting what exists and as time moves on there will be even fewer performance problems as hardware with modern OpenGL support and good IPC saturates the market.

Just my opinion, if someone knowledgeable thinks I am wrong please enlighten us.

Well wait, I heard dx12 could utilize built in gpu's to improve performance from post processing effects. Couldn't it be used in dolphin for that purpose as well? Ishiruka's build is very post processing heavy due to the ssao shader and probably whatever else Tino decides to include.

Between those two, I prefer the one that is compatible with 3D Vision. And that will probably be DX12 (I hope).