Hello fellow Dolphins
So, I've read that a high-end CPU matters more for great performance in Dolphin than a high-end GPU. But with normal PC-games, the GPU matters most.
I have an AMD FX-8150 with an AMD7850 - It's great for normal pc-gaming, but the CPU isn't really great for Dolphin.
In the future, will it then be possible that a great GPU matters more than the CPU? Will that be possible with optimisation for the GPU, so that a REALLY high-end dual-/quad- isn't necessary anymore?
Excuse me for my bad English btw, I hope you understand me and thanks
Greetz!
Hehe aw that sucks, could you perhaps explain why, please?
I'd rather have you explain why you think GPUs can do the same kind of jobs CPUs are good to do.
(08-10-2012, 06:39 AM)delroth Wrote: [ -> ]I'd rather have you explain why you think GPUs can do the same kind of jobs CPUs are good to do.
I already expected that question lol.
What I mean is: My CPU+GPU can play all new games on ultra with decent fps. Why is it that I need a better CPU to play SMarioG fluently, while the graphics of the Wii are much worse than the new PC-games from today?
So that's why I'm thinking: could we get better performances if Dolphin is better optimised for GPU's (and CPU's apparently?)
Because gamecube and wii games are not meant to be played on pc,and cpu can "translate" them so you can play them.
Even if all of the "translations" were done on gpu,it would be still slower then regural games.
Imagine reading book in english and then reading book in latin ( I presume you dont speak latin )
Even if you have dictionary,reading book in latin would take a lot more time
It doesnt metter even if book in latin was simple child's book,and book in english was dealing with more advance terms and context,you would still read book in english faster.
To answer the OPs original question (which is vague at best so I'll answer both possible interpretations).
1. The CPU and GPU perform different tasks in dolphin. Optimizing the GPU code better would only reduce the load put on the gpu, it would not affect the cpu requirements. In other words optimizing the GPU code better would only help the GPU do its task faster, it would not help the cpu get its work done any faster.
2. GPUs are not designing to run the same type of code as a cpu. A cpu emulator has far too many function calls and data dependencies to run well on a GPU even if it were possible to port it over. Programming limitations exist even with openCL and direct compute that make certain tasks impossible or borderline impossible to do on a gpu, so there is that to consider as well. And finally JIT recompilation would be borderline impossible, you would need to make an interpreter.