Another thing you might want to do to improve the performance in Dolphin is change your GPU.
I recommend getting a *last-generation* AMD GPU with the TeraScale2 architecture (HD5000 or HD6000 Series).
The latest AMD GPUs with a GCN architecture (HD7000,HD8000,Rx and Fury Series) are still horribly unoptimized (the drivers have unusually high CPU overhead).
But last-gen GPUs are so optimized they perform even better than NVIDIA's GPUs, especially when using the Direct3D backend (e.g. 2100fps [HD6000 Series] vs. 1100fps [R9 Series] in text/logo screens at native res. That's a ~2x difference in driver overhead.)
Another advantage of getting an older generation AMD GPU is that it doesn't have any of those "smart" power-saving features which introduce a lot of performance-related / stuttering problems.
What's even funnier about these older cards is they end up being even more power-efficient than the current ones, despite lacking the smart power-saving "gimmicks" because they run the VRAM at 100 MHz when idle instead of 300 MHz.
FYI, Current NVIDIA GPUs have a VLIW5 architecture, just like last-gen (TeraScale2) AMD GPUs, which is extremely efficient.
I recommend getting a *last-generation* AMD GPU with the TeraScale2 architecture (HD5000 or HD6000 Series).
The latest AMD GPUs with a GCN architecture (HD7000,HD8000,Rx and Fury Series) are still horribly unoptimized (the drivers have unusually high CPU overhead).
But last-gen GPUs are so optimized they perform even better than NVIDIA's GPUs, especially when using the Direct3D backend (e.g. 2100fps [HD6000 Series] vs. 1100fps [R9 Series] in text/logo screens at native res. That's a ~2x difference in driver overhead.)
Another advantage of getting an older generation AMD GPU is that it doesn't have any of those "smart" power-saving features which introduce a lot of performance-related / stuttering problems.
What's even funnier about these older cards is they end up being even more power-efficient than the current ones, despite lacking the smart power-saving "gimmicks" because they run the VRAM at 100 MHz when idle instead of 300 MHz.
FYI, Current NVIDIA GPUs have a VLIW5 architecture, just like last-gen (TeraScale2) AMD GPUs, which is extremely efficient.