Dolphin, the GameCube and Wii emulator - Forums

Full Version: Unfortunately, the r5991 made a slower performance than r5985,,
You're currently viewing a stripped down version of our content. View the full version with proper formatting.
Pages: 1 2
Anyone else have "unable to stretch data to buffer" error on all the newer builds for SMG2? Only affects DX9 plugin.
The original post seems to be the result of a bit of bad luck compatibility-wise - I'd stick with the older version until a newer version "fixes" the problem (could just be a unique incompatibility or chance). If that doesn't work, then there may be a slightly more troublesome underlying issue requiring the good ol' random button press fixing method Wink
Yeah, but as I mentioned it's unlikely the developers will look into it unless there is more widespread reports. Right now it's limited to a select individual or two which suggests a misconfiguration or some other task using the cpu while trying to use Dolphin. Make sure you don't have EFB to Ram enabled except for games that need it, despite the option being faster since rodolfoosvaldobogado's hybrid EFB to Ram commits, it's still not as fast as EFB to Texture. Ensure your EFB scale settings are set for performance and Anti-Aliasing isn't enabled or is set appropriately. Generally nothing more than 2x EFB scale and 4x SSAA should be used on your hardware for most games, and even that has a tendency to bring your hardware to it's knees or cause the framerate drops you are experiencing. I recommend EFB to Texture, EFB to Scale 1x and Anti-Aliasing Disabled for best performance. If you want games to look more pretty since 1x EFB scale looks ugly, enable 4x SSAA with EFB scale set to 1x for a balance between performance and quality.
Whatever it was that was caused a slight frame drop in r5991 is gone, it's back to speed with r5994.
Maybe that's just how the build compiled over at dolphin-emu.org
(07-29-2010, 11:59 AM)RupeeClock Wrote: [ -> ]Background processes? I somewhat doubt that because I tested r5986, and immediately tested r5991 afterwards.

It doesn't matter if they've been run directly after each other, during dx11 development performance differed up to 25% in these cases even if I changed just one trivial line.
So, let me put emphasize on this: Test your builds multiple times before claiming any decrease/increase in FPS, thanks.
Weird, I noticed 1% speed increase so far...
Pages: 1 2