VertexLoader: cache NativeVertexFormat #809
Conversation
| static NativeVertexLoaderMap s_native_vertex_map; | ||
| // TODO - change into array of pointers. Keep a map of all seen so far. | ||
|
|
||
| void Init() | ||
| { | ||
| MarkAllDirty(); | ||
| for (VertexLoader*& vertexLoader : g_VertexLoaders) | ||
| vertexLoader = nullptr; | ||
| for (auto& vertexLoader : s_VertexLoaders) |
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
This comment was marked as off-topic.
This comment was marked as off-topic.
Sorry, something went wrong.
|
LGTM. @JMC47 any noticeable performance improvement here? |
|
Increases Super Mario Galaxy's intro from 167 fps to 224 fps on my computer. Umm. If this looks good to people, it should probably be merged. 20 FPS increase in both SMG1 and SMG2 at pretty much all times. But the weird thing is that when I tested the build that this was supposed to fix a performance regression in, I couldn't find a performance regression. As such it's just... faster. I don't know, but, if it fixes a performance regression, that's good enough for me. |
This fix a performance regression of PR dolphin-emu#672.
|
Thanks, sorry again for the regression. Didn't notice the additional map lookup. |
VertexLoader: cache NativeVertexFormat
This fix a performance regression of PR #672.