Basic benchmark #69
Basic benchmark #69
Conversation
|
Merging to
|
|
I don't think this is what #67 proposes, therefore it does not have my approval. Also, this is just a hardcopy of the krypta example with the diagnostics enabled, I think it would be better to add a "benchmark" or "diagnostics" feature and conditionally compile the FrameTimeDiagnostics Plugin for all examples |
|
This is not a hardcopy of Krypta with diagnostics enabled. I did quite a bit of code reuse to save time in developing this benchmark. It does not have moving lights or occluders, but I'm not sure that would have any utility in a benchmark for this library; #67 was mostly spitting out ideas. The benchmark procedurally generates a 100x100 map filled with randomly placed occluding walls and light-emitting candles. Some candles jitter and some do not. I was partway through working on a system to delete and add candles every few seconds, but I believe that deserves its own benchmark as a test of how fast occluders and lights are created and destroyed, if that is even necessary. This benchmark provides tangible utility. The #65 regression would have been identified if this benchmark had been ran with the changes, as the map that is generated is large enough as to have many lights and occluders. In my own tests on the regression build, I get around 30 FPS, while I get significantly more, closer to 165, using the fixed build. Moreover, improvements like JFA can easily be tested for performance with this benchmark. If you think there should be changes, what would you like them to be @tigerplush? |
|
In regards to adding a benchmark or diagnostics feature to the examples, I think this is rather pointless. There are not enough lights, and the maps are too small in the other examples to provide any real value or performance difficulties. Krypta and the other two always run at 165 FPS (max monitor refresh rate for me) even with the #65 regression (which didn't really impact those, as nearly if not all lights are always in view) |
|
Will try to find time to review this week. |
Bump |
Includes changes from #66
resolves #67
I get the following warning. It seems I didn't quite setup the cameras properly or something? If someone knows what's up please let me know or fix it.