Bold Pixel Engine Blit Performance Test

Hi all, Vlad here!

Wrote a quick performance test yesterday comparing movieclips with our own blit engine. Usually tests are performed with a number of objects. To test several loads, usually a different number of objects is used to test the results. I didn’t want that since we all know that many games don’t have a fixed ammount of objects rendered. Even if object pooling is used, what concerns me is the objects rendered.

What I did was to create one object per frame and then move all the existing objects away from the starting point. This would give an instant visual reference about how well blit performs against movieclips.

Give it a try and then read on.

Tried it? Cool!

Movement is time-based in both tests, this means that test 1, ran with movie clips took longer since the spread is bigger. This indicates a frame drop, but if you run it again and notice that there are different spreads, so not only it drops but it drops in a rather inconsistent way. Obviously the test also took longer to conclude.

On the other hand, blit engine performs rather smoothly, to the point where frame drop is not inconsistent and allows the objects to stay pretty close to each other. Above that, the test is damn quick compared to the first one, proved by the small spread (although I could’ve included a clock, I admit).

This is nothing we didn’t already know, I agree! The true testing will be with animation, rotation and scaling of blit entities.