Nice review but I see a few issues with these tests:
- You don't switch only the graphics stack, but the whole OS. Sure I don't believe that GCC4 to GCC5 could give the kind of number variation we see, but something here, something there... it adds up (or removes?). At the end it's not so fair the graphic stack.
- You don't compare the rendering between the various platforms tested. I mean if something is faster because it renders wrongly that's not great. On the other hand, if the rendering is better on top of being faster, well that's awesome (more awesome than just knowing it it faster), but we don't get that information from these graphs..
- You don't switch only the graphics stack, but the whole OS. Sure I don't believe that GCC4 to GCC5 could give the kind of number variation we see, but something here, something there... it adds up (or removes?). At the end it's not so fair the graphic stack.
- You don't compare the rendering between the various platforms tested. I mean if something is faster because it renders wrongly that's not great. On the other hand, if the rendering is better on top of being faster, well that's awesome (more awesome than just knowing it it faster), but we don't get that information from these graphs..
Comment