Benchmarking OpenGL/Vulkan/etc. code is easy because the GPU basically does what you tell it to, without a lot of magic caches and whatnot in between. Benchmarking vector graphics APIs, on the other hand…
-
Show this thread
-
Most vector graphics APIs cache everything behind your back. It’s hard to figure out when, say, Skia actually decides to do something interesting, as opposed to just blitting from some texture cache somewhere.
1 reply 1 retweet 3 likesShow this thread -
I’m tempted to *not* do this for Pathfinder, and instead to make caching explicit in the low-level API. Rationale: the user knows better than I do whether, say, a path is going to be blitted repeatedly or only once. High-level APIs like <canvas> can have magic heuristics though.
3 replies 1 retweet 25 likesShow this thread -
Replying to @pcwalton
Sounds like static vs dynamic optimization trade off. User might know some frequency info, if the content is first-party, but won’t know the caching costs/gains (or track them well over time). I’d want a way to say “help me as you see fit” as well as explicit control, I think.
1 reply 0 retweets 1 like
Yeah, that’s what the high-level APIs would be for.
Loading seems to be taking a while.
Twitter may be over capacity or experiencing a momentary hiccup. Try again or visit Twitter Status for more information.