You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
Currently, the coeffs.bin file is baked into the the wasm bundle with include_bytes macro, but this is not ideal, because it makes the bundle about 6 times bigger. Also the file never changes, so while we could in theory let browser and CDN cache it forever, we invalidate it on every build, whenever we change something in wasm code.
We can instead load the file separately and then copy it into wasm memory. This could make webzebra load slower because we would have to copy the file into wasm memory, but it could also make it faster, because we can download this file in parallel with loading+parsing wasm and wasm parser will have less stuff to parse. If the file is cached in the browser, this could be even faster because we wouldn't have to load it at all. I'd like to see numbers on this method.
The text was updated successfully, but these errors were encountered:
Currently, the coeffs.bin file is baked into the the wasm bundle with
include_bytes
macro, but this is not ideal, because it makes the bundle about 6 times bigger. Also the file never changes, so while we could in theory let browser and CDN cache it forever, we invalidate it on every build, whenever we change something in wasm code.We can instead load the file separately and then copy it into wasm memory. This could make webzebra load slower because we would have to copy the file into wasm memory, but it could also make it faster, because we can download this file in parallel with loading+parsing wasm and wasm parser will have less stuff to parse. If the file is cached in the browser, this could be even faster because we wouldn't have to load it at all. I'd like to see numbers on this method.
The text was updated successfully, but these errors were encountered: