One possible way to speed this up for UInt8Array is to take its underlying ArrayBuffer and wrap it into UInt32Array(), do xors through it, and then return the result as a UInt8Array again. https://stackoverflow.com/a/68005553/759134 illustrates the idea. However, we do have to be careful about cases when the number of uin8 elements doesn't divide by 4.

LMK if you want to give it a try, otherwise I am okay merging this as-is.

One possible way to speed this up for UInt8Array is to take its underlying ArrayBuffer and wrap it into UInt32Array(), do xors through it, and then return the result as a UInt8Array again. https://stackoverflow.com/a/68005553/759134 illustrates the idea. However, we do have to be careful about cases when the number of uin8 elements doesn't divide by 4.

LMK if you want to give it a try, otherwise I am okay merging this as-is.

That's a really great idea. I didn't know that JS could do that kind of cast. That would be better and the existing code in the natives shows how to deal with any dangling bytes (i.e. if len%4 is not zero). I'll see what I can do with JS to make that work, thanks.

Got a version using UInt32Array. I tested it with playground (link to test) and using UInt32Arrays is definitely faster (estimate about 3 to 4 times faster). I'll push up one soon, apparently the offset for UInt32Arrays has to be multiples of 4 so I'll have to do a little more work on it