Calling multiple JS functions consecutively from CPP, does using official encapsulation methods lead to performance waste?

[sxstd001]

Is the performance of A superior to that of B?
Is that: A involves only one context switch, whereas B requires multiple consecutive context switches?

A:

	EM_ASM({
		gl.useProgram($0);  
		gl.uniformMatrix3fv($1, false, $2)
                gl.xxxx...
	}, program, u_mvp, matrix);

B:

  emscripten::glUseProgram(program);
  emscripten::glUniformMatrix3fv(u_mvp, false, matrix);
  emscripten::gl_xxxx....

[sbc100]

You are asking here about the code of calling from Wasm to JavaScript. This tends to vary a little between browsers, but should be very fast. Especially when you are making calls into WebGL or WebGPU like this I imagine the cost of the graphics API call makes the cost of the Wasm/JS crossing insignificant. Obviously the only what to really know this is to measure of the browser(s) you are targetting. In general I would advise against this type of optimization since its likely to be premature, and also change over time as browsers improve. @kainino0x WDYT?