Nathan Ringo (Dec 22 2020 at 00:24): Nathan Ringo (Dec 22 2020 at 00:24):Do I need to explicitly run optimization passes in some way? The generated amd64 assembly I'm getting is like, comically inefficient
bjorn3 (Dec 22 2020 at 20:55):You can set the opt_level option to speed_and_size when building the TargetIsa.
bjorn3 (Dec 22 2020 at 20:56):Also in what kind of way is it inefficient?
bjorn3 (Dec 22 2020 at 20:56):@Nathan Ringo ^
Nathan Ringo (Dec 22 2020 at 20:57):er, if I'm using SimpleJIT, where would I be building the TargetIsa? (I think it uses native or whatever)
There's a bunch of mov rdi, 0; mov rsi, rdi and there's no direct calls at all; only mov rcx, number; jmp rcx which by my understanding is significantly less efficient
Nathan Ringo (Dec 22 2020 at 21:02):like, for example, in https://github.com/bytecodealliance/simplejit-demo/blob/main/src/jit.rs
bjorn3 (Dec 22 2020 at 22:50):Those kinds of inefficiencies are coming from regalloc. Cranelift is currently switching to a new framework for backends. This includes a regalloc that shouldn't do this. For aarch64 it is the default (and only) backend. For x86_64 you have to enable it using the experimental_x64 feature flag.
bjorn3 (Dec 22 2020 at 22:51):@Nathan Ringo ^
bjorn3 (Dec 22 2020 at 22:53):The mov; jmp by the way is because is_pic is not enabled by default. x86_64 doesn't allow 64bit immediates for calls.
bjorn3 (Dec 22 2020 at 22:53):You can use a custom TargetIsa instead of the native one by using with_isa instead of new for the jit builder.
Nathan Ringo (Dec 22 2020 at 22:54):Okay, I'll look at those, thanks
Nathan Ringo (Dec 22 2020 at 23:25):hm, am I correct in thinking that experimental_x64 is only available in the Git version of Cranelift? It doesn't appear to exist in the released 0.68.0 despite being in the wasmtime repo...
bjorn3 (Dec 23 2020 at 06:53):That is possible.
Last updated: Dec 23 2024 at 13:07 UTC