enable parallel codegen by default

[spernsteiner]

Enable parallel codegen (2 units) by default when --opt-level is 0 or 1. This gives a minor speedup on large crates (~10%), with only a tiny slowdown (~2%) for small ones (which usually build in under a second regardless). The current default (no parallelization) is used when the user requests optimization (--opt-level 2 or 3), and when the user has enabled LTO (which is incompatible with parallel codegen).

This commit also changes the rust build system to use parallel codegen when appropriate. This means codegen-units=4 for stage0 always, and also for stage1 and stage2 when configured with --release-channel=source. (Other release channel settings use codegen-units=1 for stage1 and stage2, to get maximum performance for release binaries.) The build system also sets codegen-units=1 for compiletest tests (compiletest does its own parallelization) and uses the same setting as stage2 for crate tests. The overall result (for release-channel=source) is 35% reduction in make all time, and 25% reduction in make check time.

r? @alexcrichton

[thestinger]

Please don't make builds from source perform more poorly than they already do. This makes the language look very bad. Developers can be expected to pass a bunch of flags, expecting that of users and packagers is wrong. Only explicit debug builds should be using parallel code generation.

[thestinger]

cc #17665, #17496

I'm strongly against doing this by default for builds of Rust though. Doing it for --opt-level=0 and --opt-level=1 is fine and I would r+ that myself.

[alexcrichton]

This looks great, thanks @epdtry! In light of recent discussions about the "default mode" (see the issues @thestinger linked too), we're thinking that release mode should likely be the default instead of debug mode.

Until we have those flags implemented, could we hold off on the -C codegen-units=4 options for stage1/stage2? Otherwise I think all other changes here are unambiguously improvements!

[aturon]

It'd be really great to make progress here. @epdtry can you incorporate @alexcrichton's suggestion so we can go ahead and move forward?

[pcwalton]

I'm taking this over.