Comment by gwd

> The baseline runtime overhead of cgo calls has been reduced by ~30%.

This is great, but tell me more? What happened?