Re: Computed goto optimization of vanilla Lua

[Enrico Colombini <erix@...>]

On 05-Feb-16 13:19, Rodrigo Azevedo wrote:
> Conclusion, "one computed goto -O3 -fno-gcse -fno-crossjumping" may
> represent a real improvement.

I had bad surprises with -O3, such as added back and forth jumps that 
disrupted my critical code; I tend to avoid it.

That happened with a low-end microcontroller, so it may or may not apply 
here.

--
  Enrico