Re: Computed goto optimization of vanilla Lua

[Roberto Ierusalimschy <roberto@...>]

> Some numbers of a "real world" test case:
> 
> [...]
> 
> Conclusion, "one computed goto -O3 -fno-gcse -fno-crossjumping" may
> represent a real improvement.

Many thanks. But I don't think it is fair to change 'one' and '-O3', as
they are indpendent of what we are looking here (jump tables).  In a
comparison between "one -O3" and "one -O3 -fno-gcse -fno-crossjumping
-DJUMPTAB", I got seedpus in the order of 3%.

(BTW, -fno-crossjumping alone seems to be enough to avoid collapsing
all indirect jumps into one.)

-- Roberto