about -fno-crossjumping -fno-gcse

[esrrhs]

I test the two option, and the benchmark result is nearly same. here is the code

function fibonacci(n)
    if n<3 then
        return 1
    else
        return fibonacci(n-1) + fibonacci(n-2)
    end
end

print(fibonacci(39))

[edubart]

This example is too simple for benchmark what -fno-crossjumping -fno-gcse can do, you can read more about this flag and Lua interpreter optimization in general in this thread: http://lua-users.org/lists/lua-l/2020-08/msg00153.html

At some point Dibyendu Majumdar told about using -fno-crossjumping -fno-gcse in the Lua VM to speedup and I got 4% speedup of the Nelua compiler, which is much complex than that. http://lua-users.org/lists/lua-l/2020-08/msg00165.html