attractivechaos / plb Goto Github PK

Preferably, hardcode them in the tests themselves.

I suspect that instead of 999.99 in the "patmch:2t" Lua results there should be a "—", since Lua regexps do not have "|" operator.

NB: Since Lua does not have built-in PCRE-compatible regexps, consider using a library (like this one http://rrthomas.github.com/lrexlib/), and/or LPEG (http://www.inf.puc-rio.br/~roberto/lpeg/)

Hot off the presses! http://morepypy.blogspot.com/2011/04/pypy-15-released-catching-up.html , this should speed pypy up a good bit, should just work with the existing code.

How about running the benchmarks again? Since last test many runtimes improved their perfomance. New results can be interesting.

Can you add Rust next to the current supported languages for matrix multiplication and the other examples?

thanks

Need this for my slide deck :-)

If I can help, please don't hesitate to tell me.

You should provide a license for each line of code that you share, otherwise it will not be useful to anyone else but you.

My personal favorite is MIT or BSD.

Make sure that you get an OK for a license of your choice for every contributor.

See #15 for an example. This will help plain Lua benchmarks (and in some cases to LuaJIT ones).

Hint: do this to get the list of globals used by a Lua script:

luac -l -o /dev/null filename.lua | grep ETGLOBAL

I'll submit a pull request (for most of benchmarks at least) if I'll have #14 available.

Under sudoku/incoming you reference several datasets and give links in the Readme.md. Two of those links have died already. If you have the datasets offline, could you add them to the repository for posterity? Prefererably all 4 as the remaining links may also die.

Hi Attractive Chaos,

Thanks for your detailed benchmark - it's very impressive.

Given PHP is arguably one of the most popular languages in this web era, do you have plan to add PHP implementation in?

Thanks!

Jacky

Something along these lines:

-- These aliases would help plain Lua.
local io_lines = io.lines
local print = print -- Consider removing IO from all benchmarks, BTW.
local collectgarbage = collectgarbage

local pattern = (...)

local counter = 0
for l in io_lines() do
  if l:find(pattern) then print(l) end

  counter = counter + 1
  if counter % 1000 == 0 then -- Play with this value
    collectgarbage("step")
    counter = 0
  end   
end

I'll submit a pull request if you'd do #14.

Would be cool to get a report generator as well — if you don't update it by hand.