Re: Value types (Was: [Caml-list] ocamlopt LLVM support)
On 12/12/2010 08:09 PM, Benedikt Meurer wrote: The boxing involved is relevant, but boxing in general is not the issue. In this special case, the let nlen, n = if... code requires heap allocation, because of the way the pattern is compiled. This could be fixed by moving the condition out of the code and using two if's to select n/nlen separately (doesn't speed up that much). Fixing the pattern compiler to handle these cases might be interesting for general benefit. Instead of duplicating the conditional, you could also push the assignments to bound variables down the expression. For instance: let (x, y) = if b then (u, v) else (v, u) in ... can be replaced, conceptually, by: let x = dummy in let y = dummy in if b then (x - u; y - v) else (x - v; y - u); ... and similarly when the bound expression is a pattern matching. I've played with this a few months ago and could observe important speedups (27%, 20%) on two micro-benchmarks. The diff is really small: http://caml.inria.fr/cgi-bin/viewcvs.cgi/ocaml/branches/inplace_let/bytecomp/matching.ml?rev=10475sortby=dater2=10475r1=10474 -- Alain Micro benchmark 1: let () = for k = 1 to 1 do for i = 1 to 10 do let (x, y) = if i mod 2 = 0 then (1, i * 2) else (2, i * 3) in r := !r * x + y done done Micro benchmark 2: let f x y z = let a, b = match z with | Some (u, v) - u, v * 2 | None - 10, 20 in a * x + b * y let () = let r = ref 0 in for k = 1 to 2000 do for i = 1 to 10 do r := !r + f k i (Some (k, i)); r := !r + f k i None done done ___ Caml-list mailing list. Subscription management: http://yquem.inria.fr/cgi-bin/mailman/listinfo/caml-list Archives: http://caml.inria.fr Beginner's list: http://groups.yahoo.com/group/ocaml_beginners Bug reports: http://caml.inria.fr/bin/caml-bugs
Re: [Caml-list] Re: optimize div to right shift (NOT!)
On Mon, 13 Dec 2010 12:33:33 + Jonathan Kimmitt jonat...@kimmitt.co.uk wrote: A C compiler would optimize this to a right shift. Changing that to 'Int64.shift_right n 1' speeds up the code. Sorry to be a pedant, but this is not correct. The optimisation is only possible when the arguments are unsigned integers That particular program never used negative integers. which I don't think is specifiable when working in OCAML You are right, there is no way to tell ocaml that. # Int64.shift_right (-2L) 1;; - : int64 = -1L (So far, so good) # Int64.div (-1L) 2L;; - : int64 = 0L (Good) # Int64.shift_right (-1L) 1;; - : int64 = -1L (Duh) It is still possible to avoid the division, gcc generates this: movq%rdi, %rax shrq$63, %rax addq%rdi, %rax sarq%rax Or a better example with division by 8: leaq7(%rdi), %rax testq %rdi, %rdi cmovns %rdi, %rax sarq$3, %rax And division by non-power of two integers can optimized by replacing it with multiplication with its inverse (which gcc and llvm don't do for signed divisions, only for unsigned ones): http://www.hackersdelight.org/HDcode/magic.c.txt http://www.hackersdelight.org/HDcode/magicu.c.txt Best regards, --Edwin ___ Caml-list mailing list. Subscription management: http://yquem.inria.fr/cgi-bin/mailman/listinfo/caml-list Archives: http://caml.inria.fr Beginner's list: http://groups.yahoo.com/group/ocaml_beginners Bug reports: http://caml.inria.fr/bin/caml-bugs
[Caml-list] [ANN] Js_of_ocaml version 1.0
Hi, I'm happy to announce the first official release of Js_of_ocaml, a compiler from OCaml bytecode to Javascript. This tool let you write OCaml programs that run on Web browsers. Js_of_ocaml is easy to install, and use thereafter, as it works with an existing installation of OCaml, with no need to recompile any library. It comes with bindings for a large part of the browser APIs. The project page is: http://ocsigen.org/js_of_ocaml/ EXAMPLES The compiler has been used to implement some noteworthy examples, such as: - an interactive 3D view of the Earth http://ocsigen.org/js_of_ocaml/planet - a graph viewer http://ocsigen.org/js_of_ocaml/graph PERFORMANCES According to our benchmarks, with state of the art Javascript engines, the generated programs runs typically faster than with the OCaml bytecode interpreter ( http://ocsigen.org/js_of_ocaml/performances ). Js_of_ocaml performs dead code elimination in order to generate compact code: the Javascript file produced is usually smaller than the input bytecode file, and often much smaller. LINKS Project home page http://ocsigen.org/js_of_ocaml/ Download http://ocsigen.org/download/js_of_ocaml-1.0.tar.gz Get source codedarcs get http://ocsigen.org/darcs/js_of_ocaml/ Documentation https://ocsigen.org/js_of_ocaml/lib/overview FURTHER TECHNICAL DETAILS Js_of_ocaml performs a fairly faithful translation. The order of evaluation is preserved. Modular arithmetic is used for integers (but with 32 bit integer). It does not support tail calls (function calls in tail position), as this would be too expansive. However tail recursion (self call in tail position) is properly optimized. Explicit coercion functions can be used to convert Ocaml values to Javascript values, and conversely (for instance, to map OCaml mutable strings to Javascript immutable UTF-16 strings, or to map OCaml booleans to Javascript booleans). A Camlp4 syntax extension makes it possible to invoke Javascript methods in a type safe way. COMPARISON TO OCAMLJS Ocamljs is a compiler from OCaml source code to Javascript. Jake Donham has written a fair comparison of the two tools: http://ambassadortothecomputers.blogspot.com/2010/08/ocamljs-03.html Ocamljs is a back-end to the existing OCaml compiler. Thus, contrary to Js_of_ocaml, you need to perform a distinct installation of OCaml to use Ocamljs, and you have to recompile all the libraries you may need. Ocamljs follows a different philosophy: it attempts to merge OCaml datatypes with the corresponding Javascript datatypes. For instance, OCaml objects are implemented as Javascript objects. Conversely, Javascript objects are given an OCaml object type. A mixed representation of strings is used: mutable OCaml-style strings and immutable Javascript strings both have the same type. All this is good for interoperability, but can be a source of incompatibilities and can result in runtime errors not caught by the type checker. Ocamljs optimizes tail recursion, but this comes at a large performance cost. -- Jerome Vouillon ___ Caml-list mailing list. Subscription management: http://yquem.inria.fr/cgi-bin/mailman/listinfo/caml-list Archives: http://caml.inria.fr Beginner's list: http://groups.yahoo.com/group/ocaml_beginners Bug reports: http://caml.inria.fr/bin/caml-bugs
Re: [Caml-list] [ANN] Js_of_ocaml version 1.0
Jerome, Thank you, this sounds fantastic! One small question: could you expand on your last comment: On Dec 13, 2010, at 8:06 AM, Jerome Vouillon wrote: snip Ocamljs optimizes tail recursion, but this comes at a large performance cost. Do you mean all tail-calls come a large cost, or only those outside of plain tail-recursion? Either way, could you give us some more intuition as to why this happens, and why js_of_ocaml doesn't suffer from the same problem (assuming it applies to tail-recursion)? Thanks, Yitzhak - Yitzhak Mandelbaum ___ Caml-list mailing list. Subscription management: http://yquem.inria.fr/cgi-bin/mailman/listinfo/caml-list Archives: http://caml.inria.fr Beginner's list: http://groups.yahoo.com/group/ocaml_beginners Bug reports: http://caml.inria.fr/bin/caml-bugs
[Caml-list] [ANN] Lwt 2.2.0 released
Hello,
The Lwt team is pleased to announce the release of Lwt 2.2.0. You can
download it at:
http://ocsigen.org/lwt/install
Here is a list of changes from the previous version (2.1.1):
* Bugfixes:
** Fix a bug with cancellable threads causing {{{Canceled}}}
exceptions to be raised randomly
** Fix a fd-leak in Lwt_io.open_connection
* {{{Lwt_unix}}} now use libev instead of select
* Add thread local storage support to {{{Lwt}}}
* Add backtrace support to {{{Lwt}}}. Now {{{Lwt}}} exceptions can
be recored by using the syntax extension with the {{{-lwt-debug}}}
command line switch.
* Allow blocking system calls to be executed in parallels
* Change the type of many functions of {{{Lwt_unix}}}, which now
return a {{{Lwt}}} thread
* Add functions {{{Lwt_unix.readable}}} and {{{Lwt_unix.writable}}}
* Add function {{{Lwt_io.is_busy}}}
* Add functions {{{Lwt_event.delay}}} and {{{Lwt_signal.delay}}}
* Add function {{{Lwt_term.render_update}}}
* Add function {{{Lwt_ssl.embed_socket}}}
* Add module {{{Lwt_bytes}}} defining operations on bigarrays
instead of strings
* Use bigarrays in Lwt_io instead of strings for the internal buffer.
Lwt_io.make now takes a function that uses a bigarray.
* Add module {{{Lwt_switch}}}
Enjoy!
--
Jérémie Dimino, on behalf of the Lwt team
___
Caml-list mailing list. Subscription management:
http://yquem.inria.fr/cgi-bin/mailman/listinfo/caml-list
Archives: http://caml.inria.fr
Beginner's list: http://groups.yahoo.com/group/ocaml_beginners
Bug reports: http://caml.inria.fr/bin/caml-bugs
Re: [Caml-list] [ANN] Js_of_ocaml version 1.0
On Mon, Dec 13, 2010 at 09:58:27AM -0500, Yitzhak Mandelbaum wrote: One small question: could you expand on your last comment: On Dec 13, 2010, at 8:06 AM, Jerome Vouillon wrote: snip Ocamljs optimizes tail recursion, but this comes at a large performance cost. Do you mean all tail-calls come a large cost, or only those outside of plain tail-recursion? Either way, could you give us some more intuition as to why this happens, and why js_of_ocaml doesn't suffer from the same problem (assuming it applies to tail-recursion)? I meant tail calls, indeed. Tail recursion (when a function calls itself recursively in tail position) can be optimized efficiently by wrapping the function body inside a loop. This is what Js_of_ocaml does. For optimizing tail calls in general, you need to use trampolines. Instead of calling a function in tail position, you return this function and its arguments. Then, a piece of code called a trampoline takes care of invoking repeately the functions it receives this way until a final result is returned. This is expansive. You have to make sure that this piece of code is there whenever a function is invoked in tail position. Also, the JIT compilers cannot optimize the trampoline, as the functions it will have to call, and their number of arguments, are unknown. -- Jerome ___ Caml-list mailing list. Subscription management: http://yquem.inria.fr/cgi-bin/mailman/listinfo/caml-list Archives: http://caml.inria.fr Beginner's list: http://groups.yahoo.com/group/ocaml_beginners Bug reports: http://caml.inria.fr/bin/caml-bugs
[Caml-list] [ANN] OBus 1.1
Hi, I'm happy to announce the release 1.1 of OBus, a pure OCaml implementation of the D-Bus protocol. OBus aims to make it easy to use and provide D-Bus services in OCaml. It can generate interfaces to D-Bus services from introspection files, it provides integration of D-Bus methods, signals and properties to native ocaml functions, mapping between D-Bus types and OCaml types, ... But it is also possible to write low-level D-Bus application using OBus. OBus is distributed with predefined OCaml interfaces to the following services: Hal, NetworkManager, Popup notifications, PolicyKit, UDisks and UPower. Links: Archive: https://forge.ocamlcore.org/frs/download.php/539/obus-1.1.tar.gz Project page: https://forge.ocamlcore.org/projects/obus/ Manual: https://forge.ocamlcore.org/docman/view.php/26/127/manual.pdf Api documentation: http://obus.forge.ocamlcore.org/api/ Enjoy! -- Jérémie ___ Caml-list mailing list. Subscription management: http://yquem.inria.fr/cgi-bin/mailman/listinfo/caml-list Archives: http://caml.inria.fr Beginner's list: http://groups.yahoo.com/group/ocaml_beginners Bug reports: http://caml.inria.fr/bin/caml-bugs
RE: [Caml-list] Re: optimize div to right shift (NOT!)
Edwin wrote: http://www.hackersdelight.org/HDcode/magic.c.txt http://www.hackersdelight.org/HDcode/magicu.c.txt Nice! :-) Cheers, Jon. ___ Caml-list mailing list. Subscription management: http://yquem.inria.fr/cgi-bin/mailman/listinfo/caml-list Archives: http://caml.inria.fr Beginner's list: http://groups.yahoo.com/group/ocaml_beginners Bug reports: http://caml.inria.fr/bin/caml-bugs
