caml-list - the Caml user's mailing list
 help / color / mirror / Atom feed
From: Benedikt Meurer <benedikt.meurer@googlemail.com>
To: caml-list@yquem.inria.fr
Subject: Re: [Caml-list] OCamlJIT2 vs. OCamlJIT
Date: Tue, 30 Nov 2010 11:55:38 +0100	[thread overview]
Message-ID: <D6C274D9-8A00-4D6F-936A-58206CA5D358@googlemail.com> (raw)
In-Reply-To: <20101130124803.7952fca1@deb0>


On Nov 30, 2010, at 11:48 , Török Edwin wrote:

>> In short: Performance measured on a P4 "Northwood" (no long mode,
>> plain x86) 2.4GHz. OCamlJIT2 beats OCamlJIT by a factor of 1.1 to 2.0
>> in every benchmark, and - rather surprising - was even able to beat
>> ocamlopt in the number crunching benchmark (probably an issue with
>> the x86 backend of ocamlopt).
> 
> Looks like this happens only on Pentium4: on Core2 and Xeon ocamlopt
> is still faster on almabench.unsafe, or did I miss something?

Core 2 and Xeon are x86-64, P4 is x86, so completely different architecture (with different code generator backends).

>> 
>> As mentioned by Xavier Leroy and others previously, we probably went
>> as far as we could go in the direction of JITting the byte-code
>> virtual machine, while preserving its general stack-based nature and
>> instruction set. Moving even further means translating the byte-code
>> to some intermediate form suitable for use with standard compilation
>> techniques; but as we saw earlier, in an LLVM-based prototype, the
>> compilation overhead increases dramatically and the benefit of JIT
>> compilation vanishes.
> 
> An LLVM-based backend would still be interesting for static
> compilation, where compile times don't matter much.
> Did you try comparing an LLVM-based backend with ocamlopt?
> If it is faster could some of the LLVM passes be ported to ocamlopt's
> backend?

LLVM backend for ocamlopt is a totally different story. You'd have to start with the Ulambda or the Cmm intermediate representation, while a JIT approach starts with the byte-code representation (not necessarily, but that was the case for our approach).

However, I'm not sure there would be any immediate benefit from using LLVM as code generator backend for ocamlopt, since ocamlopt already does a quite good (and fast) job. So besides probably educational or research interests, what would be the practical benefit of LLVM inside ocamlopt?

> Best regards,
> --Edwin

regards,
Benedikt

  reply	other threads:[~2010-11-30 10:55 UTC|newest]

Thread overview: 44+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2010-11-30  8:36 Benedikt Meurer
2010-11-30 10:48 ` [Caml-list] " Török Edwin
2010-11-30 10:55   ` Benedikt Meurer [this message]
2010-11-30 17:01     ` bluestorm
2010-11-30 17:26       ` Török Edwin
2010-11-30 18:27         ` Basile Starynkevitch
2010-11-30 17:29       ` Benedikt Meurer
2010-11-30 17:32       ` Yoann Padioleau
2010-11-30 22:06         ` Jon Harrop
2010-11-30 22:48           ` Benedikt Meurer
2010-12-01 14:11             ` Jon Harrop
2010-12-01 15:00               ` Benedikt Meurer
2010-12-01 22:03                 ` Jon Harrop
2010-12-02  1:17                   ` Eray Ozkural
2010-12-03 10:03                   ` ocamlopt LLVM support (Was: [Caml-list] OCamlJIT2 vs. OCamlJIT) Benedikt Meurer
2010-12-03 13:34                     ` Till Varoquaux
2010-12-03 13:41                       ` Eray Ozkural
2010-12-03 14:06                         ` Török Edwin
2010-12-03 21:16                       ` Jon Harrop
2010-12-05 16:44                       ` Benedikt Meurer
2010-12-03 14:32                     ` Philippe Strauss
2010-12-03 21:22                       ` Jon Harrop
2010-12-03 21:45                         ` Philippe Strauss
2010-12-03 15:32                     ` Michael Ekstrand
2010-12-03 21:34                       ` Jon Harrop
2010-12-03 20:07                     ` Jon Harrop
2010-12-05 16:37                       ` Benedikt Meurer
2010-12-05 16:57                         ` Török Edwin
2010-12-05 20:54                           ` Benedikt Meurer
2010-12-05 20:12                         ` Jon Harrop
2010-12-05 21:21                           ` Benedikt Meurer
2010-12-05 21:44                             ` Benedikt Meurer
2010-12-06 22:38                             ` Jon Harrop
2010-12-05 22:41                           ` Erik de Castro Lopo
2010-12-05 22:34                         ` Erik de Castro Lopo
2010-12-06  8:27                           ` Benedikt Meurer
2010-12-06  9:28                             ` David Rajchenbach-Teller
2010-12-06 11:08                         ` Richard Jones
2010-12-06 20:18                           ` Jon Harrop
2010-12-01  0:16           ` [Caml-list] OCamlJIT2 vs. OCamlJIT Erik de Castro Lopo
2010-12-01  1:34             ` Yoann Padioleau
2010-12-01 12:58             ` Jon Harrop
2010-12-01 13:55               ` ivan chollet
2010-11-30 21:19       ` Jon Harrop

Reply instructions:

You may reply publicly to this message via plain-text email
using any one of the following methods:

* Save the following mbox file, import it into your mail client,
  and reply-to-all from there: mbox

  Avoid top-posting and favor interleaved quoting:
  https://en.wikipedia.org/wiki/Posting_style#Interleaved_style

* Reply using the --to, --cc, and --in-reply-to
  switches of git-send-email(1):

  git send-email \
    --in-reply-to=D6C274D9-8A00-4D6F-936A-58206CA5D358@googlemail.com \
    --to=benedikt.meurer@googlemail.com \
    --cc=caml-list@yquem.inria.fr \
    /path/to/YOUR_REPLY

  https://kernel.org/pub/software/scm/git/docs/git-send-email.html

* If your mail client supports setting the In-Reply-To header
  via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox;
as well as URLs for NNTP newsgroup(s).