caml-list - the Caml user's mailing list
 help / color / mirror / Atom feed
From: Goswin von Brederlow <goswin-v-b@web.de>
To: Jon Harrop <jon@ffconsultancy.com>
Cc: 'Yotam Barnoy' <yotambarnoy@gmail.com>,
	'Ocaml Mailing List' <caml-list@inria.fr>
Subject: Re: [Caml-list] How much optimized is the 'a option type ?
Date: Mon, 27 Jan 2014 11:03:55 +0100	[thread overview]
Message-ID: <20140127100355.GD24902@frosties> (raw)
In-Reply-To: <026101cf18dd$756c13d0$60443b70$@ffconsultancy.com>

On Fri, Jan 24, 2014 at 08:22:43AM -0000, Jon Harrop wrote:
> Yotam wrote:
> > Just to clarify, by value types you mean stuff allocated on the stack,
> right?
> 
> A value type is just an unboxed type like a C struct. A local variable of a
> value type is stored on the stack. A field inside a heap-allocated object
> that is a value type is stored inside the object and not in a separate
> object. An array of value types is a single contiguous block of memory with
> no pointers (e.g. an array of complex numbers).
> 
> With value types you can almost completely avoid the garbage collector by
> replacing pointers with indices into an array of value types that acts as a
> pool allocator. This can be useful for avoiding GC latency and for
> optimizing for collections that violate the generational hypothesis (e.g.
> long-lived Sets and Maps).
> 
> In HLVM, tuples were value types. For example, an (int * float) array is
> represented as a single contiguous block of memory. In constrast, OCaml
> represents this is an array of pointers to pairs where the second element is
> another pointer to a boxed float.
> 
> Value types and reified generics permit very efficient hash tables. For
> example, the Dictionary<int, float> type on .NET does not contain any
> pointers.
> 
> Cheers,
> Jon.

Problem is that a value is a fixed size and fits in a register. A
tuple does not. (int * float) even takes the space of 3 values on
32bit. You can unbox that in the optimizer for local use but in memory
and in function calls you need to pass this as box. Otherwise
polymorphic functions break.

Putting larger structures into an array without boxing also only works
for immutable objects and by copying them every time they are passed
around. You can't copy a mutable and you can't pass a pointer to the
middle of an array to another function or the GC might freak out.

So you have to be verry carefull where you use those unboxed arrays.
They only works within a single module like a Hashtbl.

MfG
	Goswin

  parent reply	other threads:[~2014-01-27 10:03 UTC|newest]

Thread overview: 53+ messages / expand[flat|nested]  mbox.gz  Atom feed  top
2014-01-17  7:35 Damien Guichard
2014-01-17  7:55 ` David House
2014-01-17  8:16   ` Julien Blond
2014-01-17  8:40     ` David House
2014-01-17  9:10       ` Gabriel Scherer
2014-01-17  9:22         ` Simon Cruanes
2014-01-17 17:57           ` Gerd Stolpmann
2014-01-18  1:35             ` Jon Harrop
2014-01-19  6:19               ` oleg
2014-01-21  1:51                 ` Francois Berenger
2014-01-18  1:01         ` Jon Harrop
2014-01-24 10:06         ` Alain Frisch
2014-01-24 10:16           ` Alain Frisch
2014-01-24 13:32             ` Yaron Minsky
     [not found]       ` <CAK=fH+jfi=GsMYBZzmuo=V5UAWimyxiiamY2+DkLg6F0i8XHGw@mail.gmail.com>
2014-01-17  9:11         ` David House
2014-01-17 11:23           ` Jonathan Kimmitt
2014-01-17 13:46             ` Nicolas Braud-Santoni
2014-01-17 13:56               ` Frédéric Bour
2014-01-17 14:02               ` Yaron Minsky
2014-01-17 14:09                 ` Simon Cruanes
2014-01-17 22:52                   ` Yaron Minsky
2014-01-18  1:37                   ` Jon Harrop
2014-01-17 14:24                 ` Gabriel Scherer
2014-01-17 22:29                   ` Yaron Minsky
2014-01-18  1:27                 ` Jon Harrop
2014-01-18  1:18             ` Jon Harrop
2014-01-20 10:16             ` Goswin von Brederlow
2014-01-20 11:23               ` Jonathan Kimmitt
2014-01-21  2:05                 ` Francois Berenger
2014-01-22 21:22                   ` Jon Harrop
2014-01-22 21:26               ` Jon Harrop
2014-01-23  9:29                 ` Goswin von Brederlow
2014-01-23 23:20                   ` Jon Harrop
2014-01-23 23:28                     ` Yotam Barnoy
2014-01-24  8:22                       ` Jon Harrop
2014-01-24  8:34                         ` Andreas Rossberg
2014-01-24 16:56                           ` Jon Harrop
2014-01-27 15:29                             ` Goswin von Brederlow
2014-01-27 16:18                               ` Yotam Barnoy
2014-01-29  7:56                                 ` Goswin von Brederlow
2014-01-29  8:32                                 ` Jon Harrop
2014-01-29 16:11                                   ` Yotam Barnoy
2014-01-30 18:43                                     ` Yotam Barnoy
2014-02-01 15:58                                       ` Goswin von Brederlow
2014-01-30 21:31                                     ` Jon Harrop
2014-01-30 21:43                                       ` Yotam Barnoy
2014-01-31  8:26                                         ` Jon Harrop
2014-02-01 15:40                                 ` Goswin von Brederlow
2014-01-27 10:03                         ` Goswin von Brederlow [this message]
2014-01-17 14:36 ` Markus Mottl
2014-01-17 15:49   ` Yotam Barnoy
2014-01-17 16:22     ` Markus Mottl
2014-01-20 10:09   ` Goswin von Brederlow

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=20140127100355.GD24902@frosties \
    --to=goswin-v-b@web.de \
    --cc=caml-list@inria.fr \
    --cc=jon@ffconsultancy.com \
    --cc=yotambarnoy@gmail.com \
    /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).