24 May
2006
24 May
'06
1:14 p.m.
Hello,
I am looking for a cross-platform implementation of an atomic
integer.
Under Linux, a build an c++ class "atomic" around asm/atomic.h,
(which I can use as if it where an int), but I'd like to have a
solution that also works on Windows XP and Mac OS X.
Thanks for any suggestions,
Maarten
24 May
24 May
1:42 p.m.
On 5/24/06, Maarten de Boer <mdeboer(a)iua.upf.edu> wrote:
Hello,
I am looking for a cross-platform implementation of an atomic
integer.
Under Linux, a build an c++ class "atomic" around asm/atomic.h,
(which I can use as if it where an int), but I'd like to have a
solution that also works on Windows XP and Mac OS X.
PortAudio developer Ross Bencina has started a new project to
address this need, including a set of lock-free algorithms based
on the atomic primitives.
http://www.audiomulch.com/~rossb/code/lockfree/liblfds/
I believe this will be very useful to many audio developers.
But, it is still in early stages of design. There are pointers to
existing implementations in this Wiki...
http://works.music.columbia.edu/LFDS
--
joq
4:18 p.m.
-----BEGIN PGP SIGNED MESSAGE-----
Hash: SHA1
Jack O'Quin wrote:
On 5/24/06, Maarten de Boer
<mdeboer(a)iua.upf.edu> wrote:
if you feel like having a dependency to the qt-core libs you might
consider
using their atomic low-level-operations.
so long...
nuekki
Hello,
I am looking for a cross-platform implementation of an atomic
integer.
Under Linux, a build an c++ class "atomic" around asm/atomic.h,
(which I can use as if it where an int), but I'd like to have a
solution that also works on Windows XP and Mac OS X.
PortAudio developer Ross Bencina has started a new project to
address this need, including a set of lock-free algorithms based on
the atomic primitives.
http://www.audiomulch.com/~rossb/code/lockfree/liblfds/
I believe this will be very useful to many audio developers. But,
it is still in early stages of design. There are pointers to
existing implementations in this Wiki...
http://works.music.columbia.edu/LFDS -----BEGIN PGP SIGNATURE-----
Version: GnuPG v1.4.3 (GNU/Linux)
Comment: Using GnuPG with Thunderbird - http://enigmail.mozdev.org
iD8DBQFEdKNr+k24EnBNzsMRAgfEAJ0fDPKoYJEplKTDNV6AJ63E3QjF3wCfTMQr
s7R5O9c7DJHg6GTDS7aZcV8=
=Mebj
-----END PGP SIGNATURE-----
10:44 p.m.
On Thu, 2006-05-25 at 02:24 +0200, Jens M Andreasen wrote:
Uhmm, I was reliying on you using a contemporary implementation. For a
philosophical discussion you might prefer to look elsewhere
--
mvh // Jens M Andreasen
>> I am
looking for a cross-platform implementation of an atomic
>> integer.
sizeof(int) is your friend
int speedy; as well, since int is defined to be the fastest integer on
your platform of choice.
10:58 p.m.
On Thu, 2006-05-25 at 02:44 +0200, Jens M Andreasen wrote:
On Thu, 2006-05-25 at 02:24 +0200, Jens M Andreasen
wrote:
Uhmm, I was reliying on you using a contemporary implementation. For a
philosophical discussion you might prefer to look elsewhere
not sure what you mean, but on sparcs, int writes are not atomic unless
you only use the lower 24 bits.
>>
I am looking for a cross-platform implementation of an atomic
>> integer.
sizeof(int) is your friend
int speedy; as well, since int is defined to be the fastest integer on
your platform of choice.
11:23 p.m.
On Wed, 2006-05-24 at 20:58 -0400, Paul Davis wrote:
--
On Thu, 2006-05-25 at 02:44 +0200, Jens M Andreasen
wrote:
Right, I tried to retract some of what I said that is only valid for the
cheap stuff available today
BTW: What sparc? I came in with 64bit datapaths ...
On Thu, 2006-05-25 at 02:24 +0200, Jens M
Andreasen wrote:
Uhmm, I was reliying on you using a contemporary implementation. For a
philosophical discussion you might prefer to look elsewhere
not sure what you mean, but on sparcs, int writes are not atomic unless
you only use the lower 24 bits.
>>
I am looking for a cross-platform implementation of an atomic
>> integer.
sizeof(int) is your friend
int speedy; as well, since int is defined to be the fastest integer on
your platform of choice.
25 May
25 May
7:33 a.m.
not sure what you mean, but on sparcs, int writes are
not atomic unless
you only use the lower 24 bits.
right, this is exactly the point! 32 bits int might be atomic on a certain
platform, but i have no idea about many others.
in the meantime, a part from joq suggestion of LFDS, and niklas mention of
qt-core libs, i also came accross th glib atomic operations
http://developer.gnome.org/doc/API/2.0/glib/glib-Atomic-Operations.html
and the MacOSX atomic functions as in CarbonCore/DriverSynchronization.h
maarten
3 p.m.
On Wed, 24 May 2006, Paul Davis wrote:
Does someone have a good reference on this? I think the writes just are
not atomic, but you can use some tricks [1] to implement atomic behaviour
by spinning until the operation succeeds.
[1] http://www.freepatentsonline.com/5666546.html?highlight=5434995
.. I'm not sure maybe this is just one variation...
But, reading the Linux 2.6.16.17 code for sparc (atomic.h and atomic.S),
only the writes are protected, not reads. So as long as you have only one
writer, you should be safe. Also, the 24bit limitation only applies if you
are using the low 8bits for spinning.
Anyways, a good (accurate and available online) reference for this would
be nice to have, to be 100% sure before making any architectural
decisions. For me, I'd be happy to just exclude these platforms, or use
mutexes as a crude workaround on the affected platforms (assuming there're
only few of them).
Cross-platform atomic integer operations would be nice, but you can do
quite a lot with just atomic read/writes, which most of the platforms
already provide for 'int's. For example, these are sufficient to implement
a simple one-writer-one-reader lock-free queue (which is necessary in most
audio apps).
OTOH, it is probably wise to explicitly mark all code where ints are
assumed to be atomic w.r.t. read/writes (or any other non-standard
properties). Especially if sparc-style atomic properties become more
widely used later on...
--
links, my public keys, etc at http://eca.cx/kv
>
sizeof(int) is your friend
not sure what you mean, but on sparcs, int writes are
not atomic unless
you only use the lower 24 bits. 
3:28 p.m.
On Thu, 2006-05-25 at 19:57 +0300, Kai Vehmanen wrote:
Does someone have a good reference on this? I think
the writes just
are not atomic, but you can use some tricks [1] to implement atomic
behaviour by spinning until the operation succeeds.
Do we still care about 32 bit sparc?
Lee
9:24 p.m.
New subject: [linux-audio-dev] Re: crossplatform atomics
On 5/26/06, Lee Revell <rlrevell(a)joe-job.com> wrote:
On Thu, 2006-05-25 at 19:57 +0300, Kai Vehmanen
wrote:
Doing audio on a 32 bit sparc.... mmm. Why not buy a spend 150 euro
and by a 2-3 ghz intel/amd machine new?
Does someone have a good reference on this? I
think the writes just
are not atomic, but you can use some tricks [1] to implement atomic
behaviour by spinning until the operation succeeds.
Do we still care about 32 bit sparc?
Lee 
26 May
26 May
5:43 a.m.
New subject: [linux-audio-dev] Re: crossplatform atomics
On Friday 26 May 2006 01:24, Loki Davison wrote:
On 5/26/06, Lee Revell <rlrevell(a)joe-job.com>
wrote:
I think the idea was that since one (old) architecture does not have atomic
ints there might be more.
Since I really could not care less about 32bit sparc, the more interesting
question would be; is it possible that non atomic ints will crop up in future
hardware designs?
To me it just sounds, really, really, really, improbable.
0.02 SEK
/Robert
On Thu, 2006-05-25 at 19:57 +0300, Kai Vehmanen
wrote:
Doing audio on a 32 bit sparc.... mmm. Why not buy a spend 150 euro
and by a 2-3 ghz intel/amd machine new? Does someone have a good reference on this? I
think the writes just
are not atomic, but you can use some tricks [1] to implement atomic
behaviour by spinning until the operation succeeds.
Do we still care about 32 bit sparc?
Lee
5:44 a.m.
On Thu, 25 May 2006, Lee Revell wrote:
On Thu, 2006-05-25 at 19:57 +0300, Kai Vehmanen
wrote:
Not necessarily much, but I'd like to understand this fully. There might
be other archs affected as well. Again looking at the Linux kernel code -
for instance armv6 atomic_set needs spinning (but not for armv4)... so
it's not just sparc32. There might not be _that_ many sparc32s out
there, but there are over 3 billion arm cores out there. ;)
But, but, even on these platforms, it's just the writes. It'd be
interesting to know if there are any cases where you could be reading
bogus values due to concurrent access to an 'int'.
--
links, my public keys, etc at http://eca.cx/kv
Does someone have a good reference on this? I
think the writes just
Do we still care about 32 bit sparc? 
13 Jul
13 Jul
2:32 p.m.
On Wed, 2006-05-24 at 17:13 +0200, Maarten de Boer wrote:
Hello,
I am looking for a cross-platform implementation of an atomic
integer.
Under Linux, a build an c++ class "atomic" around asm/atomic.h,
(which I can use as if it where an int), but I'd like to have a
solution that also works on Windows XP and Mac OS X.
Thanks for any suggestions,
Maarten
Reviving an ancient thread, but has anyone looked at this?
http://www.hpl.hp.com/research/linux/atomic_ops/
It's pretty robust by the looks of it, nicely autotoolized, portable (to
various compilers and architechtures).
You can even apt-get install lib-atomicops-dev (in sid at least)
Looks like exactly what I've been looking for...
-DR-
6653
days inactive
6703
days old
linux-audio-dev@lists.linuxaudio.org
13 comments
10 participants
participants (10)
-
Dave Robillard -
Jack O'Quin -
Jens M Andreasen -
Kai Vehmanen -
Lee Revell -
Loki Davison -
Maarten de Boer -
Niklas Klügel -
Paul Davis -
Robert Jonsson