[perl #36250] [TODO] STM threading design

[Chip Salzenberg]

# New Ticket Created by Chip Salzenberg
# Please include the string: [perl #36250]
# in the subject line of all future correspondence about this issue.
# <URL: https://rt.perl.org/rt3/Ticket/Display.html?id=36250 >

Do detailed design & documentation on STM-based threading system.

STM is Software Transactional Memory. It's wicked cool. Really.
Google it. You'll be glad you did.

BTW, if you're wondering how this can work efficiently in Parrot:
Remember that threading contention only applies to those variables
which are marked 'shared'. So the STM system need only look at their
accesses transactioned.
--
Chip Salzenberg <ch...@pobox.com>

[Larry Wall]

On Sat, Jun 11, 2005 at 02:23:10PM -0700, Chip Salzenberg wrote:
: STM is Software Transactional Memory. It's wicked cool. Really.

: Google it. You'll be glad you did.

I'm particularly interested by the scalability of

http://www.cambridge.intel-research.net/~rennals/faststm.html

(There's an implementation on Source Forge as well.)

: BTW, if you're wondering how this can work efficiently in Parrot:

: Remember that threading contention only applies to those variables
: which are marked 'shared'. So the STM system need only look at their
: accesses transactioned.

On the other hand, some of the parallelizable primitives we're
building into Perl 6 (junctions, hyperops, and especially pipes) may
well induce more accidental "sharing" than appears on the surface
(think of a lexical that is visible to both ends of a pipeline).
So STM is probably a good direction to be moving in any case.

And if we need special syntax to keep such accidental sharing under
control, now would be a good time to discover that rather than later.

[Note: followups directed to p6l by default.]

Larry