> On Apr 9, 2015, at 12:07, Carl Hoefs <[email protected]> wrote:
>
> OS X 10.10
>
> I have a highly CPU-bound project that parallelizes well. A small case has
> about 5000 tasks that can execute concurrently (max is ~100,000). I would
> like to use GCD queues to take advantage of the 24 cores I have on my
> MacPro. Does GCD automatically limit the number of tasks that become
> current on its queues to the number of cores (hw.logicalcpu)?
Yes.
> Or is it the
> programmer's responsibility to manage this?
>
> Suppose I do something like this:
>
> dispatch_group_t workgroup = dispatch_group_create();
>
> dispatch_apply(4950,
> dispatch_get_global_queue(DISPATCH_QUEUE_PRIORITY_HIGH,
> 0),
> ^(size_t i) {
>
> dispatch_group_enter(workgroup);
>
> // Code block for executing task i . . .
>
> dispatch_group_leave(workgroup);
>
> });
>
> dispatch_group_wait(workgroup, DISPATCH_TIME_FOREVER);
>
Note that your use of a dispatch_group here is redundant; dispatch_apply always
executes synchronously with respect to the caller (i.e. when the above
dispatch_apply has returned, all 4950 executions of your block have already
completed).
--
Clark Smith Cox III
[email protected]
_______________________________________________
Cocoa-dev mailing list ([email protected])
Please do not post admin requests or moderator comments to the list.
Contact the moderators at cocoa-dev-admins(at)lists.apple.com
Help/Unsubscribe/Update your Subscription:
https://lists.apple.com/mailman/options/cocoa-dev/archive%40mail-archive.com
This email sent to [email protected]
