[concurrency-interest] Enforcing total sync order on modern hardware

Marko Topolnik marko at hazelcast.com
Tue Mar 24 08:46:58 EDT 2015


On Tue, Mar 24, 2015 at 1:30 PM, Alexander Terekhov <TEREKHOV at de.ibm.com>
wrote:

> IRIW is about TSO, not SC (which is TSO + store-load barrier). And TSO is
> release-acquire consistency + write atomicity.
>

That's a good observation---I wasn't sure if TSO implied an ordering across
the entire multiprocessor system, or an ordering of each individual
processor's stores. If it is generally understood that it means the former,
then it's all IRIW needs to be safe. However, IRIW specifically avoids my
scenario where one of the reading processors is also the writer an thus
excluded from the TSO guarantees as provided by Intel, for example.

---
Marko
-------------- next part --------------
An HTML attachment was scrubbed...
URL: <http://cs.oswego.edu/pipermail/concurrency-interest/attachments/20150324/d97b9d1e/attachment.html>


More information about the Concurrency-interest mailing list