Re:Intel Discontinues the Intel Developer Forum; IDF17 Cance...
by ww520 on Monday April 17, 2017 @09:42PM
(#14135899)
NUMA style architecture has been existed for a long while now. It doesn't seem to take over the world.
[ Parent
| on HN ]
Re:Intel Discontinues the Intel Developer Forum; IDF17 Cance...
by dom0 on Tuesday April 18, 2017 @03:05PM
(#14141869)
NUMA is used, when it is actually needed, which is not that often. Most other stuff gets by without requiring NUMA enlightenment (if you think competent parallel software developers are hard to come by... finding competent NUMA people will make you reconsider that assessment).
Re:Intel Discontinues the Intel Developer Forum; IDF17 Cance...
by adrianratnapala on Tuesday April 18, 2017 @12:59AM
(#14136599)
But the practical case for it should grow as long transistor counts increase while the speed of light doesn't.
Re:Intel Discontinues the Intel Developer Forum; IDF17 Cance...
by wtallis on Tuesday April 18, 2017 @03:00AM
(#14136983)
And NUMA-like architectures on a single CPU die have been increasingly common between Intel's multiple ring busses on larger Xeons and AMD's 4-core clusters on Ryzen. Even per-core L2 caches violate the assumption that a given memory address is equally accessible from any processor core. You can't pretend that memory is all equidistant from the processor cores unless you want everything running equally slow.
Re:Intel Discontinues the Intel Developer Forum; IDF17 Cance...
by visionscaper on Tuesday April 18, 2017 @03:30AM
(#14137059)
We already have practical use cases: training and evaluation of large neural networks.