Memory barriers: Difference between revisions
Jump to navigation
Jump to search
| [unchecked revision] | [unchecked revision] |
Content added Content deleted
mNo edit summary |
No edit summary |
||
| Line 8: | Line 8: | ||
Ridiculous Fish's [http://ridiculousfish.com/blog/archives/2007/02/17/barrier/ controversial article] about using and abusing memory barriers. (and [http://www.thinkingparallel.com/2007/02/19/please-dont-rely-on-memory-barriers-for-synchronization/some bashing of this article] just for a different POV) |
Ridiculous Fish's [http://ridiculousfish.com/blog/archives/2007/02/17/barrier/ controversial article] about using and abusing memory barriers. (and [http://www.thinkingparallel.com/2007/02/19/please-dont-rely-on-memory-barriers-for-synchronization/some bashing of this article] just for a different POV) |
||
[http://www.gelato.unsw.edu.au/lxr/source/Documentation/memory-barriers.txt Documentation with explanations on what memory barriers are used in linux kernel and why.] |
|||
Revision as of 21:21, 1 February 2009
Memory barriers ensure ordering of load/store commands.
Usually CPUs and compilers reorder these instructions for more efficient execution. Lock-free algorithms often require that perceived execution order be strict, that's when memory barriers or memory fences come in handy.
More detailed description in following articles:
Memory Ordering in Modern Microprocesors, Part 1 and Part 2
Ridiculous Fish's controversial article about using and abusing memory barriers. (and bashing of this article just for a different POV)
Documentation with explanations on what memory barriers are used in linux kernel and why.