| commit | 2c86963b093c1a0887dfc6b32c6e5ea3a80f2922 | |
| author | Tony Luck <tony.luck@intel.com> | |
| Fri, 25 Sep 2009 15:42:16 +0000 (25 08:42 -0700) | ||
| committer | Tony Luck <tony.luck@intel.com> | |
| Fri, 25 Sep 2009 15:42:16 +0000 (25 08:42 -0700) | ||
| tree | d1ed58dd0a644dd670e3724a575112fedd4b4250 | treesnapshot (tar.gz zip) |
| parent | 53cddfcc0e760d2b364878b6dadbd0c6d087cfae | commitdiff |
[IA64] implement ticket locks for Itanium
Back in January 2008 Nick Piggin implemented "ticket" spinlocks
for X86 (See commit 314cdbefd1fd0a7acf3780e9628465b77ea6a836).
IA64 implementation has a couple of differences because of the
available atomic operations ... e.g. we have no fetchadd2 instruction
that operates on a 16-bit quantity so we make ticket locks use
a 32-bit word for each of the current ticket and now-serving values.
Performance on uncontended locks is about 8% worse than the previous
implementation, but this seems a good trade for determinism in the
contended case. Performance impact on macro-level benchmarks is in
the noise.
Signed-off-by: Tony Luck <tony.luck@intel.com>
Back in January 2008 Nick Piggin implemented "ticket" spinlocks
for X86 (See commit 314cdbefd1fd0a7acf3780e9628465b77ea6a836).
IA64 implementation has a couple of differences because of the
available atomic operations ... e.g. we have no fetchadd2 instruction
that operates on a 16-bit quantity so we make ticket locks use
a 32-bit word for each of the current ticket and now-serving values.
Performance on uncontended locks is about 8% worse than the previous
implementation, but this seems a good trade for determinism in the
contended case. Performance impact on macro-level benchmarks is in
the noise.
Signed-off-by: Tony Luck <tony.luck@intel.com>