code review 4671051: sync: improve RWMutex performance

sync: improve RWMutex performance
The new implementation features wait-free
fast path for readers which significantly improves
performance/scalability on read-mostly workloads.
Benchmark results on HP Z600 (2 x Xeon E5620, 8 HT cores, 2.40GHz)
are as follows:
benchmark                              old ns/op    new ns/op    delta
BenchmarkRWMutexUncontended               179.00        96.60  -46.03%
BenchmarkRWMutexUncontended-2              89.10        49.10  -44.89%
BenchmarkRWMutexUncontended-4              44.70        24.70  -44.74%
BenchmarkRWMutexUncontended-8              23.30        12.90  -44.64%
BenchmarkRWMutexUncontended-16             16.80         8.75  -47.92%
BenchmarkRWMutexWrite100                   79.60        26.80  -66.33%
BenchmarkRWMutexWrite100-2                305.00        33.00  -89.18%
BenchmarkRWMutexWrite100-4                245.00       113.00  -53.88%
BenchmarkRWMutexWrite100-8                330.00       147.00  -55.45%
BenchmarkRWMutexWrite100-16               371.00       152.00  -59.03%
BenchmarkRWMutexWrite10                    78.30        29.80  -61.94%
BenchmarkRWMutexWrite10-2                 348.00       165.00  -52.59%
BenchmarkRWMutexWrite10-4                 447.00       199.00  -55.48%
BenchmarkRWMutexWrite10-8                 564.00       180.00  -68.09%
BenchmarkRWMutexWrite10-16                492.00       192.00  -60.98%
BenchmarkRWMutexWorkWrite100             1077.00      1037.00   -3.71%
BenchmarkRWMutexWorkWrite100-2            659.00       596.00   -9.56%
BenchmarkRWMutexWorkWrite100-4            509.00       361.00  -29.08%
BenchmarkRWMutexWorkWrite100-8            603.00       351.00  -41.79%
BenchmarkRWMutexWorkWrite100-16           750.00       607.00  -19.07%
BenchmarkRWMutexWorkWrite10               990.00       951.00   -3.94%
BenchmarkRWMutexWorkWrite10-2            1119.00      1070.00   -4.38%
BenchmarkRWMutexWorkWrite10-4            1300.00      1199.00   -7.77%
BenchmarkRWMutexWorkWrite10-8            1424.00      1291.00   -9.34%
BenchmarkRWMutexWorkWrite10-16           1981.00      1786.00   -9.84%

[dvyukov, 2011-07-06 14:25:42]

Hello golang-dev@googlegroups.com (cc: golang-dev@googlegroups.com),

I'd like you to review this change to
https://go.googlecode.com/hg/

[rsc, 2011-07-12 05:09:39]

LGTM

Minor changes below.

http://codereview.appspot.com/4671051/diff/6001/src/pkg/sync/rwmutex.go
File src/pkg/sync/rwmutex.go (right):

http://codereview.appspot.com/4671051/diff/6001/src/pkg/sync/rwmutex.go#newcode8
src/pkg/sync/rwmutex.go:8: "sync/atomic"
Please sort import list.

http://codereview.appspot.com/4671051/diff/6001/src/pkg/sync/rwmutex.go#newco...
src/pkg/sync/rwmutex.go:19: // The RWMutex is fair for both readers and writers,
I'd suggest just dropping this now.
It was documenting a bug and you fixed the bug.
Otherwise simplify to

// Readers and writers are given equal priority.
// Neither can starve the other.

http://codereview.appspot.com/4671051/diff/6001/src/pkg/sync/rwmutex.go#newco...
src/pkg/sync/rwmutex.go:33: // To ensure that the lock eventually becomes
available,
Delete.

http://codereview.appspot.com/4671051/diff/6001/src/pkg/sync/rwmutex.go#newco...
src/pkg/sync/rwmutex.go:68: // If there are readers in flight, wait for them to
depart.
// Wait for active readers.

http://codereview.appspot.com/4671051/diff/6001/src/pkg/sync/rwmutex.go#newco...
src/pkg/sync/rwmutex.go:81: // Announce to readers there are no pending writers.
s/are no pending writers/is no active writer/

http://codereview.appspot.com/4671051/diff/6001/src/pkg/sync/rwmutex_test.go
File src/pkg/sync/rwmutex_test.go (right):

http://codereview.appspot.com/4671051/diff/6001/src/pkg/sync/rwmutex_test.go#...
src/pkg/sync/rwmutex_test.go:202: if (foo % writeRatio) == 0 {
Drop unnecessary ()

[dvyukov, 2011-07-12 08:22:18]

Hello rsc@golang.org (cc: golang-dev@googlegroups.com),

Please take another look.

[rsc, 2011-07-12 16:17:48]

LGTM

[rsc, 2011-07-12 16:24:26]

*** Submitted as d65164bf0d63 ***

sync: improve RWMutex performance
The new implementation features wait-free
fast path for readers which significantly improves
performance/scalability on read-mostly workloads.
Benchmark results on HP Z600 (2 x Xeon E5620, 8 HT cores, 2.40GHz)
are as follows:
benchmark                              old ns/op    new ns/op    delta
BenchmarkRWMutexUncontended               179.00        96.60  -46.03%
BenchmarkRWMutexUncontended-2              89.10        49.10  -44.89%
BenchmarkRWMutexUncontended-4              44.70        24.70  -44.74%
BenchmarkRWMutexUncontended-8              23.30        12.90  -44.64%
BenchmarkRWMutexUncontended-16             16.80         8.75  -47.92%
BenchmarkRWMutexWrite100                   79.60        26.80  -66.33%
BenchmarkRWMutexWrite100-2                305.00        33.00  -89.18%
BenchmarkRWMutexWrite100-4                245.00       113.00  -53.88%
BenchmarkRWMutexWrite100-8                330.00       147.00  -55.45%
BenchmarkRWMutexWrite100-16               371.00       152.00  -59.03%
BenchmarkRWMutexWrite10                    78.30        29.80  -61.94%
BenchmarkRWMutexWrite10-2                 348.00       165.00  -52.59%
BenchmarkRWMutexWrite10-4                 447.00       199.00  -55.48%
BenchmarkRWMutexWrite10-8                 564.00       180.00  -68.09%
BenchmarkRWMutexWrite10-16                492.00       192.00  -60.98%
BenchmarkRWMutexWorkWrite100             1077.00      1037.00   -3.71%
BenchmarkRWMutexWorkWrite100-2            659.00       596.00   -9.56%
BenchmarkRWMutexWorkWrite100-4            509.00       361.00  -29.08%
BenchmarkRWMutexWorkWrite100-8            603.00       351.00  -41.79%
BenchmarkRWMutexWorkWrite100-16           750.00       607.00  -19.07%
BenchmarkRWMutexWorkWrite10               990.00       951.00   -3.94%
BenchmarkRWMutexWorkWrite10-2            1119.00      1070.00   -4.38%
BenchmarkRWMutexWorkWrite10-4            1300.00      1199.00   -7.77%
BenchmarkRWMutexWorkWrite10-8            1424.00      1291.00   -9.34%
BenchmarkRWMutexWorkWrite10-16           1981.00      1786.00   -9.84%

R=rsc
CC=golang-dev
http://codereview.appspot.com/4671051

Committer: Russ Cox <rsc@golang.org>