Proc. Intl. Conf. on Parallel Processing, 1993.
a) the Mellor−Crummey and Scott scalable exclusive locktailnextspinnextspinnextspinnilb) the Mellor−Crummey and Scott scalable reader−writer locktailreadercountstatestatefirstnextnextstatewriterspinspinnextspinniltailstatenextspinprevlockstatenextspinprevlockstatenextspinprevlocknilnil∆∆ScalableLocks
××SimpleLocks∆×∆-Nodelay∆××-25300
250usecusecdelay∆-delay
×∆∆Lock
××Time∆(usec)
×∆150
×∆∆××∆×∆∆∆∆∆××∆∆∆×∆∆∆∆∆×∆×∆××∆×∆××××∆××∆××××∆×0
1
3579111315
NumberofProcessors
300LockTime(usec)
150
Readers/Writers3:1Readers/Writers7:1-Readers/Writers15:1|-ReadersOnly
×-
∆-
0
∆∆×∆×∆.×....∆...×...∆×.......∆×...∆.....×||.....×∆.....||......×.|.∆.....|×..|...|∆..∆×|||×|....×|...∆.....×|.....∆∆.|×.|...∆×13579111315
NumberofProcessors
