D. Johnson et al., LOW-COST, HIGH-PERFORMANCE BARRIER SYNCHRONIZATION ON NETWORKS OF WORKSTATIONS, Journal of parallel and distributed computing, 40(1), 1997, pp. 131-137
Citations number
22
Categorie Soggetti
Computer Sciences","Computer Science Theory & Methods
Circulating active barrier (CAB) is a new low-cost, high-performance h
ardware mechanism for synchronizing multiple processing elements (PEs)
in networks of workstations at fine-grained programmed barriers. CAB
is significantly less complex than other hardware barrier synchronizat
ion mechanisms with equivalent performance, using only a single conduc
tor, such as a wire or copper run on a printed-circuit board, to circu
late barrier packets between PEs. When a PE checks in at a barrier, th
e CAB hardware will decrement the count associated with that barrier i
n a bit-serial fashion as a barrier packet passes through, and then wi
ll monitor the packets until all PEs have checked in at the barrier. T
he ring has no clocked sequential logic in the serial loop. A cluster
controller (CC) generates packets for active barriers, removes packets
when no longer needed, and resets counters when all PEs have seen the
zero-count. A hierarchy of PEs can be achieved by connecting the CCs
in intercluster rings. When using conservative timing assumptions, the
expected synchronization times with optimal clustering are shown to b
e under 1 mu s for as many as 4096 PEs in multiprocessor workstations
or 1024 single-processor workstations. The ideal number of clusters fo
r a two-dimensional hierarchy of N PEs is shown to be [N(D + G)/(I + G
)](1/2), where G is the gate propagation delay, D is the inter-PE dela
y, and I is the intercluster transmission time. CAB allows rapid, cont
ention-free check-in and proceed-from-barrier and is applicable to a w
ide variety of system architectures and topologies. (C) 1997 Academic
Press.