### Table 1: Comparison between multi-processors, vector machines, and VLIWs

2003

### Table 1: Message Start-up Overheads in Several Current Commercial and Research Multiprocessors. Machine Network Channel Cycle

1996

"... In PAGE 11: ... Thus, it restricts the fastest rate at which messages can be dispatched from a processor. This is denoted as sw, where we have sw = 1=tsw (messages/ sec) (1) Representative values of tsw of some current machines are shown in Table1 . For systems with high value of tsw, as is the case with many of the recent machines like IBM/SP, CM-5, the achievable message injection can be quite limited.... ..."

Cited by 2

### Table 1: Message Start-up Overheads in Several Current Commercial and Research Multiprocessors. Machine Network Channel Cycle

1996

"... In PAGE 12: ... Thus, it restricts the fastest rate at which messages can be dispatched from a processor. This is denoted as sw, where we have sw = 1=tsw (messages/ sec) (1) Representative values of tsw of some current machines are shown in Table1 . For systems with high value of tsw, as is the case with many of the recent machines like IBM/SP, CM-5, the achievable message injection can be quite limited.... ..."

Cited by 2

### Table 1. CPU time for Parallel External A* in GIOP on a multiprocessor machine.

2006

Cited by 9

### Table 2. CPU time for Parallel External A* in Optical Telegraph on a multiprocessor machine.

2006

Cited by 9

### Table 5: Comparing MATLAB with BLAS and LAPACK for the CEM test. The tests are done on one processor on a SUN Enterprise 10000. It is a multiprocessor machine with shared memory.

"... In PAGE 14: ... In each test block Arnoldi needs less than half the time. In Table5 we show comparisons between MATLAB and LAPACK/BLAS [1] for matrix vector product and Gaussian elimination. For the matrix vector product, the BLAS routine zgemv is 14 times faster than MATLAB, and the Gaussian elimination routine in LAPACK zgesv is 1.... ..."

### Table 6: Task parallel performance on Thinking Machines CM-5 message-passing distributed- memory multiprocessor

### Table 2.1: Multiprocessor Latency and Bandwidth. Latency Bandwidth n1=2 Theoretical Machine OS

### Table 3: Data parallel performance on Thinking Machines CM-5 message-passing distributed-memory multiprocessor Circuit Processors

"... In PAGE 22: ... It is important to notice that there are cases in which a combination of data and task parallelism provides better performance over either type of parallelism individually. Compared to Table3 , the results on the 128 processor runs show that the combined task... ..."