The SPP Convex

Machine data

8 Processors on 1 hypernode
Cache size: 1 Mbyte
Cache line size : 32 bytes

Performance on the benchmark

Memory allocated : NEARSHARED

Performance with one processor

Performance with two processors

Performance with four processors

Performance with eight processors

Comparison Plots

Results

Effective Prefetch time : 90.3 ns
TLB miss time : 116 ns (without bus contention)
Cache miss time : 1110 ns (with 1 processors)
Cache miss time : 1125 ns (with 2 processors)
Cache miss time : 1240 ns (with 4 processors)
Cache miss time : 1430 ns (with 6 processors)
Cache miss time : 2279 ns (with 8 processors)


Tests run by Shivanand Bhajekar <shivanan@isi.edu>