Bisection bandwidth – Wikipedia

From Wikipedia:Theoretical support for the importance of this measure of network performance was developed in the PhD research of Clark Thomborson (formerly Clark Thompson).[3] Thomborson proved that important algorithms for sorting, Fast Fourier transformation, and matrix-matrix multiplication become communication-limited—as opposed to CPU-limited or memory-limited—on computers with insufficient bisection width. F. Thomson Leighton’s PhD research[4] tightened Thomborson’s loose bound [5] on the bisection width of a computationally-important variant of the De Bruijn graph known as the shuffle-exchange graph. Based on Bill Dally’s analysis of latency, average case throughput, and hot-spot throughput of m-ary n-cube networks[2] for various m, It can be observed that low-dimensional networks, in comparison to high-dimensional networks (e.g., binary n-cubes) with the same bisection width (e.g., tori), have reduced latency and higher hot-spot throughput.[6]
Sent from my iPad

Posts navigation

1 2 3 16 17 18 19
Scroll to top