Skip to main content

Table 11 Cache misses for parallel DL trace algorithms, in millions, on Xeon4

From: String correction using the Damerau-Levenshtein distance

A B PP_DL_TRACE PP_LSDL_TRACE PP_Strip_TRACE L vs D S vs D S vs L
40000 40000 247 636 16 -157.0% 93.6% 97.5%
80000 80000 1312 2431 15 -85.3% 98.9% 99.4%
120000 120000 4940 6814 34 -37.9% 99.3% 99.5%
160000 160000 ** 12,774 53    99.6%
200000 200000 ** 28,908 85    99.7%
240000 240000 ** 30,529 110    99.6%
280000 280000 ** 40,803 154    99.6%
320000 320000 ** 53,892 179    99.7%
360000 360000 ** 91,621 796    99.1%
400000 400000 ** 188,325 2727    98.6%