Mercurial > hg > silvet
comparison testdata/timing/results.txt @ 115:fbf9b824aaf3 bqvec-openmp
Report on last couple of tests
author | Chris Cannam |
---|---|
date | Wed, 07 May 2014 09:48:56 +0100 |
parents | 2169e7a448c5 |
children | c3c768ac4340 |
comparison
equal
deleted
inserted
replaced
114:b2f0967cb8d1 | 115:fbf9b824aaf3 |
---|---|
206 commit:62b7be1226d into bqvec branch commit:19f6832fdc8a | 206 commit:62b7be1226d into bqvec branch commit:19f6832fdc8a |
207 | 207 |
208 real 0m44.650s | 208 real 0m44.650s |
209 user 2m19.997s | 209 user 2m19.997s |
210 sys 0m0.343s | 210 sys 0m0.343s |
211 | |
212 commit:c4eae816bdb3, as commit:ac750e222ad3 but with some logic to | |
213 make using the shifts optional (though on by default). Performance | |
214 *should* be unchanged here. | |
215 | |
216 real 0m43.979s | |
217 user 2m19.297s | |
218 sys 0m0.360s | |
219 | |
220 commit:b2f0967cb8d1, as commit:c4eae816bdb3 but storing the templates | |
221 as float arrays and then pulling them out into individual | |
222 one-per-shift-factor double arrays each of which is explicitly | |
223 allocated with the proper alignment. Uses more memory, and the code is | |
224 ugly, but gets aligned starts for slightly more of the vector ops. | |
225 | |
226 real 0m50.856s | |
227 user 2m44.937s | |
228 sys 0m0.463s | |
229 |