| Age | Commit message (Collapse) | Author | Lines | |
|---|---|---|---|---|
| 2024-09-23 | add a simple read benchmark. No AVX or anything. | Birte Kristina Friesel | -3/+19 | |
| 2024-09-19 | Add AVX512 copy variant. Not particularly efficient yet, might be missing sth | Birte Kristina Friesel | -5/+272 | |
| 2024-07-19 | mbw.c: fix dfatool output format | Birte Kristina Friesel | -3/+3 | |
| 2024-07-16 | use distinct dfatool keys for copy types | Birte Kristina Friesel | -4/+11 | |
| 2024-06-17 | pthread=1: allocate threads after initializing global variables | Birte Kristina Friesel | -22/+21 | |
| 2024-05-23 | explicitly specify NUMA regions for source and target memories | Birte Kristina Friesel | -11/+32 | |
| 2024-05-23 | Add NUMA support | Birte Kristina Friesel | -1/+59 | |
| 2023-06-07 | adjust for new dfatool output format | Daniel Friesel | -3/+3 | |
| 2023-05-02 | use monotonic clock for benchmarks. gettimeofday is a really bad idea | Daniel Friesel | -10/+10 | |
| 2023-05-02 | add n_threads to parameter output | Daniel Friesel | -0/+5 | |
| 2023-05-02 | add multi-threaded benchmarks for NUMA evaluation and the likes | Daniel Friesel | -43/+163 | |
| 2023-04-28 | produce dfatool-compatible output, add run script | Daniel Friesel | -9/+7 | |
| 2023-01-17 | Removed unused header | Borislav Stanimirov | -1/+0 | |
| 2022-04-26 | Bump to v1.5 | Andras Horvath | -6/+6 | |
| 2022-04-24 | fix mempcpy not available on mac | Willian Z | -2/+2 | |
| 2020-05-14 | Fix the test for -t which was inverted | Steve Schnepp | -2/+2 | |
| Found because GCC 6.4.0 emits a warning: warning: comparison of 0 > unsigned expression is always false [-Wtautological-compare] | ||||
| 2015-05-06 | Added review changes for pull request | Stephen Bates | -3/+1 | |
| 2015-05-05 | Added -n 0 option to run loops forever | Stephen Bates | -3/+10 | |
| 2014-02-17 | Add version number. v1.4 | Andras HORVATH | -0/+4 | |
| 2014-02-17 | Fix labeling of the displayed results. | Andras HORVATH | -9/+14 | |
| 2013-10-05 | Fixed small bug with my bug fix ;-) | James Slocum | -1/+1 | |
| 2013-10-05 | Bug Fix: advance was not guaranteed to be divisable by long_size, changed ↵ | James Slocum | -4/+5 | |
| the MBLOCK test by convering to char* arrays and tracking bytes. | ||||
| 2013-10-03 | Fixed 2 bugs. | James Slocum | -3/+6 | |
| Fixed a segfault, and read error. 1) the for(t=0; t<array_size; t+=block_size) loop could write past the end of the buffer and segfault 2) The 'a' buffer was not being advanced, causing the same memory to be read over and over, this caused heavy cache hits and unrealistically high results. | ||||
| 2012-03-11 | Re-indenting, cosmetic changes | Andras HORVATH | -208/+221 | |
| 2012-03-06 | ...and get rid of the extra variable | Andras HORVATH | -1/+0 | |
| 2012-03-06 | Fix MCBLOCK test: actually copy to the full buffer, not just its first ↵ | Andras HORVATH | -2/+2 | |
| $blocksize. | ||||
| 2012-03-06 | git import | Andras HORVATH | -0/+275 | |
 |
index : mbw | |
| Memory Bandwidth Benchmark | derf |
| summaryrefslogtreecommitdiff |