Date
Attendees
Agenda:
- Progress since the last meeting
- Some results from a poor-man's script running over 33 files produced in the test run(s) but without isolating from file-system overheads, caches, etc...
gzip | bzip2 / bzip2 -9 | lz4 / lz4 -9 | xz / xz -9 | |||||||||
quantization | tcompress | tdecompress | Factor | tcompress | tdecompress | Factor | tcompress | tdecompress | Factor | tcompress | tdecompress | Factor |
---|---|---|---|---|---|---|---|---|---|---|---|---|
seconds/file | seconds/file | seconds/file | seconds/file | seconds/file | seconds/file | seconds/file | seconds/file | |||||
4 | 4.73 | 0.67 | 6.73 | 5.79/5.91 | 2.27/2.39 | 9.96/9.96 | 0.36/15.06 | 0.24/0.21 | 3.69/5.43 | 53.82/58.61 | 1.00/0.91 | 8.20/8.20 |
8 | 6.85 | 0.70 | 5.55 | 6.18/5.88 | 2.39/2.33 | 8.20/8.20 | 0.42/16.55 | 0.24/0.18 | 3.34/4.78 | 45.61/54.42 | 0.97/0.97 | 8.20/8.22 |
16 | 9.33 | 0.70 | 4.70 | 5.97/5.94 | 2.45/2.42 | 7.02/7.02 | 0.42/16.39 | 0.30/0.27 | 3.11/4.20 | 46.30/53.70 | 1.21/1.21 | 6.96/6.96 |
32 | 11.45 | 0.79 | 4.04 | 6.09/6.00 | 2.61/2.48 | 6.14/6.14 | 0.45/11.42 | 0.30/0.24 | 2.94/3.79 | 42.36/50.55 | 1.45/1.36 | 6.00/6.03 |
64 | 11.24 | 0.85 | 3.62 | 6.06/6.18 | 2.58/2.55 | 5.48/5.48 | 0.61/8.18 | 0.24/0.18 | 2.82/3.27 | 39.12/47.03 | 1.64/1.52 | 5.29/5.30 |
128 | 12.97 | 0.82 | 3.38 | 6.79/6.52 | 2.73/2.64 | 4.88/4.88 | 0.52/7.06 | 0.30/0.30 | 2.66/2.99 | 42.85/49.33 | 1.70/1.73 | 4.75/4.76 |
Vanilla | 4.52 | 0.88 | 1.716 | 9.79/10.12 | 4.61/4.39 | 1.80/1.80 | 0.42/3.97 | 0.27/0.33 | 1.50/1.50 | 23.21/29.12 | 4.18/4.15 | 1.87/1.88 |
- Plans for a subsequent round of tests?
Action items
- Sophie Reed single epoch catalog comparison (with help from Paul Price )
- Robert Gruendl will work out limits on BSCALE that would effectively lower the saturation.