Hi there, Starting with the huge work of Simon, an in depth optimization of the infrastructure of Sage is under way (Categories, Caching...). I'd like to be able to measure somehow the benefits for the end user. Unfortunately right now there is no systematic way to measure the variation of speed of Sage from one release to the other. Or, maybe there is one but I don't know about it ?
In MuPAD, we has some notion of section in tests files, and at the beginning of each section we could indicate the expected time of the section in some more or less machine independent speed units. Any speed variation of more than a certain threshold (maybe something like 20%) was reported as test failure. I don't know if it could be feasible in Sage. In the mean term and in some orthogonal direction, I think it would be useful to have some representative benchmarks. More precisely, I'm suggesting that we should gather, in each mathematical field, a few relatively large (say from a few seconds to around 1 min) typical computations that should be representative on what is done in this field. With those under the hand we should record for each Sage version the speed of a (or a few) reference machine. This would also allow to compare the speed of different machines for our computations. What do you think ? Cheers, Florent -- To post to this group, send an email to sage-devel@googlegroups.com To unsubscribe from this group, send an email to sage-devel+unsubscr...@googlegroups.com For more options, visit this group at http://groups.google.com/group/sage-devel URL: http://www.sagemath.org
