Commit Graph

302 Commits

Author SHA1 Message Date
Daniel Lemire f143d4e3f4 Adding cache access stats. 2018-11-28 10:53:57 -05:00
Daniel Lemire 1f8f03bbf7 Adding cache access stats. 2018-11-28 10:53:41 -05:00
Daniel Lemire b858a404f7 Adding missing include 2018-11-28 10:29:57 -05:00
Daniel Lemire c1805783fc Tweaking performance. 2018-11-27 21:13:31 -05:00
Daniel Lemire 8648c4108e MOre cleaning. 2018-11-27 20:42:35 -05:00
Daniel Lemire ba0f6fea51 Cleaning. 2018-11-27 17:38:53 -05:00
Daniel Lemire 58ac242770 Ok. Let us benchmark this thing. 2018-11-27 15:05:50 -05:00
Daniel Lemire a43b0772e1 Lots and lots of cleaning. 2018-11-27 14:37:59 -05:00
Daniel Lemire 5fae7b2100 Still working 2018-11-27 10:10:39 -05:00
Daniel Lemire 50defa510f Stupid work. 2018-11-26 16:55:24 -05:00
Daniel Lemire 86a75462c5 Adding the ability of doing a dump. 2018-11-23 22:20:57 -05:00
Daniel Lemire 5e50536c06 Better documentation 2018-11-23 15:46:16 -05:00
Daniel Lemire 08ae836aa1 Removing dead file. 2018-11-20 20:53:42 -05:00
Daniel Lemire 5bdf19bb18 Removing parsers that are unfair. 2018-11-20 20:08:02 -05:00
Daniel Lemire 21ee490d18 Added memcpy. 2018-11-20 18:06:03 -05:00
Daniel Lemire f1228d0b98 Comment. 2018-11-20 17:55:25 -05:00
Daniel Lemire e4d4158e3f Added dependencies. 2018-11-20 16:43:22 -05:00
Daniel Lemire bbff6c3edb Added another ref. 2018-11-20 14:32:12 -05:00
Daniel Lemire 7647cb2e49 Added dropbox 2018-11-20 14:09:43 -05:00
Daniel Lemire 1fcd2688f8 Better documentation. 2018-11-20 12:59:06 -05:00
Daniel Lemire bbeb64a70b Cleaning documentation. 2018-11-20 12:54:06 -05:00
Daniel Lemire 78e75a8bae Even faster. 2018-11-20 11:56:10 -05:00
Daniel Lemire 7dd590c43c Saving faster version. 2018-11-20 11:02:39 -05:00
Daniel Lemire 006eb76bcd Adding a special case. 2018-11-16 21:49:37 -05:00
Daniel Lemire 47ae00895a Forgot to save... 2018-11-09 21:42:44 -05:00
Daniel Lemire 17f5d0517d Opting for a more common intrinsic. 2018-11-09 21:41:15 -05:00
Daniel Lemire 3c7a17cbc1
Delete LIMITATIONS.md 2018-11-09 21:32:56 -05:00
Daniel Lemire 78ec4142ff
Delete rapidjsondesign.md 2018-11-09 21:32:47 -05:00
Daniel Lemire 76074a821f Various cleaning steps. 2018-11-09 21:31:14 -05:00
Daniel Lemire 0e5b939568 Merge branch 'master' of github.com:lemire/simdjson 2018-11-09 15:16:25 -05:00
Daniel Lemire c1a7e79862 Lifting the mem limit. (Dirty commit.) 2018-11-09 15:16:05 -05:00
Daniel Lemire df65de4ae2 Tuning presentation and fixing a problem with minifier benchmark. 2018-10-23 21:36:32 -04:00
Daniel Lemire 18633e02d2 Added more thorough testing. 2018-10-23 20:19:33 -04:00
Daniel Lemire f0af315315 Elaborating on the limitations. 2018-10-23 14:03:30 -04:00
Daniel Lemire 9738af68c8 Fixing up the code point parsing. I think that what is there is now correct.
I believe that there was a case of early optimization.
2018-10-19 22:07:22 -04:00
Daniel Lemire 8315f4c888 Cleaning up the code. 2018-10-17 21:31:22 -04:00
Daniel Lemire 35381279c3 Maybe we can do away with the fast ASCII trick. 2018-10-17 21:05:38 -04:00
Daniel Lemire e517414080 We include character-encoding validation. 2018-10-17 19:22:09 -04:00
Daniel Lemire 355e5d2ed3 Checking for unescaped chars. 2018-10-17 15:08:49 -04:00
Daniel Lemire 9fc8d8444b We want to allow more than just arrays and objects, as per the JSON spec. 2018-10-17 13:57:42 -04:00
Daniel Lemire 57d7544b52 Adding... 2018-10-17 13:23:45 -04:00
Daniel Lemire 6cc5131f7a Adding an allparserscheckfile program. 2018-10-17 12:00:44 -04:00
Daniel Lemire 0453d546f8 Adding remark. 2018-10-16 20:43:37 -04:00
Daniel Lemire 8f704fdb7c Making the test tougher. 2018-10-08 16:32:54 -04:00
Daniel Lemire 7eb7cd265a We can now parse crazy things like pi to 100 digits. 2018-10-08 15:24:16 -04:00
Daniel Lemire 70c122074f Tests. 2018-10-08 14:41:36 -04:00
Daniel Lemire e2a3f751cf Counting numbers. 2018-10-04 09:48:00 -04:00
Daniel Lemire aeacd26366 Adding mispredicted branch counts. 2018-10-04 09:47:34 -04:00
Daniel Lemire 930533b6da Normalizing the number of cycles. 2018-10-04 09:33:41 -04:00
Daniel Lemire 2ad9891b66 I think NO_PDEP_PLEASE should be defined by default. It seems
to be generally better/faster. More instructions, but also
more instructions per cycle, so it ends up being a net win.
2018-10-03 21:42:27 -04:00