Show enters and exits. Hide enters and exits.
| 06:57:17 | brixen | joachimm: ping |
| 08:27:41 | boyscout | Enumerable#each_cons, #each_slice - 359e964 - Marc-Andre Lafortune |
| 08:27:41 | boyscout | 1.8.7: Enumerable#each_with_index - e6240bb - Marc-Andre Lafortune |
| 08:31:48 | boyscout | CI: e6240bb success. 2647 files, 10179 examples, 32469 expectations, 0 failures, 0 errors |
| 20:33:53 | rue | dbussink: Still around? |
| 21:45:29 | evan | slava: what PHP to machine code thing are you watching? |
| 21:47:24 | slava | evan: http://www.youtube.com/watch?v=kKySEUrP7LA |
| 21:47:44 | slava | evan: its yet another one of those "our language has serious semantics issues so these are the hacks we use" talks :-) |
| 21:50:03 | slava | evan: so what's your new JIT all aobut? |
| 21:50:05 | slava | using LLVM again? |
| 21:52:14 | evan | giving it a shot again, yes. |
| 21:52:30 | evan | there are a number of projects now trying to use it similarly |
| 21:52:43 | evan | which means move people pushing on it to get good performance out of the JIT |
| 21:52:46 | evan | they've already done some work on it |
| 21:53:42 | slava | are you going to do type feedback? |
| 21:54:03 | evan | yeah |
| 21:54:13 | evan | i'm working on a roadmap for the JIT |
| 21:54:22 | evan | type feedback is on the roadmap, but it's not at the top :) |
| 21:54:52 | evan | there are a number of things on the list above it that are easier to do |
| 21:55:02 | evan | but are based on type feedback / IC contents |
| 22:02:15 | slava | I'm adding SIMD primitives soon |
| 22:02:22 | slava | gonna rice out raytracer and nbody benchmarks :) |
| 22:04:46 | evan | hehe |
| 22:05:00 | evan | are you going to expose them to the high level to be used nearly directly? |
| 22:05:10 | slava | yeah, as 4-element vectors |
| 22:05:44 | slava | the vector words in our math library will be overloaded on the accelerated data types, to use intrinsics directly intsead of mapping over the array with an arithmetic op |
| 22:06:13 | evan | right |
| 22:07:21 | slava | we have pretty good cross-platform bindings for advanced opengl shader stuff, full screen graphics and game controller/joystick apis, so simd is a natural next step |
| 22:07:39 | slava | once the gc is incremental factor should be a great game dev platform |
| 22:10:26 | slava | evan: I'd be interested in seeing your roadmap |
| 22:43:02 | evan | slava: http://gist.github.com/109758 |
| 22:44:08 | slava | ok |
| 22:44:29 | evan | Oh, somewhere in there is generic primitive inlining |
| 22:44:43 | evan | the first 3 or 4 are pretty much in order |
| 22:44:49 | evan | after that, it's fuzzier |
| 22:56:22 | slava | evan: cool |
| 23:22:47 | rue | 3. ??? |