libera/#petalisp - IRC Chatlog

9:43:46 rotateq shame on me i have read just parts of the petalisp main paper yet

9:51:15 heisig rotateq: There is not such thing as a Petalisp main paper.

9:51:29 heisig Just some rumors spread around at some of the ELSs.

9:51:37 rotateq heisig: but I think you know what I mean with that :)

9:51:43 heisig Yes, I do :)

9:52:10 heisig But this is really one of the problems I face - I don't want to publish things that are likely to change soon.

9:52:35 heisig And I don't want to publish performance numbers when the implementation is still lacking.

9:52:38 rotateq but good to know you're (quite well?) back from your traveling

9:52:52 heisig Yes, I am :)

9:52:58 rotateq i know what you mean, or when the perfectionism triggers

9:53:19 heisig Yes, the perfectionism is a huge issue for me.

9:53:59 heisig There are some parts that I consider 'perfect', so the others that aren't bother me a lot.

9:54:06 rotateq it often was in my life for a long time and still is

9:55:10 heisig The worst part was when I presented some preliminary performance results at our computing center, and they were complaining that the performance was still off by a factor of 4 from what you can do in C or Fortran.

9:55:26 heisig The result is that I developed sb-simd :)

9:55:27 rotateq pff...

9:56:08 rotateq then fight back saying how much more performant things on fpga can be vs assembly

9:56:19 rotateq ah I see

9:56:26 hayley "Wait, you guys don't have a GPU backend?"

9:56:36 heisig hayley: Heh!

9:56:41 rotateq QPU :D

9:57:57 heisig But I'm getting there. I can now clearly see a version of Petalisp that is presentable.

9:58:10 heisig Which is good, because I'd like to write my thesis later this year.

9:59:26 rotateq around christmas like the other one? ^^ iirc

10:00:32 heisig More like starting from next week.

10:01:24 rotateq sounds good

10:02:04 heisig I really should finish that vectorizer today. But again - perfectionism got the best of me, and I want to handle all the edge cases correctly.

10:02:29 heisig Gah! Maybe I should trade sleep for coding time...

10:04:55 rotateq and no one will be mad at you for that?

10:06:42 heisig Right, I shouldn't do that. Also, trading in sleep is not a viable long term strategy anyway.

10:07:50 heisig Maybe I should just focus and stop spamming IRC.

13:41:16 rotateq heisig: but when I first saw the extended euclidean in section 3.2 I instantly laughed (in a very positive way) and knew that's from TAoCP book 2 chapter 4 :)

13:45:45 heisig rotateq: Yes, I was positively surprised when I realized I could solve that problem with TAoCP :)

13:48:43 rotateq see you later :)

13:48:57 heisig See you!

13:49:07 rotateq it's my favorite chapter