Logs: freenode/#haskell
| 2021-05-11 20:28:24 | <shapr> | Rembane: I don't know enough about erlang to be able to answer that |
| 2021-05-11 20:28:59 | → | geowiesnot joins (~user@87-89-181-157.abo.bbox.fr) |
| 2021-05-11 20:29:17 | × | geekosaur quits (930099da@rrcs-147-0-153-218.central.biz.rr.com) (Quit: Connection closed) |
| 2021-05-11 20:29:41 | × | jamm_ quits (~jamm@unaffiliated/jamm) (Ping timeout: 250 seconds) |
| 2021-05-11 20:29:42 | → | geekosaur joins (930099da@rrcs-147-0-153-218.central.biz.rr.com) |
| 2021-05-11 20:29:51 | <shapr> | m_shiraeeshi: good question, GPUs might be good at graph reduction? I couldn't find any research about doing graph reduction on GPUs |
| 2021-05-11 20:30:12 | <Rembane> | shapr: That's a good answer. What I was intending to ask before Erlang got in there was: how are the cores going to communicate? Message passing? |
| 2021-05-11 20:30:15 | <shapr> | m_shiraeeshi: the graphcore hardware I mentioned near the end might be a graph reduction coprocessor, I haven't dug into that |
| 2021-05-11 20:30:41 | → | merijn joins (~merijn@83-160-49-249.ip.xs4all.nl) |
| 2021-05-11 20:31:37 | × | wonko7 quits (~wonko7@62.115.229.50) (Ping timeout: 252 seconds) |
| 2021-05-11 20:32:13 | <shapr> | Rembane: I don't know enough about that part, I assume there's a program root 'main' and if there's a place where work could be sent off to another core, that work gets marked as "needed by the main node" or maybe explicitly passed to a core? |
| 2021-05-11 20:32:36 | <shapr> | I haven't read all of the reduceron papers yet |
| 2021-05-11 20:33:43 | <m_shiraeeshi> | wait, what do those bitcoin farmers do with GPUs? |
| 2021-05-11 20:33:53 | <Rembane> | shapr: That sounds good. Spontaneously, DMA and IRQs sounds not too bad. But maybe they are when things become too parallel. |
| 2021-05-11 20:34:23 | <Rembane> | The Reduceron is such a fantastically cheesy name <3 |
| 2021-05-11 20:34:52 | <shapr> | I agree, I wish they had a logo |
| 2021-05-11 20:34:56 | <shapr> | I'd buy the t-shirt |
| 2021-05-11 20:35:16 | × | heatsink quits (~heatsink@108-201-191-115.lightspeed.sntcca.sbcglobal.net) (Remote host closed the connection) |
| 2021-05-11 20:35:32 | <Rembane> | Me too. And all the swag! |
| 2021-05-11 20:35:33 | catern- | is now known as catern |
| 2021-05-11 20:35:42 | <Rembane> | Stickers, plushy, everything! |
| 2021-05-11 20:35:49 | × | merijn quits (~merijn@83-160-49-249.ip.xs4all.nl) (Ping timeout: 245 seconds) |
| 2021-05-11 20:36:07 | → | apache8080 joins (~rishi@wsip-70-168-153-252.oc.oc.cox.net) |
| 2021-05-11 20:36:16 | → | heatsink joins (~heatsink@108-201-191-115.lightspeed.sntcca.sbcglobal.net) |
| 2021-05-11 20:37:53 | <shapr> | Rembane: we could make one and ask Runciman |
| 2021-05-11 20:37:57 | <shapr> | or Naylor |
| 2021-05-11 20:38:01 | catern | is now known as catern- |
| 2021-05-11 20:38:06 | → | dominicusin[m] joins (dominicusi@gateway/shell/matrix.org/x-gdindupjdyjnygcd) |
| 2021-05-11 20:38:09 | × | Varis quits (~Tadas@unaffiliated/varis) (Remote host closed the connection) |
| 2021-05-11 20:38:10 | × | cfricke quits (~cfricke@unaffiliated/cfricke) (Quit: WeeChat 3.1) |
| 2021-05-11 20:38:13 | catern- | is now known as catern |
| 2021-05-11 20:38:20 | → | hyiltiz joins (~quassel@unaffiliated/hyiltiz) |
| 2021-05-11 20:38:31 | <Rembane> | shapr: That's a very cunning plan! |
| 2021-05-11 20:38:34 | <shapr> | Rembane: I gave this talk at a "futures of programming" Recurse Center event, and I got several interesting questions |
| 2021-05-11 20:38:47 | <shapr> | One was "does graph reduction work for non-FP languages?" |
| 2021-05-11 20:39:07 | → | JaakkoLuttinen[m joins (jluttinema@gateway/shell/matrix.org/x-ysahyblxieeuvzlx) |
| 2021-05-11 20:39:17 | <shapr> | and one I couldn't find the answer for: can you do graph reduction on a GPU? |
| 2021-05-11 20:39:20 | → | mly joins (mlydisenco@gateway/shell/matrix.org/x-rpdrjxcbyachxvja) |
| 2021-05-11 20:40:24 | → | acidjnk_new2 joins (~acidjnk@p200300d0c72b9509002dc6dbf76c6cb5.dip0.t-ipconnect.de) |
| 2021-05-11 20:40:41 | <Taneb> | shapr: Runciman's retired now I believe, sadly |
| 2021-05-11 20:40:57 | × | ubert quits (~Thunderbi@p200300ecdf005e5ce6b318fffe838f33.dip0.t-ipconnect.de) (Ping timeout: 250 seconds) |
| 2021-05-11 20:40:59 | × | Guest87676 quits (~laudiacay@67.176.215.84) (Ping timeout: 246 seconds) |
| 2021-05-11 20:41:11 | <shapr> | oh, didn't know that |
| 2021-05-11 20:41:19 | → | jtojnar joins (jtojnarmat@gateway/shell/matrix.org/x-csatkhbdsihyljoz) |
| 2021-05-11 20:41:23 | → | sm[m] joins (simonmicma@gateway/shell/matrix.org/x-ohzxdfhmrklaasyg) |
| 2021-05-11 20:41:31 | <Rembane> | My hunch is that GPUs are very bad at graph reductions, but I can't say why. |
| 2021-05-11 20:41:48 | → | Ericson2314 joins (ericson231@gateway/shell/matrix.org/x-lfojhdlxyjhgujwr) |
| 2021-05-11 20:41:50 | → | nut joins (~gtk@roc37-h01-176-170-197-243.dsl.sta.abo.bbox.fr) |
| 2021-05-11 20:41:52 | <Uniaika> | Rembane: because they are hardware specialised in dealing with triangles :-P |
| 2021-05-11 20:42:00 | → | CrabMan joins (phi-matrix@gateway/shell/matrix.org/x-nbtkwhyasulaaowv) |
| 2021-05-11 20:42:02 | <Rembane> | Uniaika: Indeed, and triangles aren't graphs. :) |
| 2021-05-11 20:42:15 | → | domenkozar[m] joins (domenkozar@NixOS/user/domenkozar) |
| 2021-05-11 20:42:18 | → | heck-to-the-gnom joins (heck-to-th@gateway/shell/matrix.org/x-fidipsyutsvkvjuh) |
| 2021-05-11 20:42:23 | <maerwald> | can you make it a graph? You're an expert, right? |
| 2021-05-11 20:42:28 | <Rembane> | Maybe immutable data is more important than the paradigm of the language? |
| 2021-05-11 20:42:33 | <Taneb> | (just remembered I'm connecting to IRC from a server named for Colin Runciman. Feels significant) |
| 2021-05-11 20:42:35 | <maerwald> | also make it perpendicular |
| 2021-05-11 20:42:38 | → | Guillaum joins (guiboumatr@gateway/shell/matrix.org/x-hddjdnxujganspku) |
| 2021-05-11 20:42:49 | <Uniaika> | heya Taneb and domenkozar[m] :) |
| 2021-05-11 20:42:54 | <Uniaika> | welcome back Guillaum |
| 2021-05-11 20:42:55 | <shapr> | Taneb: I met Runciman a few times at various ICFPs, enjoying talking to him |
| 2021-05-11 20:42:58 | × | hyiltiz quits (~quassel@unaffiliated/hyiltiz) (Ping timeout: 276 seconds) |
| 2021-05-11 20:43:02 | → | rednaZ[m] joins (r3dnazmatr@gateway/shell/matrix.org/x-ixpddsckemdoyohz) |
| 2021-05-11 20:43:06 | × | tromp quits (~tromp@dhcp-077-249-230-040.chello.nl) (Remote host closed the connection) |
| 2021-05-11 20:43:10 | <Taneb> | shapr: he was pretty cool, he supervised my master's thesis |
| 2021-05-11 20:43:13 | → | lnxw37d4 joins (lnxw37d4ma@gateway/shell/matrix.org/x-gjxdbkbobeywhhqj) |
| 2021-05-11 20:43:14 | <shapr> | oh nice! |
| 2021-05-11 20:43:18 | shapr | looks up that thesis |
| 2021-05-11 20:43:28 | <Taneb> | (it was awful for me reasons, don't) |
| 2021-05-11 20:43:36 | <Uniaika> | Rembane: immutable data, composition, recursion |
| 2021-05-11 20:43:47 | <shapr> | Taneb: you wrote a thing! I bet it has value! |
| 2021-05-11 20:43:57 | <shapr> | Taneb: also, if you can look back and wince, you know you've improved! |
| 2021-05-11 20:44:10 | <Taneb> | I was looking at where I was and wincing at the time |
| 2021-05-11 20:44:18 | <shapr> | I wrote fermat's last margin and it got me a Haskell job, two years later that code looks so nasty |
| 2021-05-11 20:44:37 | → | fgaz joins (fgazmatrix@gateway/shell/matrix.org/x-ldjqbjnkddlzxldw) |
| 2021-05-11 20:44:47 | × | Kaiepi quits (~Kaiepi@47.54.252.148) (Remote host closed the connection) |
| 2021-05-11 20:45:05 | → | beardhatcode joins (robbertbea@gateway/shell/matrix.org/x-cfdagqzyyobujzsa) |
| 2021-05-11 20:45:08 | → | mjlbach joins (atriusmatr@gateway/shell/matrix.org/x-ckuqwldaelnnpcjm) |
| 2021-05-11 20:45:10 | → | Kaiepi joins (~Kaiepi@47.54.252.148) |
| 2021-05-11 20:45:19 | <Rembane> | Uniaika: Now when you say that, can we turn any program into a program that only consists of immutable data, composition and recursion? |
| 2021-05-11 20:45:25 | → | unclechu joins (unclechuma@gateway/shell/matrix.org/x-uykkngjnnorqmldi) |
| 2021-05-11 20:45:35 | → | jlv joins (jlvmatrixo@gateway/shell/matrix.org/x-zyjyjmkduzwviwai) |
| 2021-05-11 20:45:40 | → | srid joins (sridmatrix@gateway/shell/matrix.org/x-qnewniuqanmbhoms) |
| 2021-05-11 20:45:41 | <shapr> | Rembane: what about the SSA transform? |
| 2021-05-11 20:45:49 | → | joe[m]4 joins (joemonoida@gateway/shell/matrix.org/x-qjapyfooxpskbrvm) |
| 2021-05-11 20:45:55 | → | bitonic joins (bitonicmat@gateway/shell/matrix.org/x-gzxyabnifphublbd) |
| 2021-05-11 20:45:56 | → | lambdaclan joins (lambdaclan@gateway/shell/matrix.org/x-oiuqiomezalbjuiy) |
| 2021-05-11 20:46:03 | <shapr> | I have the vague impression that SSA is one way to turn an imperative program into a functional-ish program |
| 2021-05-11 20:46:34 | <shapr> | so I wonder if static single assignment would let you get the same benefits? |
| 2021-05-11 20:46:49 | → | alexfmpe joins (alexfmpema@gateway/shell/matrix.org/x-xsmftrlsukuvngtm) |
| 2021-05-11 20:46:55 | <Rembane> | shapr: Hm... yes! That's a good transform. Are there any of these machines around? I want to do experiments! |
| 2021-05-11 20:47:00 | <Uniaika> | Rembane: truth be told, I'm more and more into making my programs interpreters for an eDSL |
| 2021-05-11 20:47:04 | <Uniaika> | so that I can implement https://twitter.com/thingskatedid/status/1386077306381242371 |
| 2021-05-11 20:47:10 | <Uniaika> | this would be fucking rad |
| 2021-05-11 20:47:12 | → | jeffcasavant[m] joins (jeffcasava@gateway/shell/matrix.org/x-yddhjhigtppehlja) |
| 2021-05-11 20:47:42 | → | loyon joins (loyonmatri@gateway/shell/matrix.org/x-pzuovwsjgvbvvmqy) |
| 2021-05-11 20:47:55 | → | michaelpj joins (michaelpjm@gateway/shell/matrix.org/x-zqgjpvmridqqtfur) |
| 2021-05-11 20:48:10 | → | M66bdd7f1f556[m] joins (m66bdd7f1f@gateway/shell/matrix.org/x-xwowgliunifnmlvg) |
| 2021-05-11 20:48:11 | × | jgt_ quits (~jgt@92-247-237-116.spectrumnet.bg) (Ping timeout: 240 seconds) |
| 2021-05-11 20:48:15 | → | hyiltiz joins (~quassel@31.220.5.250) |
All times are in UTC.