Newest at the top
| 2024-11-14 17:21:27 +0100 | <geekosaur> | llvm still lacks support for pre-CPSed code |
| 2024-11-14 17:20:48 +0100 | aljazmc | (~aljazmc@user/aljazmc) aljazmc |
| 2024-11-14 17:19:31 +0100 | <tomsmeding> | :) |
| 2024-11-14 17:19:01 +0100 | tromp | (~textual@92-110-219-57.cable.dynamic.v4.ziggo.nl) |
| 2024-11-14 17:16:39 +0100 | <EvanR> | ok |
| 2024-11-14 17:16:35 +0100 | <tomsmeding> | EvanR: it definitely is not |
| 2024-11-14 17:16:16 +0100 | <Inst> | probably MY skill issue :( |
| 2024-11-14 17:16:14 +0100 | <EvanR> | is llvm not the default now anyway |
| 2024-11-14 17:14:34 +0100 | <bailsman> | Inst: I compiled my benchmark with -O2 -fllvm. Does not seem meaningfully different. Is -O2 the wrong optimization level? |
| 2024-11-14 17:12:36 +0100 | <Inst> | try compile with -fllvm |
| 2024-11-14 17:12:30 +0100 | <lambdabot> | Unknown command, try @list |
| 2024-11-14 17:12:30 +0100 | <Inst> | @bailsman |
| 2024-11-14 17:11:19 +0100 | <EvanR> | in any case idiomatic haskell is a starting point for getting into the weeds for optimization |
| 2024-11-14 17:10:34 +0100 | <EvanR> | not necessarily, sometimes idiomatic haskell is faster |
| 2024-11-14 17:10:03 +0100 | <bailsman> | If you write idiomatic haskell, you get as-slow-as-you-would-expect, if you try to write in-place code, you get way-slower-than-you-would-expect. |
| 2024-11-14 17:09:25 +0100 | <EvanR> | in the case of arrays, for lookup tables |
| 2024-11-14 17:09:22 +0100 | <bailsman> | I agree with your conclusion - stop trying to be clever and just learn what idiomatic haskell code looks like. |
| 2024-11-14 17:08:53 +0100 | <EvanR> | but as a looping mechanism |
| 2024-11-14 17:08:44 +0100 | <EvanR> | in the case of list, usually not as a data structure |
| 2024-11-14 17:08:09 +0100 | <EvanR> | list and arrays in haskell are both good for certain purposes |
| 2024-11-14 17:07:25 +0100 | <EvanR> | "they are both called list" isn't that inspiring |
| 2024-11-14 17:06:59 +0100 | <EvanR> | the C version of linked list is just a bad thing to compare to haskell list unless you are careful to emulate what the haskell version did |
| 2024-11-14 17:06:20 +0100 | <bailsman> | To me the fact that the Haskell Vector is ~100ms, Haskell map is ~25ms, C allocate-new-linked-list-and-copy version is ~15ms, C array in place is ~2ms is suggestive of the fact that indeed allocating a list is slow, and it's indeed what Haskell is doing, but it's still better than trying to do an array in Haskell. |
| 2024-11-14 17:06:10 +0100 | <EvanR> | before claiming stuff about what stuff compiles to you should check it |
| 2024-11-14 17:05:01 +0100 | <EvanR> | but the specific reasons are off |
| 2024-11-14 17:04:50 +0100 | <EvanR> | "straight list processing and immutable structures are probably better in haskell than C-like mutable array munging" though is what I've been saying for days |
| 2024-11-14 17:03:44 +0100 | <geekosaur> | gc only gets involved when the pointer reaches the end of the nursery |
| 2024-11-14 17:03:31 +0100 | <geekosaur> | the nursery/gen 0 is a bump-pointer allocator |
| 2024-11-14 17:03:20 +0100 | <geekosaur> | not magic |
| 2024-11-14 17:03:13 +0100 | <bailsman> | I agree - I'm not really sure. Some GC magic probably. But the point is that it's builtin and optimized, so it's much faster than trying to emulate in-place updates, which compiles to a morass of work and not 5 asm instructions like the c version. |
| 2024-11-14 17:02:06 +0100 | <geekosaur> | because it's probably not what actually happens |
| 2024-11-14 17:01:56 +0100 | <geekosaur> | bailsman, what do you think is going on during an allocation? |
| 2024-11-14 17:01:22 +0100 | <EvanR> | yes it is |
| 2024-11-14 17:01:08 +0100 | <bailsman> | No it isn't. |
| 2024-11-14 17:00:57 +0100 | <EvanR> | allocating nodes in haskell is much faster |
| 2024-11-14 17:00:38 +0100 | <EvanR> | are you allocating nodes with malloc |
| 2024-11-14 17:00:31 +0100 | <EvanR> | that's... not going to be an apples to apples comparison |
| 2024-11-14 17:00:30 +0100 | <bailsman> | So I think I'm concluding that map is "the best you can do in haskell" because it's optimized and a builtin, and any attempt to do in place algorithms is just going to be massively slow. |
| 2024-11-14 17:00:06 +0100 | <bailsman> | To test my theory, I wrote a C version of the benchmark. Updating a linked list by allocating nodes one by one and copying over the values takes 14ms, approximately as long as Haskell takes to do map. Updating 1M records inplace in an array takes 2ms. |
| 2024-11-14 16:59:12 +0100 | <geekosaur> | and the tree example is probably closer to your actual AST than a list zipper example would be |
| 2024-11-14 16:58:51 +0100 | <geekosaur> | even if it won't work with your structure as is, the wiki page I pointed you to earlier describes what you can do with a zipper |
| 2024-11-14 16:58:42 +0100 | lortabac | (~lortabac@2a01:e0a:541:b8f0:55ab:e185:7f81:54a4) (Quit: WeeChat 4.4.2) |
| 2024-11-14 16:57:58 +0100 | <geekosaur> | you can do anything to the focused node including remove or replace it, and moving the zipper will reknit the tree |
| 2024-11-14 16:57:30 +0100 | <ph88> | ok cool, thanks geekosaur ! |
| 2024-11-14 16:57:24 +0100 | <geekosaur> | that'd be a zipper |
| 2024-11-14 16:57:24 +0100 | <ph88> | can zipper do this too ? |
| 2024-11-14 16:56:59 +0100 | <ph88> | what if i don't only want to change the variable `a` but i also want to inspect the nodes and modify/replace them ? |
| 2024-11-14 16:56:55 +0100 | <geekosaur> | if you just want something Traversable-style, any generics library will give you that |
| 2024-11-14 16:56:38 +0100 | <geekosaur> | which it sounded like you wanted |
| 2024-11-14 16:56:27 +0100 | <geekosaur> | although the default traversals are all of the Traversable variety, unlike a zipper which lets you move at will |