« first day (4108 days earlier)      last day (918 days later) » 

1:26 AM
^ this
 
1:44 AM
There's probably a truism for "premature generalization" that's as true as Knuth's "premature optimzation" 😉
 
 
9 hours later…
10:52 AM
I wonder though, if it's of any use; you know how you can do something generic, but then it's easier to adapt it to something specific.
 
user92578
11:31 AM
17/19 tests pass, 18 required for full 10 points... yet another task where I got 7 points after ~30 minutes, grind to 9 points took a few hours... Sucks because I'd love to solve this fully, but with a thousand things to do I've kinda ran out of justifications to keep working on this
 
user92578
Annoyingly locally on the uni servers the 18th test passes fast enough, but not on the submission server... However the 17th test there is just below the limit so I'm not really even close to being fast enough for the 18th when it doubles the size of the input
 
2:39 PM
:(
 
2:57 PM
Yikes! What's the assignment problem?
 
 
1 hour later…
user92578
4:13 PM
Greedy algorithm for a set cover
 
user92578
Possibly some shenanigans related to my usage of a priority queue going on... Don't want to rebuild it every iteration, so I reinsert sets when their size changed as pairs (size, set), and pop from the heap until set.size matches the size in the pair
 
user92578
I only process sets that I know have elements that I have just covered, so that shouldn't be a slow down anymore...
 
user92578
Would need to approximately half my runtime for the 18th test to pass, and half twice for all 19 to pass
 
user92578
5:16 PM
Noo I found an unnecessary loop but it wasn't enough speedup
 
Ahhhh, yeah, I remember that one kicking my butt too!
 
user92578
5:32 PM
I feel like I must be just missing some big idea here, all though I think I've tried all the tips provided... Micro-optimizations just won't get me down enough
 
user92578
Wish I could just allow myself to let it go :P
 
user92578
6:14 PM
Lazy constructing of the sets looks promising but I'm still off like 500ms or so...
 
user92578
4700ms for last passing level, scaling factor somewhere a bit above 2
 
user92578
6:27 PM
9944ms locally o_O ...
 
user92578
Time to start rolling on the server gods
 
user92578
no luck, fuck
 
user92578
yah server is juuuust enough slower that this is never going to work
 
user92578
6:51 PM
Okay I think I've got something, reserving vector memory results in a massive speedup, just gotta find the right amount to stay under the memory limits :P
 
user92578
7:02 PM
Fuck this, I'm onto my last submission, the unlimited submission test server is too slow, locally I'm passing easily...
 
user92578
 
user92578
Man what a waste of time
 
user92578
8083ms, not even close to passing the 19th
 
user92578
Solved by literally reserving like 16 pointers to vectors containing owner sets of each vertex
 

« first day (4108 days earlier)      last day (918 days later) »