<adrien>
first component of your tuple is the mean?
<thelema>
adrien: I started using BatFloat.compare, switched to Pervasives.compare, and am currently using a custom comparison to try to eliminate any possible errors in compare
<thelema>
first component of the tuple is the group, second is the mean (time)
<adrien>
and is the output unsorted or in reverse?
<adrien>
hmmm, ok, unsorted
<thelema>
not same as input order, not reversed
<adrien>
tried printing right after the call to List.sort?
<thelema>
no, I guess I can try that. I'll admit that maybe there's something broken in List.group
wtetzner has joined #ocaml
<thelema>
thanks for the help - the output is sorted properly. Apparently List.group is breaking something
<adrien>
:-)
<thelema>
ah... :( List.group starts with re-sorting the list.
* thelema
makes a proper comparison function
struktured has joined #ocaml
<thelema>
(pmap intersect,merge-based intersect) @86.48 us(84.69 us,88.91 us) is 53.5% faster than next
<thelema>
(filter-based intersect) 186.08 us
<thelema>
Is this type of output reasonable for a benchmark of three intersection functions?
ttamttam has quit [Ping timeout: 256 seconds]
Anarchos has quit [Quit: Vision[0.9.7-H-090423]: i've been blurred!]
<mfp>
thelema: at 1st sight, one would think there are only 2 functions being benchmarked
<thelema>
true - I'm cleaning up the () so they only appear in the case of multiple functions having the same mean
<mfp>
then you wonder how to read the 1st line: if there's 2 functions, pmap intersect and merge-..., why are there 3 times there? @86.48 us(84.69 us,88.91 us)
<thelema>
The last two times are the 95% Confidence Interval
<mfp>
or are those upper-lower bounds?
<mfp>
k
<thelema>
I'm currently leaning towards not showing those
<thelema>
I added them originally for debugging (see above re: List.sort)
<thelema>
[pmap union; merge-based union; merge-unsafe union] @84.05 us is 51.3% faster than
<thelema>
fold-based union 172.76 us
<thelema>
Here's where I'm at now.
<thelema>
fwiw, there's a progress indicator as each function is measured and analyzed
<mfp>
nice :)
<thelema>
It's just a print w/ %! that the measurement has started, and another that the analysis has started
<thelema>
I guess I can put a progress meter 1..100% complete
<thelema>
although I'd have to assume terminal escape codes...
<mfp>
wouldn't that interfere with the timings, though?
<thelema>
not really - I'm doing the timings in batches, with the batch size auto-calibrated based on timer cost and resolution and the time it takes to run the function once.
<mfp>
sounds great
<thelema>
so the number of batches is usually reasonable, and I can do printing in between batches
<mfp>
are you also doing graphs with the p. dist. function and such, like that Haskell lib?
<mfp>
ah yes, -> criterion
<thelema>
no graphs - any recommendations on a graphing lib?
<mfp>
useful for blogging purposes if anything ;)
<thelema>
quite true
<mfp>
hmm there was some lib to output via gnuplot, wasn't there?
* mfp
searches the forge
<thelema>
if you've not used anything, no worries.
<mfp>
not used, but IIRC I downloaded it once :)
<thelema>
that almost counts.
<thelema>
I think someone here on IRC was working on plotting - I'll search the IRC logs to find them
<thelema>
hmm, why are all the example plots in reverse video...
<mfp>
libplplot-ocaml is available in Debian
<thelema>
I'll see how easy it is to odb-ify.
<adrien>
there's plplot and archimedes
<adrien>
archimedes is using oasis but is young (but under development)
<thelema>
adrien: yes, archimedes...
<thelema>
I don't need anything fancy (I think)
<adrien>
when I used it during the summer, simple plotting in archimedes was trivial; more advanced use was still a bit rough
<thelema>
I like trivial
raichoo has quit [Quit: Lost terminal]
Smerdyakov has joined #ocaml
<hcarty>
thelema: If you get get the distributions in array or function form then you the end user can use whichever plotting library they like
<hcarty>
I tried to make PLplot's OCaml interface easy to use (ex. Quick_plot.func [sin; cos] (-3.14, 3.14);). But Archimedes is probably easier to get up and running since it's available through odb.
<thelema>
hcarty: float array
<thelema>
hcarty: I need to get a findlib-enabled lablgtk re-installed
<thelema>
grr, stupid https error on ocamlcore
Boscop has quit [Ping timeout: 260 seconds]
edwin has quit [Remote host closed the connection]
arubin has quit [Quit: arubin]
Morphous has quit [Ping timeout: 248 seconds]
lamawithonel has joined #ocaml
ztfw has quit [Remote host closed the connection]
Morphous has joined #ocaml
lamawithonel has quit [Read error: Connection reset by peer]
lamawithonel has joined #ocaml
ygrek has quit [Ping timeout: 248 seconds]
sepp2k has quit [Read error: Connection reset by peer]
dsheets1 has joined #ocaml
sebz has joined #ocaml
dsheets has quit [Ping timeout: 260 seconds]
sebz has quit []
Smerdyakov has quit [Quit: Leaving]
Cyanure has quit [Ping timeout: 240 seconds]
Cyanure has joined #ocaml
mnabil has quit [Remote host closed the connection]
Cyanure has quit [Remote host closed the connection]