ChanServ changed the topic of #picolisp to: PicoLisp language | Channel Log: https://irclog.whitequark.org/picolisp/ | Picolisp latest found at http://www.software-lab.de/down.html | check also http://www.picolisp.com for more information
orivej has joined #picolisp
<m_mans> Hi all!
mtsd has joined #picolisp
orivej has quit [Ping timeout: 248 seconds]
mtsd has quit [Ping timeout: 240 seconds]
<beneroth> hi m_mans
rob_w has joined #picolisp
rob_w has quit [Changing host]
rob_w has joined #picolisp
miskatonic has joined #picolisp
<cess11_> Good morning.
<Regenaxer> Hi cess11_
miskatonic has quit [Quit: ERC Version 5.3 (IRC client for Emacs)]
<beneroth> hi cess11_, hi Regenaxer :)
<Regenaxer> Hi beneroth
<beneroth> sounds like ARM64 servers will come... RedHat supports them now, too
<Regenaxer> Good thing
<Ober> who needs cache coherency?
<Ober> check the memory and lru every ctx
<beneroth> Ober, ?
<tankfeeder> ~75% faster
<tankfeeder> Regenaxer: ^^^
<tankfeeder> but primeseq is monster
<tankfeeder> now
<Regenaxer> Looks wrong
<Regenaxer> the 'make' result is not used
<Regenaxer> I would do (link 2) and (link 3)
<tankfeeder> because prime? should use Lst in for D
<Regenaxer> then always step 2
<Regenaxer> yes
<Regenaxer> But I though this is the returned list anyway
<Regenaxer> you build the list of primes
<Regenaxer> *and* use the list so far for trial divisions
<Regenaxer> very similar as it was before
<Regenaxer> Do you still have yesterday's version?
<Regenaxer> Before, you divided by 3, 5, 7, 9 right?
<tankfeeder> yesterdays version of which function?
<Regenaxer> Now just by 3, 5, 7, 11 etc
<Regenaxer> all
<Regenaxer> The one where I suggested to divide only by primes, not all odd numbers
<tankfeeder> i dont have prev versions, but i can rewrite
<Regenaxer> it was much simpler, yes
<tankfeeder> there is small number of changes
<Regenaxer> The 'prime?' function is fine, but does not need (or (= N 2)
<Regenaxer> Without the (queue 'R I) it should also be faster
<Regenaxer> I think I understand
<Regenaxer> you do that for we "from" value
<Regenaxer> ok
aw- has joined #picolisp
<Regenaxer> So I would just remember the value of (made) for the result
<Regenaxer> (link) concatenates
<Regenaxer> so you don't need R
<Regenaxer> and no 'queue'
<aw-> beneroth: source of arm64 servers news?
<tankfeeder> so we discuss only primeseq, write ?
<tankfeeder> Primes between 100 and 150: (101 103 107 109 113 127 131 137 139 149)
<tankfeeder> Number of primes between 7700 and 8000: 31
<tankfeeder> i need it for this output, required for task
<Regenaxer> yes, only primeseq, but prime? can omit the 'or'
<Regenaxer> cause we don't test for '2', right?
<Regenaxer> "need it for this output": Yes, so remember the start value (made)?
<Regenaxer> after (while (> A I)
<Regenaxer> (setq R (made))
<Regenaxer> Then continue to (link), and finally return R as now
<Regenaxer> Can omit the two (queue 'R I) calls
<Regenaxer> as the result is built with (link ..)
<beneroth> aw-, heise writes that SUSE and Canonical also support ARM on enterprise/server contracts
<beneroth> so I guess they see a marketing for ARM64 servers :)
<beneroth> s/marketing/market
<aw-> beneroth: interesting, thanks
<tankfeeder> Regenaxer:
<tankfeeder> after (while (> A I)
<tankfeeder> (setq R (made))
<Regenaxer> yes
<tankfeeder> i dont need numbers before 100
<Regenaxer> hmm
<tankfeeder> only after
<Regenaxer> must be the next one
<Regenaxer> true
<tankfeeder> and before 150
<Regenaxer> It must be the *next* one, and then (setq R (tail 1 (made))) :)
<Regenaxer> ie the next cell
<Regenaxer> A bit nasty
<Regenaxer> ha!
<Regenaxer> Remember the length!
<Regenaxer> (setq R (length (made)))
<Regenaxer> return (nth (make ...) R)
<Regenaxer> (or (dec R)) to be exact
<Regenaxer> something like that, you know what I mean ;)
<Regenaxer> I think it will be faster than the queuing stuff, though not simpler
<Regenaxer> (nth (make ... (setq R (length (made))) ... (link)) (inc R))
<beneroth> aw-, you're welcome
<tankfeeder> prime? rewrite gives ~5% faster
<Regenaxer> ok
<tankfeeder> now primeseq
<Regenaxer> good
<tankfeeder> also i should avoid (last (take N)))
<Regenaxer> 'get' ?
<tankfeeder> maybe another version like take but takeN
<tankfeeder> something like (setq @@ (link I)) return @@
<tankfeeder> above doest help, so last is ok then
<Regenaxer> ok
<tankfeeder> new version
<tankfeeder> i dont see it
<tankfeeder> oops
<tankfeeder> wrong
<tankfeeder> so
<tankfeeder> first version ~78secs, now ~18secs
<Regenaxer> cool
<Regenaxer> But why wrong?
<tankfeeder> wrong channel
<tankfeeder> last issue
<tankfeeder> "Number of primes between 7700 and 8000: "
<tankfeeder> should 30
<tankfeeder> not 31
<tankfeeder> i will fix it too.
<Regenaxer> good
<tankfeeder> ...
<tankfeeder> 7951 7963 7993 8009)
<tankfeeder> 8009 thats why 31 items in list
<tankfeeder> (inc 'I 2) pass out range, ha
<Regenaxer> :)
<m_mans> btw, I've tested libaio with PicoLisp, seq. and random write cases, no much overhead, it worked fine even with single thread. I used my laptop and USB HDD
<Regenaxer> m_mans, good to hear!
<beneroth> m_mans, cool, thanks for testing!
Regenaxer has quit [Ping timeout: 248 seconds]
aw- has quit [Quit: Leaving.]
Regenaxer has joined #picolisp
<tankfeeder> final version is ready
<tankfeeder> too nasty.
<tankfeeder> but works as should.
<Regenaxer> Cool tankfeeder
<Regenaxer> afp
<tankfeeder> i gonna commit.
<Regenaxer> yes, good
<tankfeeder> new task done.
<tankfeeder> this is last task in Prime category on rosetta
<Regenaxer> Why not (for X 20 ... ?
<Regenaxer> Hmm, I see
<Regenaxer> Only if the 'and' fires, right
<Regenaxer> OK, so fine, isn't it?
<tankfeeder> yea
<tankfeeder> i will commit
<Regenaxer> Good
<tankfeeder> back to this task
<tankfeeder> damn, nasty.
<Regenaxer> T
rob_w has quit [Quit: Leaving]
miskatonic has joined #picolisp
orivej has joined #picolisp
<rick42> beneroth: cool. arm servers. thanks for heads up
alexshendi has left #picolisp [#picolisp]
beneroth is now known as bene|off
miskatonic has quit [Quit: ERC Version 5.3 (IRC client for Emacs)]
tankfeeder has quit [Ping timeout: 250 seconds]
tankfeeder has joined #picolisp