egg|nomz|egg changed the topic of #kspacademia to: https://gist.github.com/pdn4kd/164b9b85435d87afbec0c3a7e69d3e6d | Dogs are cats. Spiders are cat interferometers. | Космизм сегодня! | Document well, for tomorrow you may get mauled by a ネコバス. | <UmbralRaptor> egg|nomz|egg: generally if your eyes are dewing over, that's not the weather. | <ferram4> I shall beat my problems to death with an engineer.
<armed_troop>
I see queries about testing but I'm missing the context :P
<egg>
armed_troop: eh, just preparing for the cohen release
<armed_troop>
ahhhh, ok
<egg>
armed_troop: linux binary got a fair bit bloated because I forgot to ifdef something out on *nix, but other than that I don't think there are any real issue
<egg>
s/$/s
<Qboid>
egg meant to say: armed_troop: linux binary got a fair bit bloated because I forgot to ifdef something out on *nix, but other than that I don't think there are any real issues
<armed_troop>
hmmmm, can just rebuild it without the bloat? or is that too much effort?
icefire has quit [Quit: Leaving]
<egg>
armed_troop: that would entail retagging, redoing the profile-guided optimization thing, etc.
<egg>
too much work, don't care
<egg>
armed_troop: it'll go away the following moon anyway :-p
<kmath>
<KatBarger> We are still accepting applications from graduate students, even though we are passed the deadline. We hope to get… https://t.co/NrQZjuUem4
<egg>
!wpn bofh
* Qboid
gives bofh a ytterbium duck which strongly resembles a circle
<egg>
!wpn whitequark
* Qboid
gives whitequark a ventral trampoline ordinal
<bofh>
!wpn egg
* Qboid
gives egg a cranky octagon
tawny has quit [Ping timeout: 182 seconds]
<egg>
bofh: do you want to test cohen on linux?
icefire has joined #kspacademia
* UmbralRaptor
wonders if MNRAS is just bad about making published data easy to use.
* UmbralRaptor
🔪 things, and wonders if no one has done an all sky stellar parameter catalog from DSS, IRAS, 2MASS, etc.
tawny has joined #kspacademia
<UmbralRaptor>
Hrm. Anything hotter than 2400 K would be in the Rayleigh Jeans tail for J-band…
* UmbralRaptor
pokes the RAVE catalog with a VizieR.
<egg>
bofh: phl's algorithm for the number of ice cubes to put in the shaker: take the number of people, add one, and take the number of regions of the plane separated by that number of lines in general position
<egg>
UmbralRaptor: hm, is cohen the dog release?
<bofh>
egg: seems reasonable, tbh.
<egg>
bofh: cohen is going to be the first release which falls on a new year
<UmbralRaptor>
egg: dog?
<egg>
UmbralRaptor: ^
<egg>
UmbralRaptor: new moon, so sometimes it falls on the new year of a lunisolar calendar :-p
<kmath>
<thehill> #BREAKING: Trump considering privatizing International Space Station, turning it into real estate venture: report… https://t.co/F2fJLiU4s6
<egg>
bofh: UmbralRaptor: So the january 2023 release of principia will be a kitty
<egg>
bofh: it's 3d vectors, and "intrinsics" means we wrote the vector ops with SSE2 vector intrinsics (otherwise MSVC is nonvectorizing)
<bofh>
I am absolutely unsurprised there's negative benefit for the degree 4 case (also use unaligned loads if stuff is crashing!).
<egg>
bofh: no it's not that it crashes, it's that IACA crashes
<egg>
bofh: and the benefit is positive, albeit smol
<egg>
bofh: also phl might have swapped some numbers
<egg>
bofh: fixed
<egg>
bofh: IACA crashes and in the process shows that it's written in go of all things >_>
<bofh>
wait, *IACA* crashes? lmfao.
<egg>
bofh: :D
<egg>
bofh: also phl is now having fun optimizing 3d != and ==
<egg>
bofh: and you appear to be an auroch
<egg>
does that mean you can be observed with the ANBOscope
<bofh>
3d != and == are prolly most optimized down to a single cmpps
<bofh>
also, wait, when did I become an auroch?
<egg>
bofh: well there's an anonymous auroch on the google sheet
<egg>
so I guess that's you
<egg>
bofh: and nah, they're not optimized much by MSVC apparently
<bofh>
I mean by hand :p
<egg>
bofh: phl went with _mm_movemask_pd (and then eventually a bitant and a cmp)
<egg>
bitand even
<bofh>
oh right, I kerp forgetting you're working with doubles
* egg
slaps bofh with a double trout
<egg>
bofh: ah so there's the failure mode where IACA crashes and starts talking about goroutines, but there's also the one where it starts looping forever (it does that on the unoptimized Dot) :-p
<bofh>
rofl
<bofh>
incredible.
<bofh>
I would honestly just benchmark those routines using RDTSC & a loop tbh.
<egg>
bofh: well we have benchmarks for the polynomials
<egg>
bofh: but since we're poking at intrinsics it seems fun to tinker with IACA