The h Bar - 2014-02-15

« first day (1199 days earlier) ← previous day next day → last day (3732 days later) »

1:29 AM

Looking at my contributions, not particularly spectacular at all - 5 answers (one will no doubt be deleted soon physics.stackexchange.com/questions/95306/… at a loss of 27 rep [edit+upvote+accept]), an dthe only answer to my questions was due to me putting a bounty on it.... not good

@Amaterasu Are closed questions deleted?

@KyleKanos i would imagine so, they are often voted for deletion... so now, i will be losing 37 when it does

@Amaterasu Hmm, this search shows closed but not-deleted posts from as far back as 2010

yes, but the recent discussions seemingly aimed at homework questions have me quite convinced that the question (hence my answer and associated rep) will be nuked

That discussion is more about what constitutes a homework question than nuking closed homeworks

1:54 AM

well, to save any disappointmnt, I am seeing my rep for physics at 291, instead of the 328 stated

i am toying with dleting my answer there altogetehr

@Amaterasu Don't!

Until it's actually deleted (which I am doubting will happen), you still have that rep

regardless, i am toally ignoring that tag now

More. If you retain the rep for six months you get to keep it even if the question is deleted.

(I think it is six months. There is some magic period.)

that is a long way off

Yep, it's six months. See meta.stackoverflow.com/a/5222/2509 as long as you have at least +3 on the post.

2:02 AM

ironically the 2 particular areas of physics that I am at least marginally proficient in are atmospheric physics (PhD in it, but little interest in it here) and Physics education (inc the maligned homework tag - but also where there is little interest)

With the +3 requirement you have a problem right now.

i can't do anything about it

@Amaterasu Why are you so care about the rep in this site? I don't think you have strong interest in it

it is often to have rep change +-100 by rep recalcuate

nevermind

i am just hoping to get an answer for my latest question

@hwlau i am actually concerned about my contributions to the site

It is fair.

Usually, If the question is really that bad that need to be deleted, then I will just let it go. I will just move on and answer other questions.

2:17 AM

i can't answer most questions

i am specialised in 2 fields that are profoundly unpopular here

atmospheric physics (where most of my questions are based - and seemingly only answered if I put a bounty on it) and physics education (which include the maligned homework, reference-request, specific-reference, education tags)

also @hwlau I only have 291 (effective) rep

@Amaterasu Ok, I guess it may be better for you to promote the Earth science proposal here. Then you can have a better community to ask and answer questions

i have

How do you promote it?

in my profile, not much else I can do

How about talk to some students and prof in your adjuct university?

2:28 AM

143

Proposed Q&A site for those interested in the geology, meteorology, oceanography, and environmental sciences.

Currently in commitment.

they are not interested (time)

@Amaterasu Have you looked into that?

@KyleKanos yes, I am committed to it

reason?

Okay

2:29 AM

but, it is approaching 2 years for that site

@Amaterasu Possibly time consuming fix for that: contact Atmospheric faculty at PhD-granting institutes to get their grad students in on it

@Amaterasu I remember they change the rule that the definition and commission must be finished 1 year.

then the proposal is in trouble

3

Q: What is the scope of this site?

Proposal: Popular Science As the proposal is called Popular Science, would questions looking for answers at a popular level say about life sciences be on topic too? Are popular questions about all sciences welcome?

science proposal-scope

That's kinda funny

@Amaterasu No, less than half year past and there it already reach 71%. The steady grow of the commitment suggests that it is like success in 4 months.

2:38 AM

hope so

2:49 AM

This doesn't make sense. Copying a 1D slice of the 2nd dimension of a 4D array takes 17 seconds (out of 791 total seconds run time), but coping that 1D slice back into the 4D array takes half that time!

I would not know

3:10 AM

the stupid irony is that I am getting introduced as a physicist all over the place... when it is not quite like that - I have a PhD in physics (a very narrow field), but are just a school teacher (and weekend-independent-researcher)

4:07 AM

@KyleKanos Snow/ice, then earthquakes... the South might be the start of the rapture

@KyleKanos That's interesting -- it could be a caching issue

But if you can rewrite it so it's not doing anything on the 2nd dimension that would obviously be better!

@KyleKanos Have you ever compiled with -check arg_temp_created turned on?

3 hours later…

7:40 AM

@KyleKanos the issue is now fixed - flagged it and got it deleted - no point keeping the answer if it's going to be deleted anyway - now, I am ignoring the 'homework' tag

now the other issue is to actually get an answer to a question without having to put a bounty on it to get it...

1 hour later…

9:09 AM

Actually, as the edit 2 rep is also gone - the question was deleted....had to put a bounty on my question, as that is the only way it seems to be considered worthy enough to be answered

7 hours later…

3:51 PM

Watch Phys.SE users take over English.SE here:

5

12

Q: What does 'infinitesimally small' mean?

If infinite is the opposite of infinitesimal, and small is the opposite of large, then: infinitely large ---------- Means "very large" infinitely small --------- Means "very small" infinitesimally large ---- Means "very small" infinitesimally small --- Means "very large" ??? However my interp...

adjectives adverbs negation double-negation

1 hour later…

5:13 PM

50 day commit streak!

@KyleKanos It was funny watching the example questions build up with their idea of popular science questions (as someone who gets regularly poked by laypeople about such things, they're completely off the mark with the example qs)

1 hour later…

6:16 PM

@Qmechanic Daw, I miss Ben's answers here

@tpg2114 Nope, but I am wondering if that might cause issues

3 hours later…

8:58 PM

Heeyyyy. I have 603 votes on questions on 1078 votes, shouldn't that give me the Electorate badge?

@KyleKanos me too :(

Glen The Udderboat

@KyleKanos Why are you concerned with badges?

@GlenTheUdderboat I like shiny gold ones.

Glen The Udderboat

@KyleKanos I thought you weren't interested in the Olympics.

@GlenTheUdderboat I'm not. But I do like my internet badges

Glen The Udderboat

9:11 PM

@KyleKanos It appears you have your 'Electorate' badge. Happy?

user image

Yes, there it is!

Took like 40 minutes

@KyleKanos Most of the time our codes run slow, it's because we're slicing arrays that need to create copies

We had to touch almost every single line of code once to change our variable ordering in an array to avoid slicing

@tpg2114 Got it. I've found that division is sloooooooooow

And I should be able to reduce one of those too.

I have a small loop (4 iterations max) that I do a division inside each time when the value I'm dividing by is outside the loop

So I made a new variable that is rati = 1d0/rat and I'm VTune'ing it now to see if I gain anything

We got a lot out of that too

1d0... making me sad :/

We define an integer parameter in a global file, wp (working precision)

So then all of our constants are 1.0_wp

All of our declarations are REAL(KIND=wp)

And we can change one line and run in single, double, or quad

For whatever reason, the developers used qprec even though it's not quad precision

9:21 PM

@KyleKanos badges are awarded by a periodic script

@ManishEarth Horribly inefficient!

@KyleKanos ...the alternative is running expensive db queries every minute :p

@ManishEarth Way more efficient

though SE has the hardware to support that, actually. They only absolutely need like 10% of what they have

nickcraver.com/blog/2013/11/22/…

@tpg2114 Well that was awesome....I dropped a line from 12.815 s to 2.753 seconds

Glen The Udderboat

9:32 PM

@ManishEarth Has there been a proposal to ditch the badges?

And somehow, s=-rat+tfn/ufn went from 13.321 s to 0.529s in this run....that's weird

@GlenTheUdderboat Why would they do that!?

Glen The Udderboat

@KyleKanos Because nobody (present company excluded) cares?

@GlenTheUdderboat I'd argue that badges, much like reputation, represent an answerers level of trust.

@KyleKanos Chocolates might not be enough for VTune now. You might need to put a ring on it

@KyleKanos Not that weird if the memory it was accessing is no longer an array temporary or something, if the two changes are linked. Or it could also just be batshit crazy

Or the compiler might be able to optimize better

Compilers == voodoo

Compilers are magic

2 hours later…

11:37 PM

Felix von Leitner's presentation on compiler optimization performance from 2009 (PDF link) is getting a bit dated, but is worth reading.

Notice the text on slide 41: "gcc is smarter than the video codec programmer on all platforms."

That's gcc 4.3, BTW. Presumably it is still smarter these days.

« first day (1199 days earlier) ← previous day next day → last day (3732 days later) »

Transcript for

Feb '1415

The h Bar

General chat for Physics SE (physics.stackexchange.com). For M...

join this room
about this room

00:00

06:00

12:00

18:00

all times are UTC

site design / logo © 2024 Stack Exchange Inc; legal

mobile