Advanced search

Message boards : General Topics : Some Stats About You

Author Message
Profile Marius
Project administrator
Project developer
Project scientist
Avatar
Send message
Joined: 29 Jun 15
Posts: 427
Credit: 4,276
RAC: 0
Message 20262 - Posted: 13 Jul 2015, 21:27:31 UTC
Last modified: 13 Jul 2015, 21:28:05 UTC

Hi guys,

One thing we're going to try and do more of is making some informative posts, blog style. This message board software isn't exactly what we need, and we aren't quite set up to do this elsewhere on cosmologyathome.org (yet), so for the moment it's posted on my personal website.

Here's the first post: Stats about Cosmology@Home

It features some anonymous statistics about C@H users which we've been looking at in thinking about how best we can upgrade our software. Enjoy and feel free to comment about it here or on the comments system there.

Profile cykodennis
Send message
Joined: 31 May 10
Posts: 234
Credit: 4,896,378
RAC: 0
Message 20263 - Posted: 14 Jul 2015, 6:18:20 UTC - in response to Message 20262.
Last modified: 14 Jul 2015, 6:54:22 UTC

Yay! \o/

I like the idea with the blog style.

Profile pschoefer
Send message
Joined: 26 Mar 08
Posts: 9
Credit: 1,946,775
RAC: 0
Message 20264 - Posted: 14 Jul 2015, 8:48:14 UTC - in response to Message 20262.

Thanks for the interesting stats.

I think that the OS efficiency comparison is biased towards older OSs. The Win Server 2k8, Vista, and some Linux machines may have been running for a number of years, while Win 8.1 was released just 2 years ago. Hence, the Win 8.1 machines had less time to accumulate credits. It would be better to compare only the credits awarded during the last month, although I don't know if there's an easy way to get this numbers (RAC is not a very good measurement either).

OTOH, Win XP machines had plenty of time to accumulate loads of credits, but they might be significantly less powerful than modern machines. The typical Win 8.1 host might also have a newer CPU than the typical Win 7 host, but CPUs did not become that much faster over the last couple of years, at least as long as new extensions like AVX are not used.

Profile cykodennis
Send message
Joined: 31 May 10
Posts: 234
Credit: 4,896,378
RAC: 0
Message 20265 - Posted: 14 Jul 2015, 13:07:37 UTC - in response to Message 20264.

Hi,

i think it is related to the pattern of use of such machines. In C@H with its few checkpoints, it helps when the computer runs for long times without interruption.

I guess the amount of 24/7 machines running W2k8R2 or Linux is much higher, than of 24/7 running W8.1.
So, the most boincing W8.1 in 2015 are workstations, which will be shutdowned, rebooted - in short, everyday use - which can set a C@H W back to its last checkpoint.

Marius BOINC/Docker development will probably change this.

Profile Marius
Project administrator
Project developer
Project scientist
Avatar
Send message
Joined: 29 Jun 15
Posts: 427
Credit: 4,276
RAC: 0
Message 20266 - Posted: 14 Jul 2015, 17:57:15 UTC

Thanks. Both very good points I overlooked. The "total credits" is indeed from forever, as opposed to from the last month, which gives an advantage to older OS's, as well as likely to the 32 bit hosts. And the check pointing thing is interesting too. I may add an addendum to post addressing these. At the very least its definitely possible to pull out credits from just the last month from the DB.

Peter Stanton
Send message
Joined: 22 Sep 08
Posts: 2
Credit: 14,515
RAC: 0
Message 20267 - Posted: 17 Jul 2015, 2:28:04 UTC - in response to Message 20262.

I'm happy to see an update on the project. That really reassures us volunteers that the project is still alive. I honestly thought it was dead, but I think I'll allow it to get new tasks again now. :)

Profile Marius
Project administrator
Project developer
Project scientist
Avatar
Send message
Joined: 29 Jun 15
Posts: 427
Credit: 4,276
RAC: 0
Message 20268 - Posted: 17 Jul 2015, 6:59:37 UTC - in response to Message 20267.

Great to hear. Ha I appreciate the honestly, pretty natural conclusion to reach given our silence lately. Anyway, work continues to provide more substantive updates as we speak...

ritterm
Send message
Joined: 30 May 08
Posts: 25
Credit: 896,913
RAC: 0
Message 20269 - Posted: 17 Jul 2015, 19:35:15 UTC - in response to Message 20267.

Peter Stanton wrote:
I'm happy to see an update on the project. That really reassures us volunteers that the project is still alive...

+1

Marius wrote:
...work continues to provide more substantive updates as we speak...

Good to hear, Marius, thank you.

So, can volunteers have some assurance that their current efforts are doing real work that's contributing to the science?

Also, your CV notes that you are going to Paris in a few months. Congratulations on that, but does that mean your participation in C@H will end? If so, is anyone lined up to take over?

Cheers,

MarkR
____________

Profile Marius
Project administrator
Project developer
Project scientist
Avatar
Send message
Joined: 29 Jun 15
Posts: 427
Credit: 4,276
RAC: 0
Message 20270 - Posted: 17 Jul 2015, 21:31:34 UTC - in response to Message 20269.
Last modified: 17 Jul 2015, 21:35:20 UTC

> So, can volunteers have some assurance that their current efforts are doing real work that's contributing to the science?

This is something we'll try to do better, but for now a good way to see the impact of results is to look for citations to the original PICO papers (here and here).


> Also, your CV notes that you are going to Paris in a few months. Congratulations on that, but does that mean your participation in C@H will end? If so, is anyone lined up to take over?

Thanks! In fact my C@H involvement will likely improve with my move to Paris, you should look up who the director is of the Lagrange Institute which is where I'll be going :) (hint: its Ben) (granted my position is not tied to C@H in any way, but I'm sure with both of us across the hall from each other will probably really help productivity)

ritterm
Send message
Joined: 30 May 08
Posts: 25
Credit: 896,913
RAC: 0
Message 20271 - Posted: 19 Jul 2015, 19:07:40 UTC - in response to Message 20270.
Last modified: 19 Jul 2015, 19:08:57 UTC

Marius wrote:
...a good way to see the impact of results is to look for citations to the original PICO papers (here and here).


I'm not so sure I can find my way to relevant papers... :-( Can you provide a link to one or more that credits the work of C@H volunteers?

Marius also wrote:
...my C@H involvement will likely improve with my move to Paris...

More good news! :-)
____________

Profile cykodennis
Send message
Joined: 31 May 10
Posts: 234
Credit: 4,896,378
RAC: 0
Message 20272 - Posted: 19 Jul 2015, 22:52:32 UTC - in response to Message 20271.

Links as you've asked for, do not exist afaik.

The reason for it is the same, as like a with a guitarist who may credit the brand of his guitar, but not the origin of some particular parts.

As Marius made once more clear, C@H is a project which teaches a software application to deliver better results. If you want to see the real impact of the crunched work of C@H, you have to look for PICO, not for C@H.

The links which he has provided to you show, that PICO is still in use even today, for data analysis.

Profile Marius
Project administrator
Project developer
Project scientist
Avatar
Send message
Joined: 29 Jun 15
Posts: 427
Credit: 4,276
RAC: 0
Message 20273 - Posted: 22 Jul 2015, 19:19:36 UTC

Yea sorry I think my links were slightly confusing. Like cykodennis said, the software that uses the data calculated on C@H is called PICO. There have been two papers written about this software, the second one here talks some about C@H (C@H didnt exist when the first paper was written). These are old papers, but the software is up to date, and is being actively used today. When a group uses the PICO software for some results, they cite one or both of those original papers. The links I sent were a list of such citations so you can see the activity of cosmologists using PICO throughout the years.

Profile C@H Sceptic
Send message
Joined: 23 Jan 15
Posts: 17
Credit: 101,772
RAC: 0
Message 20278 - Posted: 11 Aug 2015, 17:35:20 UTC - in response to Message 20262.
Last modified: 11 Aug 2015, 17:44:39 UTC

This message board software isn't exactly what we need, and we aren't quite set up to do this elsewhere on cosmologyathome.org (yet), so for the moment it's posted on my personal website.


Okay, lets mend the message board, after all its been broken for 10 years.

    1 Why does Ben only ever post "its all wonderful" every 2 years when clearly it isnt.
    2 The server has been Broken from start. If for any reason a BOINC job isn't completed, it is unlikely to be re-run. This means that Millions of jobs are logged as "null result" - has this wrecked your science? We'd like to know.
    3 Blog stuff about how many people use Windows XP and in which continent they do it in isn't really the answer.
    4 Why haven't you actually run a BOINC job yourself for over a month? This doesn't show a lot of commitment.

Profile Marius
Project administrator
Project developer
Project scientist
Avatar
Send message
Joined: 29 Jun 15
Posts: 427
Credit: 4,276
RAC: 0
Message 20279 - Posted: 11 Aug 2015, 19:20:14 UTC - in response to Message 20278.

Hi C@H Sceptic,

1. Communication hasn't been the best, there's no denying that. We're working to improve. I check the boards daily. Ben has many other responsibilities that prevent him from directly interacting on here as often as I'm sure he would like.

2. Losing any specific job is totally OK science-wise for us. We build up the jobs by randomly sampling the cosmological parameter space, its the aggregate of all the results that is the useful quantity, not any one particular job. I do appreciate there's alot of these types of questions that people have that are not readily answered except by digging through the forums, reading the very technical papers on PICO, or asking us directly like you have. To that end I'm planning on describing the whole process in a future blog post that's hopefully a lot more accessible.

3. Sure, if that post were all we were going to do, it wouldn't be enough. Fortunately its just a start.

4. Fair enough, in my defense I've been running plenty of test jobs in developing the new server, something which you guys will hear more about soon :)

Profile Phil1966
Send message
Joined: 10 Jul 13
Posts: 26
Credit: 3,212,542
RAC: 0
Message 20567 - Posted: 8 Nov 2015, 10:00:58 UTC - in response to Message 20279.

Nice to see the reawakening of the project + the forum :)

Message boards : General Topics : Some Stats About You