U41.xx Observation Thread

[B^S] Elphidieus
[B^S] Elphidieus
Joined: 20 Feb 05
Posts: 222
Credit: 12,347,408
RAC: 0

RE: It kinda pisses me

Message 30146 in response to message 30145

Quote:

It kinda pisses me off that I and others have been wasting cpu cycles by using 'poorly' programmed clients. The same thing happened on SETI. How can the project directors be so careless (by not distributing better clients) and insensitive (to the fact they are wasting cpu resources, energy, and our emotional mojo).

poorly programmed....? I don't think so....

The standards clients are meant for the general computational processes minus those different instruction sets on the widely varied CPUs in the market. So do you expect the project directors to waste time and resources in order to programme numerous clients to be CPU- and instruction-specific....?

pardon my poor english....

Akos Fekete
Akos Fekete
Joined: 13 Nov 05
Posts: 560
Credit: 4,510,638
RAC: 0

RE: RE: It kinda pisses

Message 30147 in response to message 30146

Quote:
Quote:
It kinda pisses me off that I and others have been wasting cpu cycles by using 'poorly' programmed clients. The same thing happened on SETI. How can the project directors be so careless (by not distributing better clients) and insensitive (to the fact they are wasting cpu resources, energy, and our emotional mojo).
poorly programmed....? I don't think so....

Hi,

I've seen lots of programcodes before it and I can say that the original code was also really good. I think there are other guys who can make much more faster than my codes too. We know that the EAH staff is working on a common, faster application, and they do their best.

TauCeti
TauCeti
Joined: 1 Apr 05
Posts: 16
Credit: 454,993
RAC: 0

RE: RE: It kinda pisses

Message 30148 in response to message 30146

Quote:
Quote:

It kinda pisses me off that I and others have been wasting cpu cycles by using 'poorly' programmed clients. The same thing happened on SETI. How can the project directors be so careless (by not distributing better clients) and insensitive (to the fact they are wasting cpu resources, energy, and our emotional mojo).

poorly programmed....? I don't think so....

The standards clients are meant for the general computational processes minus those different instruction sets on the widely varied CPUs in the market. So do you expect the project directors to waste time and resources in order to programme numerous clients to be CPU- and instruction-specific....?

pardon my poor english....

I am quite sure that all the official project programmers work hard and do their very best to provide the best clients they can. But they are limited in time and - i speculate - there are not so many fulltime programmers for the e@h client at all.

It's just a question of ressources. For me, the akosf-optimizations are a prime example for suboptimal ressource (money) allocation in a project. Millions of dollars are spend (officially and by volunteers) on hardware (infrastructure, clusters) and energy whereas it looks like that the monetary budget for brainpower (skilled programmers for various architectures) falls short.

But i think it's wrong to blame anyone for this. I can imagine, that the huge optimizing potential of the application was not known before akosf started his miracles.

If the akosf-optimizations help to convice the projects money sources that more personal and less hardware is needed to achieve better throughput, it would be a Good Thing(tm) from my point of view.

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,119
Credit: 119,716,173
RAC: 74,237

RE: Wow. Big improvement on

Message 30149 in response to message 30145

Quote:
Wow. Big improvement on my laptop, from 25K to 4K seconds. I'll try it on a regular desktop, too, when I have a chance.


Yes, it is good isn't it?

Quote:
It kinda pisses me off that I and others have been wasting cpu cycles by using 'poorly' programmed clients. The same thing happened on SETI. How can the project directors be so careless (by not distributing better clients) and insensitive (to the fact they are wasting cpu resources, energy, and our emotional mojo).


Dear me.
Pisses. Wasting. Careless. Insensitive. Emotional.
Just goes to show, eh?
You really can learn something new every day ...... :-)
Cheers, Mike.

I have made this letter longer than usual because I lack the time to make it shorter. Blaise Pascal

WayneKFord
WayneKFord
Joined: 21 Aug 05
Posts: 25
Credit: 1,423,427
RAC: 15

Oops. I didn't mean to

Oops. I didn't mean to personalize my criticism, if it was construed that way. Obviously, Mike would like me to choose more politically correct expressions. But that wouldn't make me much fun at dinner parties, would it?

'Poor programming' is suggested when you can get a factor of two or more performance improvement merely by re-writing the code. Why is that debated? I agree with the comment below that suggests more brain power can be more effective than more compute power.

One of the fallicies of these distributed computing projects seems to be that the code is written for the lowest comment denominator. Perhaps that constraint should be relaxed a bit, relative to getting the best performance overall. (concrete example: don't waste resources debugging Win98 implementations and writing generic code to run on pentium-I cpus)

From the SETI project message boards, I've decided that having compute farms of last century Pentium II's is probably not a societally beneficial thing to do but lots of people seem to be willing to heat their homes that way in search of the Holy Grail. Save the energy and buy a modern computer.

Metod, S56RKO
Metod, S56RKO
Joined: 11 Feb 05
Posts: 135
Credit: 762,738,423
RAC: 14,991

Speedup report: on dual

Speedup report: on dual Opteron 280, CPU time for long results (z1_1462.0) dropped from 3340-3420 secs to 2010-2140 secs. Comparing U41.02 versus D40.

[edit] All of U41.02 results are still pending. [/edit]

[edit 2] This means that this box could crunch nearly 43 long results per day. Luckily there's more than one project attached. [/edit 2]

Metod ...

Mike Hewson
Mike Hewson
Moderator
Joined: 1 Dec 05
Posts: 6,119
Credit: 119,716,173
RAC: 74,237

RE: Oops. I didn't mean to

Message 30152 in response to message 30150

Quote:
Oops. I didn't mean to personalize my criticism, if it was construed that way. Obviously, Mike would like me to choose more politically correct expressions.


Nah! Heaven forbid! Polite will do. :-)

Quote:
But that wouldn't make me much fun at dinner parties, would it?


Bingo! That's the big challenge with electronic communication. We don't have the usual non-verbal cues that actually dominates human conversation, and thus 'slips between cup and lip' are that much easier. Not that I'd be up to web cams though...... :0)

Quote:
'Poor programming' is suggested when you can get a factor of two or more performance improvement merely by re-writing the code. Why is that debated?


It isn't. The resources to do that are though...

[shameless rhetoric]
Could I take this moment to generally remind users that they are ( very kindly ) volunteering their resources to a project which hasn't got enough? That is why we are doing it aren't we? So should we be ( recurrently it seems ) surprised to discover that E@H, LIGO et al have some finities/limitation(s) at some level? Thus choices have to be made by someone/somewhere that will create distinct outcomes which by necessity are more or less suitable for certain sets of goals/specifications. Can we all see the wider view on that? So, by all means please debate technical merit honestly and directly - it is the key to improvement - but please keep the language neutral. Sigh.....
[/shameless rhetoric]

Cheers, Mike.

I have made this letter longer than usual because I lack the time to make it shorter. Blaise Pascal

AnRM
AnRM
Joined: 9 Feb 05
Posts: 213
Credit: 4,346,941
RAC: 0

No invalids overnight on any

No invalids overnight on any box using U41.02/3.....it's rock and roll time on the Akos SSE3 Express! Our 2 lonely Celerons suddenly think they are AMDs and can finally hold their own....well done, Akos....Cheers, Rog.
(PS...Of course the down side to this incredible speed is we have more machines maxing out earlier. This makes the Rosetta and Malaria projects happy though)

Akos Fekete
Akos Fekete
Joined: 13 Nov 05
Posts: 560
Credit: 4,510,638
RAC: 0

RE: No invalids overnight

Message 30154 in response to message 30153

Quote:
No invalids overnight on any box using U41.02/3.....it's rock and roll time on the Askof SSE3 Express! Our 2 lonely Celerons suddenly think they are AMDs and can finally hold their own....well done, Askof....Cheers, Rog.
(PS...Of course the down side to this incredible speed is we have more machines maxing out earlier. This makes the Rosetta and Malaria projects happy though)

Good news. I will put U41.04 to the download page (it is about 1-2% faster than U41.03). I'm glad to hear that these clients can save time for other projects too! My PCs also run other project in their free time.

PS: I don't want to offend you, but could you call me to Akos? Thanks. :-)
/ Sorry, but it is a nice Hungarian name, as far as i know. :-) /

M. Schmitt
M. Schmitt
Joined: 27 Jun 05
Posts: 478
Credit: 15,872,262
RAC: 0

RE: PS: I don't want to

Message 30155 in response to message 30154

Quote:
PS: I don't want to offend you, but could you call me to Akos? Thanks. :-)
/ Sorry, but it is a nice Hungarian name, as far as i know. :-) /

You are a "Schlitzohr" :-)))

cu Michael

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.