MilkyWay@Home Project

Post by **debs3759** » Sun Mar 01, 2009 1:12 am

Yeah, he finally accepted that one thing that would help (even if only while he finds a fix to the other problems) is increasing the work size in order to reduce server hits while he gets his server problem fixed :)

Post by **danielmid** » Tue Mar 03, 2009 8:40 pm

Buga went and snuck ANOTHER ATI card into his farm. Just passed 3 mil,[sarcasm]Congrats Buga![/sarcasm]

Post by **Labbie** » Tue Mar 03, 2009 8:46 pm

Yeah, I noticed that too.

Post by **Labbie** » Tue Mar 03, 2009 9:23 pm

And one of his machines is 4th by RAC on the project.

http://milkyway.cs.rpi.edu/milkyway/top_hosts.php

Post by **MzSnowleopard** » Tue Mar 03, 2009 9:55 pm

so Buga... how's that Buga Bunker coming ?

Post by **Buga1** » Wed Mar 04, 2009 5:39 am

Who! What! Me!......... Never. LMAO

I thought I told you all. Maybe it was just KM in that night in FC. The days have been melting together.

Rick

Post by **Knightmare** » Wed Mar 04, 2009 6:42 am

Buga1 wrote:Who! What! Me!......... Never. LMAO

I thought I told you all. Maybe it was just KM in that night in FC. The days have been melting together.

Rick

'Twas just little ole me that knew about it...lol

Post by **debs3759** » Wed Mar 04, 2009 9:44 am

danielmid wrote:Buga went and snuck ANOTHER ATI card into his farm. Just passed 3 mil,[sarcasm]Congrats Buga![/sarcasm]

Tht's just showing off!

It'll be a few months before I can afford any decent GPUs, but I'll have a few of the best when it comes :)

Post by **MzSnowleopard** » Wed Mar 04, 2009 4:10 pm

showing off AND inspiring- Buga's antics tend to 'encourage' others to want more- spend more- have better computers. <coughs> i7's

Post by **Sarge_II** » Wed Mar 04, 2009 7:25 pm

IE7? Nah, I use FF.

Post by **MzSnowleopard** » Wed Mar 04, 2009 8:01 pm

l7.. you know- that comp Buga is masterminding for KM

Post by **Sarge_II** » Wed Mar 04, 2009 9:43 pm

MzSnowleopard wrote:l7.. you know- that comp Buga is masterminding for KM

That may be what they're call it on public forums, but we all know it's really a T3.

Post by **Buga1** » Thu Mar 05, 2009 4:17 am

I have a T3. LMAO. But it is a few generations behind now. Had a B4 Also but that now gone. But it wouldnt do much crunching. It's better at crashing. LOL

And I wasn't trying to be sneaky. I didnt know I was supposed to post about anything new I bought. My Bad. LOL Just jokin'.

But I think we will pass Seti here soon in MW, to become the #1 project we are on. Maybe in a month. Unless we get other "SNEAKY" ATI's a crunching. LMAO

Rick

Post by **Sarge_II** » Thu Mar 05, 2009 5:14 am

Buga1 wrote:I have a T3. LMAO. But it is a few generations behind now. Had a B4 Also but that now gone. But it wouldnt do much crunching. It's better at crashing. LOL

And I wasn't trying to be sneaky. I didnt know I was supposed to post about anything new I bought. My Bad. LOL Just jokin'.

But I think we will pass Seti here soon in MW, to become the #1 project we are on. Maybe in a month. Unless we get other "SNEAKY" ATI's a crunching. LMAO

Rick

"Get in, if you want to live."

Post by **Sarge_II** » Tue Mar 10, 2009 1:12 am

Code: Select all

State	#
Results ready to send	766
Results in progress	72,755
Workunits waiting for validation	47
Workunits waiting for assimilation	0
Workunits waiting for deletion	0
Results waiting for deletion	0
Transitioner backlog (hours)	0

Anybody else not getting WUs from MW?

Post by **Dune Finkleberry** » Tue Mar 10, 2009 6:02 am

Sarge_II wrote:

Code: Select all

State	#
Results ready to send	766
Results in progress	72,755
Workunits waiting for validation	47
Workunits waiting for assimilation	0
Workunits waiting for deletion	0
Results waiting for deletion	0
Transitioner backlog (hours)	0

Anybody else not getting WUs from MW?

I'm good.... so far.

I'll keep a closer eye on it though.

Post by **Chris S** » Tue Mar 10, 2009 3:17 pm

They seem to be coming through to me OK at the moment, but the guys with the high end GPU cards are pinching them all.

I tell you folks, and you heard it from me first

GPU crunching is going to screw Boinc big time

No project can possibly keep up with the flow of workunits required.....

Post by **debs3759** » Tue Mar 10, 2009 5:36 pm

I'm sure they can keep up with the flow over time (asthey improve the infrastructure),but not so sure they can keep up-to-date with actually using and processing all the results they get :)

Post by **Dune Finkleberry** » Tue Mar 10, 2009 6:45 pm

Chris S wrote:They seem to be coming through to me OK at the moment, but the guys with the high end GPU cards are pinching them all.

Well, I've certainly not got any high end equipment. Just my little ol' 5 year old 'puter.

Not bad 5 years ago, but landfill material now.

Post by **Sarge_II** » Tue Mar 10, 2009 11:37 pm

Dune Finkleberry wrote:
Sarge_II wrote:
Code: Select all
State	#
Results ready to send	766
Results in progress	72,755
Workunits waiting for validation	47
Workunits waiting for assimilation	0
Workunits waiting for deletion	0
Results waiting for deletion	0
Transitioner backlog (hours)	0
Anybody else not getting WUs from MW?
I'm good.... so far. I'll keep a closer eye on it though.

Apparently it was just a problem last night, lasting at most an hour.

My RAC has gone up more than I expected now that I have a mini-farm again. Since the used computer I bought a week ago has a single processor and I had an RAC of around 800 with a duo, I thought I'd max out at about 1200, but I am now in the mid-1400s.

Post by **Dark Angel** » Wed Mar 11, 2009 4:57 am

Chris S wrote:They seem to be coming through to me OK at the moment, but the guys with the high end GPU cards are pinching them all.

I tell you folks, and you heard it from me first

GPU crunching is going to screw Boinc big time

No project can possibly keep up with the flow of workunits required.....

Actually that's a little short sighted...they do this to get super computer ability without the expense...they will adapt given time either by upgrading infrastructure or by making the work units longer/bigger/more complex.

Post by **Gavin Shaw** » Thu Mar 12, 2009 12:08 am

12 Mar 2009 AEST:

Sorry for the lack of communication recently. I'm back from my flu so it should be a bit better now :) I started up some new searches, which should hopefully do a better job avoiding the edges of the search space which was causing some of the weird acting very long workunits. Should be starting up a few new types of searches in the next day or so as well, more on that later.

Post by **Chris S** » Sat Mar 14, 2009 11:10 am

I'm looking ahead to maybe a years time. If all the main projects get requests from their crunchers to provide GPU apps, and the majority start using them, then I can see some problems.

As already said, firstly they'll have to provide extra workunits of a few orders of magnitude, can the projects infrastructure support that? Next, longer workunits or more complex ones will evolve to deal with that workflow. This could result in slower or older PC's being forced out of Boinc.

Finally, where is the money and time coming from in the projects, to deal with all this massive influx of information? Seti has had the begging bowl out for the last 2 years without much return.

Boinc started out as a peoples DC project that anyone with a computer, slow or fast, could join in with. If this GPU phenomena takes over from CPU crunching it will become a playground for the rich who can afford top end Graphics cards at £350 each, and fit 3 at a time.

GPU capability seems to be increasing exponentially, with most cards having more ram than the host PC these days, so its not going to get any better as time goes on. My 2c worth is that projects need to send out two types of workunits, GPU only ones, and CPU only ones, to let everybody have a a fair chance to get involved.

Nobody ever really saw this GPU thingy coming over the horizon but its here and needs to be dealt with. So I will restate my case that if it isn't, Boinc is in trouble.

[smilie=eek.gif]

Post by **debs3759** » Sat Mar 14, 2009 1:22 pm

Chris S wrote:Boinc started out as a peoples DC project that anyone with a computer, slow or fast, could join in with. If this GPU phenomena takes over from CPU crunching it will become a playground for the rich who can afford top end Graphics cards at £350 each, and fit 3 at a time.

hehe, have you got frostbite?

I have been following Ice's progress with his GPUs with some sense of amusement (yeah,I know it's not just him, but he is very vocal about it!). I think those who talk about it as an addiction are right, there do seem to be a few crunchers who crave that extra credit (I am one of them, BTW), and some of them act as if life wouldn't be worth livving without the ability to overtake everyone else in the stats.

Yeah, there will be a great division between those who can and those who can't (use a GPU), but that doesn't mean the rest of us can't enjoy the competitive spirit, or even (God forbid!) just enjoy doing the science :)

Post by **Chris S** » Sat Mar 14, 2009 2:18 pm

Hehehe, yes you might be about right ......

Well, that seems to endorse my view that there should be a place for everyone, whatever their computing ability and depth of pocket.

Post by **Gavin Shaw** » Thu Mar 19, 2009 11:42 pm

It is official...

MilkyWay is the number one crunched project by credits for Calm Chaos, having overtaken Seti overnight...

Seti is now in second place, with ABC at a distant third...

Post by **zerocool** » Thu Mar 19, 2009 11:51 pm

lol nice one well done one and all. happy chrunchin

Post by **zerocool** » Thu Mar 19, 2009 11:53 pm

looks like i7 and cuda buggs invading bugas area lol

Post by **Chris S** » Tue Mar 24, 2009 8:55 am

Looks like I was right in my prediction that GPU crunching is hurting MIlkyway. Unless money and time can be found to increase a projects infrastructure, I can see that the next generation of video cards will cause even more problems.

Anyway, if GPU is thwe way to go, how long before someone brings out a 4 procesor card dedicated to crunching not gaming?

Post by **Gavin Shaw** » Thu Mar 26, 2009 1:30 am

26 Mar 2009 AEST:

Sorry for the recent lack of communication, but we've been working on quite a few things behind the scenes here.
First, my new laptop just came in the mail, so I'll be able to seriously start working on a CUDA version of the milkyway application. We'll be developing this for OS X, Linux and Windows.
Second, we're going to be doing a few server-side changes to how things are run, which should significantly help with workunit reporting and work availability. We'll be splitting MilkyWay@Home into two separate applications/backends, Milkyway regular and Milkyway GPU. This will allow us to keep current workunit sizes for non-GPU applications, while farming a lot more work out to the GPUs (so they need to contact the server less frequently). We'll be doing different optimization methods on the regular Milkway and the GPU Milkyway so they'll both be doing equally valid research. In the next day or two I'll be posting the code that the application will be using for the GPU version. After testing to make sure that it's working correctly we'll swap to only awarding credit to the GPU applications for GPU milkyway (that way our servers wont be as bogged down, which will mean better work availability). Also, before a credit firestorm erupts again, we have no plans to award less credit/work to the GPU applications, so you don't need to worry about that.

-------------------------------------------------------------

This could be a good thing...

Post by **Dark Angel** » Fri Mar 27, 2009 5:59 am

Gavin Shaw wrote:26 Mar 2009 AEST:

Sorry for the recent lack of communication, but we've been working on quite a few things behind the scenes here.
First, my new laptop just came in the mail, so I'll be able to seriously start working on a CUDA version of the milkyway application. We'll be developing this for OS X, Linux and Windows.
Second, we're going to be doing a few server-side changes to how things are run, which should significantly help with workunit reporting and work availability. We'll be splitting MilkyWay@Home into two separate applications/backends, Milkyway regular and Milkyway GPU. This will allow us to keep current workunit sizes for non-GPU applications, while farming a lot more work out to the GPUs (so they need to contact the server less frequently). We'll be doing different optimization methods on the regular Milkway and the GPU Milkyway so they'll both be doing equally valid research. In the next day or two I'll be posting the code that the application will be using for the GPU version. After testing to make sure that it's working correctly we'll swap to only awarding credit to the GPU applications for GPU milkyway (that way our servers wont be as bogged down, which will mean better work availability). Also, before a credit firestorm erupts again, we have no plans to award less credit/work to the GPU applications, so you don't need to worry about that.

-------------------------------------------------------------

This could be a good thing...

I hate to say it but contrary to what they are saying this looks like step one of how to justify lowering credit for a GPU app cruncher...I'm not yet in that group but I wouldn't put it past them to try it...they've project normed to SETI why not project norm the GPU app too...I mean it should be granting just as much credit as say the stock SETI app...least that's how they think.

Post by **Dances with Werewolves** » Fri Mar 27, 2009 11:45 am

If he is working on a CUDA app. wont that shut out all the ATI cards? CUDA is Nvidia.

Post by **Gavin Shaw** » Fri Mar 27, 2009 12:39 pm

Dances with Werewolves wrote:If he is working on a CUDA app. wont that shut out all the ATI cards? CUDA is Nvidia.

CUDA is Nvidia so yes it will not support ATI cards. However, they may still allow the ATI app already available...

Post by **Labbie** » Fri Mar 27, 2009 1:31 pm

Nah, the BOINC developers are working on recognizing ATI cards already. That version should be available within a month or so.

Post by **debs3759** » Fri Mar 27, 2009 3:08 pm

Dark Angel wrote:I hate to say it but contrary to what they are saying this looks like step one of how to justify lowering credit for a GPU app cruncher...I'm not yet in that group but I wouldn't put it past them to try it...they've project normed to SETI why not project norm the GPU app too...I mean it should be granting just as much credit as say the stock SETI app...least that's how they think.

And I just paid a couple of bills instead of buying 2 * 4850 and 2 * 4870 (the 2 * 4850 were a single deal on eBay, the 2 * 4870 would have been because I saw them cheaper than most, and one came with 4 GB of DDR2 DIMMs).

If they really do take that stance and reduce the credits, I will have to kick myself REALLY HARD for being responsible instead of feeding the habit :)