cluster computer

Message boards : Number crunching : cluster computer
Message board moderation

To post messages, you must log in.

1 · 2 · 3 · Next

AuthorMessage
wouter

Send message
Joined: 10 Jan 11
Posts: 10
Credit: 4,204
RAC: 0
Message 1067445 - Posted: 16 Jan 2011, 22:04:12 UTC

I have a lot of bad computers.
Pentium 3 and 4.

So I thought if I let them all work together that I could do more work.

but I know nothing about cluster computers.

can anyone help me?

already thanks
ID: 1067445 · Report as offensive
-BeNt-
Avatar

Send message
Joined: 17 Oct 99
Posts: 1234
Credit: 10,116,112
RAC: 0
United States
Message 1067449 - Posted: 16 Jan 2011, 22:18:36 UTC - in response to Message 1067445.  
Last modified: 16 Jan 2011, 22:21:24 UTC

I have a lot of bad computers.
Pentium 3 and 4.

So I thought if I let them all work together that I could do more work.

but I know nothing about cluster computers.

can anyone help me?

already thanks


I believe that the BOINC client along with relevant parts of the Seti@Home applications would have to support clustering but I'm not totally sure because #1. I've never messed with clustering for BOINC and #2. I'm not sure what all it would entail. There are several articles out there on how to setup Windows and Linux clusters. There is an article in the wiki about it but's it's pretty vague, and I think it is talking about actually running them seperately instead of as a one 'super computer' http://www.boinc-wiki.info/Creating_a_diskless_cluster.

Here is another thread about it here, http://setiathome.berkeley.edu/forum_thread.php?id=47382 the last post says:

Using VCS, I have succesfully ran BOINC as a service group in a 2 node cluster (Solaris 10) using veritas cluster.

Basically, the boinc installation was out on a SAN..during failover the disk group and virtual hostname was moved to Node 1 (or vice versa) and boinc was then relaunched on the failover node.

It defeats the purpose though...as only one node in the cluster can run the boinc software in any given time.

This might not be what you mean by cluster though...because VCS is really only for high availability...


But your results may vary since he was running it on a high availability cluster instead of a real time one. Interesting wish I had enough time and machines to play with it for sure!
Traveling through space at ~67,000mph!
ID: 1067449 · Report as offensive
wouter

Send message
Joined: 10 Jan 11
Posts: 10
Credit: 4,204
RAC: 0
Message 1067457 - Posted: 16 Jan 2011, 22:51:56 UTC - in response to Message 1067449.  

thanks
ID: 1067457 · Report as offensive
Profile Bernie Vine
Volunteer moderator
Volunteer tester
Avatar

Send message
Joined: 26 May 99
Posts: 9954
Credit: 103,452,613
RAC: 328
United Kingdom
Message 1067459 - Posted: 16 Jan 2011, 22:56:00 UTC - in response to Message 1067457.  

I think this is what you want

PVM
ID: 1067459 · Report as offensive
archae86

Send message
Joined: 31 Aug 99
Posts: 909
Credit: 1,582,816
RAC: 0
United States
Message 1067460 - Posted: 16 Jan 2011, 22:56:33 UTC - in response to Message 1067445.  

I have a lot of bad computers.
Pentium 3 and 4.

So I thought if I let them all work together that I could do more work.

Have you considered the cost of power? For machines turned on more than a small fraction of a year, the power cost disadvantage of machines of that era is so bad that even when they are free it is cheaper to turn them off and spend part of the money that would have been spent to power them to buy modern hardware with equal or greater compute power and far lower power consumption.

Seriously, unless power is somehow free to you, and you don't care about the person or entity who is paying for it, this might not be wise even if you can work out how to do it.

ID: 1067460 · Report as offensive
Iona
Avatar

Send message
Joined: 12 Jul 07
Posts: 790
Credit: 22,438,118
RAC: 0
United Kingdom
Message 1067470 - Posted: 16 Jan 2011, 23:23:54 UTC - in response to Message 1067460.  

I have a lot of bad computers.
Pentium 3 and 4.

So I thought if I let them all work together that I could do more work.

Have you considered the cost of power? For machines turned on more than a small fraction of a year, the power cost disadvantage of machines of that era is so bad that even when they are free it is cheaper to turn them off and spend part of the money that would have been spent to power them to buy modern hardware with equal or greater compute power and far lower power consumption.

Seriously, unless power is somehow free to you, and you don't care about the person or entity who is paying for it, this might not be wise even if you can work out how to do it.


Firstly, I'll admit that I know nothing about 'clusters', but what I do know about, is the cost of running 2 X 3.2Ghz P4 machines against the cost of running a single Core 2 6600 machine. There is no comparison - the Core 2 is cheaper to run. Cheaper to run, any way you care to measure it. Even running a fairly lowly 20% of the time, the Core 2 6600 will get through twice the amount of work that the 2 X P4s could (and they were running for far longer), but at lower cost and with far less heat being expended. As archae66 has suggested, you'll have to think carefully about that power use - given that it would not take that long for a cheap (second user) Core 2 system to pay for itself in reduced power costs. As you can probably see, I run a pair of 'Core 2s' and I've not looked back.



Don't take life too seriously, as you'll never come out of it alive!
ID: 1067470 · Report as offensive
Profile soft^spirit
Avatar

Send message
Joined: 18 May 99
Posts: 6497
Credit: 34,134,168
RAC: 0
United States
Message 1067473 - Posted: 16 Jan 2011, 23:27:43 UTC - in response to Message 1067470.  

not to mention a single decent GPU will put out more than a room full of P3-P4's.
Janice
ID: 1067473 · Report as offensive
wouter

Send message
Joined: 10 Jan 11
Posts: 10
Credit: 4,204
RAC: 0
Message 1067588 - Posted: 17 Jan 2011, 7:47:50 UTC - in response to Message 1067473.  

Thank you all.

but I just wanted my old computers for seti
and I do not think it really helps when I always miss my deadline.

So I was thinking I let them work together.
and that I get all my deadlines.
ID: 1067588 · Report as offensive
tbret
Volunteer tester
Avatar

Send message
Joined: 28 May 99
Posts: 3380
Credit: 296,162,071
RAC: 40
United States
Message 1067600 - Posted: 17 Jan 2011, 9:05:17 UTC - in response to Message 1067445.  
Last modified: 17 Jan 2011, 9:24:05 UTC

I don't know that this will help at all, but here goes nothing.

The P-III? You can probably safely ditch it.

The P-4, however, is a different story.

IF your P-4 has a PCIe slot (it doesn't have to be a PCIe-2) - that's a PCIe x16 slot, the long one -- you can pick-up an nVidia GT 240 video card for (right now) $42.00 after rebate from Newegg. I have no idea if that's the cheapest out there right now or not.

Unless you have a really flaky power supply (I mean *really* flaky) you shouldn't have any problem running that card in your P-4.

So, let's say you are thinking, "Why on Earth would I spend another $42.00 on this old computer?"

Because the video cards are MUCH faster SETI crunchers than the computers.

As an example - I've got an old P4 1.8GHz machine that does not have a PCIe slot. It's average SETI "credit" is something below 200 per day. I have recently told BOINC not to get any new SETI tasks, and I will retire it from crunching for SETI.

I've got another old P-4 3GHz computer that has a PCIe slot on the motherboard. I was using the CPU to do SETI crunching and it would get a few hundred credits per day. I bought a nVidia GT240 for it, stuck it in, loaded drivers, got it working, and walked away.

That computer now gets about 5,000 credits per day.

The nVidia GT240 (which is a slow video card by today's standards) can do as much work in just over an hour as the P-4 1.8GHz computer was doing all day and all night.

It just doesn't make any sense to continue to use the electricity on these old computers *unless* you can stick a CUDA-capable video card in them.

I have another "old" computer (Phenom quad core) that I stuck a nVidia GT9800 in. In theory the 9800 should be faster than the GT240. The cheapest I see the GT 9800 is about $90. I would not recommend spending the extra money on the GT 9800. The 9800 seems to get hotter than the 240 and it (at least mine) is more prone to producing errors.

The 9800 has a theoretical advantage over the 240, but in MY EXTREMELY LIMITED EXPERIENCE the 240 does as much "useful" work as the 9800, maybe more.

My 9800 was branded PNY and my 240 is branded ASUS. In my case, I had to restrict the number of CPU cores running SETI on the Phenom quad core to keep the case temperatures down. (I am using only one of four cores for SETI, plus the video card. My system monitor shows that I have one core idling, two cores doing very little, and one core busy with SETI.)

So IF your P-4 motherboard has a PCIe slot and you are willing to spend $40 on a video card, you can make it a "useful" cruncher.

The more powerful nVidia cards do 10 times the work, are 4 to 10 times as expensive, and would mean you had to spend even more money on a new power supply to run them.

You aren't going to win many races with a GT 240, but you will get a lot more work done per watt and per hour than with your CPU.

If you don't have a PCIe slot on your P-4 motherboard, you could let it do as much as it will do, anyway.

I love the idea of a computing cluster, but I have sincere doubts that the outcome would out-strip even a net $40 CUDA-capable video card.

The P-4 1.8GHz I am retiring is only the first of several "retirement parties" I am planning in the near future.
ID: 1067600 · Report as offensive
wouter

Send message
Joined: 10 Jan 11
Posts: 10
Credit: 4,204
RAC: 0
Message 1067606 - Posted: 17 Jan 2011, 10:13:20 UTC - in response to Message 1067600.  
Last modified: 17 Jan 2011, 10:35:34 UTC

I'm not sure that a p-III or p-4 can calculate a seti work within the deadline.
ID: 1067606 · Report as offensive
Profile Gundolf Jahn

Send message
Joined: 19 Sep 00
Posts: 3184
Credit: 446,358
RAC: 0
Germany
Message 1067613 - Posted: 17 Jan 2011, 10:45:04 UTC - in response to Message 1067606.  
Last modified: 17 Jan 2011, 10:45:40 UTC

I'm not sure that a p-III or p-4 can calculate a seti work within the deadline.

I have a PIII 700MHz notebook running setiathome_5.27 under NT4. It needs about 10 hours (VHAR) to 38 hours for a work unit.

Gruß,
Gundolf
ID: 1067613 · Report as offensive
wouter

Send message
Joined: 10 Jan 11
Posts: 10
Credit: 4,204
RAC: 0
Message 1067614 - Posted: 17 Jan 2011, 10:47:38 UTC - in response to Message 1067613.  
Last modified: 17 Jan 2011, 11:12:41 UTC

thanks

and to get more off topic i have a ati hd 4XXX serie on my motherboard.

can i use him in a combination with a hd 5770 to compute for Collatz Conjecture.

i am in school right now and i will look more info up when i am home.
ID: 1067614 · Report as offensive
Claggy
Volunteer tester

Send message
Joined: 5 Jul 99
Posts: 4654
Credit: 47,537,079
RAC: 4
United Kingdom
Message 1067616 - Posted: 17 Jan 2011, 11:09:31 UTC - in response to Message 1067606.  

I'm not sure that a p-III or p-4 can calculate a seti work within the deadline.


I have a PIII 800E @896MHz that does Astropulse, Wu's take about 148 hours (6 days) with Optimised Astropulse,

Claggy
ID: 1067616 · Report as offensive
Profile Tazz
Volunteer tester
Avatar

Send message
Joined: 5 Oct 99
Posts: 137
Credit: 34,342,390
RAC: 0
Canada
Message 1067632 - Posted: 17 Jan 2011, 13:00:24 UTC - in response to Message 1067606.  

I'm not sure that a p-III or p-4 can calculate a seti work within the deadline.


My AMD K6-2 850 MHz does 1 wu in a little over a day, 2 days if any other programs are running.
</Tazz>
ID: 1067632 · Report as offensive
OzzFan Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Apr 02
Posts: 15691
Credit: 84,761,841
RAC: 28
United States
Message 1067648 - Posted: 17 Jan 2011, 14:50:55 UTC - in response to Message 1067632.  

I'm not sure that a p-III or p-4 can calculate a seti work within the deadline.


My AMD K6-2 850 MHz does 1 wu in a little over a day, 2 days if any other programs are running.


The fastest official K6-2 ran at 550MHz. There may have been a 600MHz laptop version. 850MHz sounds like an original Athlon CPU.
ID: 1067648 · Report as offensive
OzzFan Crowdfunding Project Donor*Special Project $75 donorSpecial Project $250 donor
Volunteer tester
Avatar

Send message
Joined: 9 Apr 02
Posts: 15691
Credit: 84,761,841
RAC: 28
United States
Message 1067649 - Posted: 17 Jan 2011, 14:56:11 UTC - in response to Message 1067588.  

but I just wanted my old computers for seti
and I do not think it really helps when I always miss my deadline.


For comparison, I've had SETI@Home running on an old Pentium Classic 233MHz CPU and not miss deadlines. More recently I have a dual Pentium III 700MHz that can make deadlines too, but it runs 24/7.

It's ok to miss a few deadlines while BOINC gets a better idea about how often your computer is turned on, allowed to crunch, and it's FPU abilities.

So I was thinking I let them work together.
and that I get all my deadlines.


Unfortunately, SETI@Home does not support clusters. It would treat each node in the cluster as a separate machine; it would not see them as one giant machine.

BOINC could be rewritten to support clustering, but then that would defeat the purpose of using spare CPU cycles. Many people here create spare CPU cycles just to donate to crunching (as is their choice, I'm not knocking it) by having multiple machines up and running 24/7 (including myself), but that's above and beyond the intention of the BOINC system to only use what you don't when your computer is on.
ID: 1067649 · Report as offensive
Profile ML1
Volunteer moderator
Volunteer tester

Send message
Joined: 25 Nov 01
Posts: 20147
Credit: 7,508,002
RAC: 20
United Kingdom
Message 1067655 - Posted: 17 Jan 2011, 15:21:08 UTC

Is it worthwhile putting a recent GPU card into an "old clunker" to then do mainly GPU crunching?

Or would that be wasted effort for the returns?


Happy crunchin',
Martin

See new freedom: Mageia Linux
Take a look for yourself: Linux Format
The Future is what We all make IT (GPLv3)
ID: 1067655 · Report as offensive
Profile skildude
Avatar

Send message
Joined: 4 Oct 00
Posts: 9541
Credit: 50,759,529
RAC: 60
Yemen
Message 1067657 - Posted: 17 Jan 2011, 15:25:49 UTC - in response to Message 1067655.  

It's probably better to invest in a signle GPU that you can install in an older rig than to replace everything with a new cpu/mobo/PSU/ram etc.


In a rich man's house there is no place to spit but his face.
Diogenes Of Sinope
ID: 1067657 · Report as offensive
Profile James Sotherden
Avatar

Send message
Joined: 16 May 99
Posts: 10436
Credit: 110,373,059
RAC: 54
United States
Message 1067660 - Posted: 17 Jan 2011, 15:31:16 UTC - in response to Message 1067606.  

I'm not sure that a p-III or p-4 can calculate a seti work within the deadline.

My old P4 2.53GHZ running with Opp apps Does all its work in time. AP takes me 31 hours MB any where from 5 to 9 hours. Depending on angle range. I have a bunch of vlars one will be done in about 9 hours. Now granted the P4s meager rac is low, But it Hardly ever does a bad work unit. Its been crunching SETI since 2004 24/7. Im hoping to replace it next year with another i7.
[/quote]

Old James
ID: 1067660 · Report as offensive
wouter

Send message
Joined: 10 Jan 11
Posts: 10
Credit: 4,204
RAC: 0
Message 1067663 - Posted: 17 Jan 2011, 15:58:29 UTC - in response to Message 1067660.  

Thank you all.
I think here all my questions answered.

So, Basically

1. it is better to separate all nodes.
2. it is better to invest in a GPU.
3. P-III and P-4 can both get the deadline easily.
ID: 1067663 · Report as offensive
1 · 2 · 3 · Next

Message boards : Number crunching : cluster computer


 
©2024 University of California
 
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.