However, that makes it sound like it doesn't cost anything to run, which is untrue. Full use of the processor cost electricity. About 50% higher for my desktops and 250% more for my laptop.

Yeah, that's it exactly. There is an added cost by participating. In your case, of that $650/year BOINC costs you, how much of that is added cost? $400?

and a whole lot more if I hadn't bought the AMD64's just to run boinc. LOL

I'm... not sure what to do about that. You're cheating the easy system I had thought up :(. There's no supposed to be any direct purchase of hardware otherwise unused, just existing hardware that was in place and non-maximized.

You've *massively* increased the cost per workunit by doing that. :) Probably doubling I'd guess. Would have been far more efficient to donate the money directly if you saw no side effect. But.. people don't see donated money reflected in credit scores. On the flip side, there would be almost no reduction in crunching power by giving that computer to someone who could then use it functionally, (average CPU usage is probably about 1-3%, comprised of big spikes for short durations).

Consider if everyone who bought a computer only, (and I mean only) to crunch... gave it to a school to use instead. The school would keep them crunching non-stop, and the people who bought them would pay their respective power bills, based on how long they'd be running (or running at heightened wattage) when the school otherwise wouldn't have had them on/maxed out.

Let say the school would have the computers running 8 hours a day, non-maxed. So, you're adding 16 hours maxed, and 8 hours of about a 35% increase.

So, the computer's owner would see an 8 hour reduction by 65% in the power bill they pay. And see only about a 1-3% drop in crunching results. Which is somewhat minor, but, that's not the big deal. The big deal is that the school would benefit massively because they still need computers in a computer lab for kids to sit down at, when the kids need to. Those computers have to be fast enough, when they need them, which is in big spikes for short durations.

Not really ever applicable in the real world sadly. There aren't many who hook up computers that serve no functional purpose except to crunch. Even worse, they're almost never in the same geographical area. Plus, the logistics of setting up and administering a program like this are somewhat staggering. Fun as an intellectual exercise though :)

---

Just to add to the discussion. There are benefits to some of the other projects, mainly those that don't always have work, but when they do require the results quickly. Those projects would probably have to wait if they tried to book a super-computer, where as by using BOINC they get an almost instantaneous response.

Also going by info from the Stats sites, Seti hosts on average only crunch 3 units/day, so I assume this is mainly done on the 'spare resourses' of those hosts. In the general scheme of things there are very few people who run multiple computers which run simply to crunch BOINC projects, and one assumes the owners are willing to donate the costs, whatever they are to the cause. And I also assume quite a few of those that run farms, do so at there place of work were they have permission. And if these few people were to donate cash instead of computers, including all costs, the sums gathered would be fairly small.

---

Those projects would probably have to wait if they tried to book a super-computer, where as by using BOINC they get an almost instantaneous response.

This is why I'm interested in hearing from someone familiar with the supercomputing market.

Presumably, writing a project massive enough that it requires supercomputing resources takes a while. Maybe long enough that scheduling wouldn't be an issue, they could have it timed.

Regardless, I think you're right. If not simply "faster", they'd have flexibility. Flexibility would be a big thing. No scheduling to meet. If you're late getting your project set up, you won't miss your timeslot that you booked months in advance, or be paying for that downtime.

Projects could have varying priority. For longer term ones, low, and cheaper prices. But if you want something *now*, and cause other projects to be shelved, you have to pay a premium. But, always low priority stuff available to take off the shelf (or, if *nothing* paying available, can default down to chairity projects again).

I wonder how many projects in the world *don't* exist because someone says "Feh, that'd take a supercomputer to figure out, and we can't pay for that!" And how many doors this would open to people, (If any doors at all. Who knows, maybe we burn more money in power costs than we generate in computing value).

Also going by info from the Stats sites, Seti hosts on average only crunch 3 units/day, so I assume this is mainly done on the 'spare resourses' of those hosts.

Where do you see that? Or are you just guessing after some poking around?

In the general scheme of things there are very few people who run multiple computers which run simply to crunch BOINC projects, and one assumes the owners are willing to donate the costs, whatever they are to the cause. And I also assume quite a few of those that run farms, do so at there place of work were they have permission. And if these few people were to donate cash instead of computers, including all costs, the sums gathered would be fairly small.

Willing to donate the costs for dedicated crunchers, of course. We know that, because they do it! I'm wondering how high that value is though. One further refinement we can look at.

I also agree that if people actually were given a choice to donate, they would not. People like credit and crunching. Crunching is like building a pyramid, grande, exciting, and trackable progress. Donating is like paying bills. No one likes paying bills.

I think I'll split this off into a new thread once I gather my thoughts on what kinds of questions we need answered.

---

Also going by info from the Stats sites, Seti hosts on average only crunch 3 units/day, so I assume this is mainly done on the 'spare resourses' of those hosts.

Where do you see that? Or are you just guessing after some poking around?

I took the credits/day and the active hosts numbers, then divided by the average credits.

---

I took the credits/day and the active hosts numbers, then divided by the average credits.

I must be dumb. All I see are Top Participants/Teams/Comps for stats anywhere I look. URL? :)

---

Boinc synergy stats

OR

Boincstats

---

Boinc synergy stats OR Boincstats

Excellent. Hrm.

We could do a quick 'n dirty Added Cost calc, (which we could use to get an Added Cost/WU number) as follows:

1) Take the total sum of CPU-time, for some time period.

2) Assume no one leaves their computer on any longer than they normally would, so that added-cost is only the difference between average CPU usage and full-CPU usage, not the full wattage of having the rest of the computer runing too.

3) Assume some kind of average added-wattage for full-CPU rather than normal CPU (~40 Watts?)

4) Take the number from #1, and multiply it by the number from #3. We then have kW hours. That's a neat number to know, it's how much power the entire SETI community consumed in that time period.

5) Assume some average power cost ($0.15/kWh, $0.10/kWh, whatever). Multiply it by the number from #4. We then know the power cost of SETI. That's a neat number to know too, it's how much all of SETI's crunching volunteers spent in that time period to crunch.

6) See how many workunits were completed in that time period. Divide the number from #5 by this number. That's how much it costs per SETI workunit. That's an interesting number to know, because it can be compared to supercomputer performance/cost to figure out who's more efficient and by how much. It can be used to demonstrate the feasibility of a BOINC-based distributed-computing business. It can be used to demonstrate how BOINC is (or isn't) an alternative to supercomputer renting.

...

All we'd need to calculate this.. is the first piece.

Does anyone know if CPU-Time is an available stat anywhere, project-wide? That and the number of workunits it produced. Even just a total sum for the history of BOINC would be fine.

---

Well I figure it this way. On average running Boinc is about 55 watts more than normal on my units.
Given this, times our rate , per month I am spending about 6.34 a machine to crunch Boinc.
Now times that by what I had before the great meltdown.
I was spending a little over 190 bucks more a month for juice to crunch WU's.

Cost after the meltdown?
Well that will be a grand total of 19.02 a month until I repair the rest of the machines.
Yes thats going from 30 units in the farm to 3.

---

Beach Bums Current Stats:


Come Join us at Hawaiian Beach Bums

Yes thats going from 30 units in the farm to 3.

How did you lose 27 systems???

---

Grant
Darwin NT

How did you lose 27 systems???

I asked the same thing. I'll let him elaborate himself it he wants.. but.. apparently just a huge electrical storm a couple days ago.

That's *through* surge protectors and some were hooked up to an UPS too.

That doesn't seem possible to me. I was thinking maybe a surge killed his air conditioner causing a room full of 30 of them to cook themselves silly. No idea though, or if by meltdown he means they just got spiked and died, or, actually fried parts.

*sobs a little* .. sure they were past their prime, but they still had so much life left in them. So much crunching to give :(

---

apparently just a huge electrical storm a couple days ago.

That's *through* surge protectors and some were hooked up to an UPS too.

That doesn't seem possible to me.

Most surge protectors are gimmicks.
UPSs have much better surge protection, but most of them these days are standby units so if the lightning strikes while the power is still up it'll go right through the UPSs & takeout whatever is attached.
With online UPSs the device is alway running from the battery & is never connected to the mains. The UPS may die but it's very rare for a connected device to die as well, even with a big strike.

---

Grant
Darwin NT

Most surge protectors are gimmicks.
UPSs have much better surge protection, but most of them these days are standby units so if the lightning strikes while the power is still up it'll go right through the UPSs & takeout whatever is attached.

Apparently so. I thought surge protectors prevented over-voltage, and UPS prevented under-voltage. Voltage regulators maintain a steady isolated voltage.

Surge protectors should.. you know.. protect against.. you know.. surges. Being.. surge.. protectors.. designed exactly for lightning strikes and power spikes from the grid :(

[edited to remove some stupids, mis-parsed what you said the first time]

I mentioned that he should contact the manufacturer and see how much dollar amount they gaurantee, (though if they're charming chinese cheapos, prolly nothing).

With online UPSs the device is alway running from the battery & is never connected to the mains.

I think what I'm thinking is a regulator is what you're calling an online UPS. I concurr though.

---

Surge protectors should.. you know.. protect against.. you know.. surges. Being.. surge.. protectors.. designed exactly for lightning strikes and power spikes from the grid :(

Surge protectors only have a limited life span, as well as maximum ratings. The heavier duty surge protectors generally have higher rated devices, as well as lots of them in parallel.
They do give some protection from small line surges, but they're no good at keeping lightning strikes out.

I mentioned that he should contact the manufacturer and see how much dollar amount they gaurantee, (though if they're charming chinese cheapos, prolly nothing).

Those "$10,000 insurance guarantee against any device connected to this surge protector ever being damaged by a surge of any sort" type warranty? A load of ^&*#& IMHO.

With online UPSs the device is alway running from the battery & is never connected to the mains.

I think what I'm thinking is a regulator is what you're calling an online UPS. I concurr though.

A regulator is a regulator- it just regulates the voltage coming in to keep the output stable.

A UPS is as it suggests- an Uninteruptable Power Supply.
With a standby unit (most common type) the load is connected to the mains. The better types provide some voltage regulation in the form of buck & boost- if there is a slight drop in the supply voltage they will boost it to the nominal voltage. If there is a slight rise then they will drop it to the nominal voltage. Although these actions are only good for a few mS to a few 100mS.
If the mains power drops completely, then the unit switches the load to the output provided by the battery & switches that output on as well. Better units will also do this if the input voltage exceeds or drops below what the UPS can manage with it's buck & boost feature.

With an online UPS the load is never connected to the mains supply. It is permanetley connected to the output of the UPS provided by the battery. Think of it as a mains powered inverter, with battery backup.

---

Grant
Darwin NT

Surge protectors should.. you know.. protect against.. you know.. surges. Being.. surge.. protectors.. designed exactly for lightning strikes and power spikes from the grid :(

Surge protectors only have a limited life span, as well as maximum ratings. The heavier duty surge protectors generally have higher rated devices, as well as lots of them in parallel.
They do give some protection from small line surges, but they're no good at keeping lightning strikes out.

I mentioned that he should contact the manufacturer and see how much dollar amount they gaurantee, (though if they're charming chinese cheapos, prolly nothing).

Those "$10,000 insurance guarantee against any device connected to this surge protector ever being damaged by a surge of any sort" type warranty? A load of ^&*#& IMHO.

With online UPSs the device is alway running from the battery & is never connected to the mains.

I think what I'm thinking is a regulator is what you're calling an online UPS. I concurr though.

A regulator is a regulator- it just regulates the voltage coming in to keep the output stable.

A UPS is as it suggests- an Uninteruptable Power Supply.
With a standby unit (most common type) the load is connected to the mains. The better types provide some voltage regulation in the form of buck & boost- if there is a slight drop in the supply voltage they will boost it to the nominal voltage. If there is a slight rise then they will drop it to the nominal voltage. Although these actions are only good for a few mS to a few 100mS.
If the mains power drops completely, then the unit switches the load to the output provided by the battery & switches that output on as well. Better units will also do this if the input voltage exceeds or drops below what the UPS can manage with it's buck & boost feature.

With an online UPS the load is never connected to the mains supply. It is permanetley connected to the output of the UPS provided by the battery. Think of it as a mains powered inverter, with battery backup.

The lightning could have come in through the LAN. I've seen that happen.

---

Fear will keep the local systems in line. Fear of this battle station. - Grand Moff Tarkin

The lightning could have come in through the LAN. I've seen that happen.

Yep.
Lightning stirkes that come through phone lines/data lines cause much more damage than ones that come through the power lines.

---

Grant
Darwin NT

Lightning struck my neighbors tree, jumped to my house, started a receptacle on fire(which I extinquished), and didn't affect my puters, TV's, or any other electronic device. We did find 5 dead squirrels under the tree though. May they rest in piece.

different story:

I was a kid when lightning struck our house, it chipped the plaster from 3 corners of my room, but didn't wake me. My mom freaked when she came in and I wasn't awake.

Lightning struck my neighbors tree, jumped to my house, started a receptacle on fire(which I extinquished), and didn't affect my puters, TV's, or any other electronic device. We did find 5 dead squirrels under the tree though. May they rest in piece.

A friend used to work for the local branch of our phone company.
He once did a call out to a house that had been hit by ligthning. The strike hit their shed, then ran along the clothesline (that was strung between the shed & the house) & then through the house.
He could see where the phone used to be mounted on the wall by the big black burn mark where it was blown off, and the burn mark that ran up the wall & into the roof where the phone line used to be.
Most of the light fittings had been blown off of the ceiling/wall or melted.
Amzingly the house didn't catch fire, but all the wiring had to be replaced & about 90% of the electrical goods.

Anther good strike actually occured near the main telephone exchange. He got to work to find 5 of their computers were no longer running, pulled the covers off to find the strike had hit the phone lines & come in through that way. There were nice big black burn makrs around the modem sockets & when they pulled the motherboards out for a closer look many of the earth tracks had burnt up.

Sounds like you were pretty lucky with your close call.
:-)

---

Grant
Darwin NT

firstly i'll appologise because i've not read the entire thread, but i'll just say that if people are commenting on power comsumption as a negative thing, lets keep things in perspective, the amount of power unused transformers (switched on, but the device which needs the transformer is off or not in use, things like leaving electronics on "standby") use is obscene, and much much more than our computers (i saw some hard numbers once, but can't remember the URL for the data)

now these transformers are useless when their associated device is off/not in use
where as an active PC crunching for BOINC is doing something very useful (rather than just idling)

also in the interests of saving things, the greatest cost to the environment is in the manufacture and disposal of PCs
and because with most PC hardware (the monitor is the main exception) suffers from thermal stress more than it will from wearout, it's actually more benificial environmentally to leave them on all the time (they'll last longer)

also "power saving" features don't help because that increases thermal stress too (hardware being turned on and off lots) and also to start up say a hard drive, takes a fair bit of power, it's more efficient to keep it running

i'm not saying everyone should or shouldn't do certain things, or change their
habits, i'm just pointing out some facts, some of which aren't obvious

---

Want to search the BOINC Wiki, BOINCstats, or various BOINC forums from within firefox? Try the BOINC related Firefox Search Engines

Lee, No. It is more of a research / curiosity thread.

Another lightning story:
I was working at BK when lightning hit a tree near the drive through speaker. It split the tree in half and it fell over the driveway blocking the cars. It also traveled inside the store over the wiring. Blew every breaker in the panels, stopped the broiler from running which started a grease fire, and absolutely fried the registers. That was fun. :-)

---

Fear will keep the local systems in line. Fear of this battle station. - Grand Moff Tarkin

Lee, No. It is more of a research / curiosity thread.

appologies, in that case i'll refrain from posting untill i've read the whole thread :)

---

Want to search the BOINC Wiki, BOINCstats, or various BOINC forums from within firefox? Try the BOINC related Firefox Search Engines