Harvesting & Reusing Idle Computer Cycles

Hustler writes "More on the University of Texas grid project's mission to integrate numerous, diverse resources into a comprehensive campus cyber-infrastructure for research and education. This article examines the idea of harvesting unused cycles from compute resources to provide this aggregate power for compute-intensive work."

Comments

electricity

[TedCheshireAcad]

Does anyone realize that running a CPU at 100% takes more electricity than running a CPU at 10%?

"wasted compute cycles" aren't free. I would assert they're not even "wasted".

Re:electricity

[Anonymous Coward]

The point is that they're not being used, and that they can be used for research. From the point of view of the researchers, who need these cycles, they are wasted.

Re:electricity

[TERdON]

Yeah, but it still draws a lot less letting a some computers burn some cycles, than you would have to use if you built a shiny, new, cluster. And you don't have to pay for the hardware either, because you already have it...

Re:electricity

[ergo98]

"wasted compute cycles" aren't free. I would assert they're not even "wasted".

No doubt in the era of idle loops and HLT instructions unused processor capacity does yield benefits. However from the perspective of a large organization (such as a large corporation, or a large university), it is waste if they have thousands of very powerful CPUs distributed throughout their organization, yet they have to spend millions on mainframes to perform computational work.

Re:electricity

[hotdiggitydawg]

That's a very valid point, we should not assume that this usage comes at no cost to the environment. However, the cost of building and running a separate CPU dedicated to the same purpose is even higher - twice the hardware infrastructure (motherboards, cases, power supplies, what else? monitors, gfx cards, etc.), twice the number of cycles wasted loading software infrastructure (OS, drivers, frameworks eg. Java/Mono). Add to that the fact that hardware is not easily recycled and the "green" part of me suggests that cycle-sharing is a better idea than separate boxes.

The next question is - who pays for the electricity then? University departments are notorious for sqabbling over who picks up the tab for a shared resource - and that's not even considering the wider inclusion of home users...

Re:electricity

[antispam_ben]

Does anyone realize that running a CPU at 100% takes more electricity than running a CPU at 10%?

Yes, I do, the same for RAM being accessed and for a hard disk drive when it's seeking. But this is insignificant compared to the overhead of the power supply, fans, hard disk drive spindle motors, other circuitry that runs continuously, and dare I mention all those fancy-dancy computer case lights that are popular now.

The incremental cost of these otherwise-unused cycles is so low that they can be considered free.

So someone prove me wrong, what's the electricity cost of running a CPU at full cycles for a year vs. running at typical load? What's the cost of the lowered processor life due to running at a higher temperature. Chip makers will tell you this is a real cost, but practically, the machine is likely to be replaced with the next generation before the processor has a heat-related problem.

Regardless, the cost is MUCH lower, in both electricity and capital, than buying other machines specifically to do the work assigned to these 'free cycles'.

Re:electricity

[Profane MuthaFucka]

First, figure the (Watts fully loaded) - (watts at idle) and call it something like margin watts. Then, figure out how much a kilowatt hour of electricity costs in your area. Say 7 cents.

Since a watt is a watt, and for rough purposes you can either choose to ignore or treat power supply inefficiency as a constant, you can get an idea of what it costs.

Chip: 2.2Ghz Athlon 64
Idle: 117 watts
Max: 143 watts
difference: 25 watts
Kilowatt hour / 25 watts = 40 hours.

It takes 40 hours for a loaded chip to use a kilowatt hour more electricity than an idle chip. Over a year, this will cost you $15.34 in electricity. Since your power supply isn't 100 percent efficient, it'll be more. Say 20 bucks a year.

Re:electricity

[AaronGTurner]

Those of us developing Campus Grids do take this into account in costing models!

Re:electricity

[Anonymous Coward]

Does anyone realize that a CPU runs at 100% all the time? It just depends on what it does.

Re:electricity

[Jeff DeMaagd]

My questions are in relation to the public distributed computing projects.

Who pays for that extra electricity? What if the program was poorly written and destabilizes the computer?

Few to none of the distributed computing projects don't factor this in. It's a nice way of cost-shifting, I think.

I think it is a good way for an organization to make better use of their computers though, I really don't want any part of it.

Re:electricity

[Smiffa2001]

What would be amusing was if global warming research was being done with the 'spare' cycles:

"Sir, we've completed the study and all the results are in. It's pretty shocking..."
"Go on..."
"Well, since we started, it's gotten much worse compared to before. The rate of change increased. We think it's the increased power use..."
"D'Oh!!!"

NOTE: Scientific accuracy might be impaired during the length of this feature. Thank you for reading.

Re:electricity

[kesuki]

So someone prove me wrong, what's the electricity cost of running a CPU at full cycles for a year vs. running at typical load?

I can't tell you a whole year, but I can tell you for a month. Alright let's go back to DivX ;-) days when it was a hacked M$ codec... I was paying $20 a month ($10 over the 'minimum') for electricity. The first month I started doing DivX ;-) encoding, from various sources... my monthly bill shot up to $45. So, $25 a month more than at idle, per computer. (this assumes you run at full load)

Keep in mind that's still pretty cheap, if you've got decent, fairly new computers you're getting a pretty good mips/$ ratio. If however you've got a lab full of pentium 2's and you're running all this specialized software on it, well frakly the cost ratio slides downhill fast. Especially when you consider you could replace 'just what needs to be replaced' to get those systems up to entry level semptron-64's for less than a year's worth of electricity.

The cost to benefit ratio of doing stuff like this entirely depends on the class of desktop computers you're running it on. So your point is only valid if the technology powering the desktops is less than 3-4 years old, otherwise it makes more sense to either A. upgrade the desktops and continue with idle cycle usage plan or B. buy a real cluster.

Re:electricity

[hazem]

What all of you working from the electricity cost issue are missing is that at most universities, money for capital is different than money for operations. Capital money is hard to get. An increase in your operations cost just kind of get ignored if they're not too big.

This has political ramifications.

The goal: get a great, powerful, cluster of compute power.

You can't go to the administration and say, "We need to spend $150k on a compute cluster". The answer will be "we don't have one now, and everything's just fine. No."

So, you, being resourceful, implement this campus-wide cluster system that taps spare resources. Power bills go up a bit - nobody cares.

Now, a couple years later, lots of projects are using the cluster. But the thing isn't working well because the power's not there during normal peak usage.

At his point you go the administration, "we're losing tuition-paying students, and several grants are at risk because our compute cluster is not powerful enough. We need to spend $250k on a new compute cluster.

And THAT is how you manipulate your operations budget to augment your capital budget.

Re:electricity

[Doc Ruby]

What about the power consumption to produce the hardware? That power is invested in a maximum count of cycles, amortized across the computer's lifetime. If the computer is at 10% CPU, the manufacturing/delivery power investment only pays off 1/10th what it would at 100%. So the question is, of course, the value of the extra 90%. Of course, if the value is greater than the costs of the electricity (a good bet), but less than the costs of manufacturing 10x more CPUs (running at 10%), then this approach is the best alternative. I'd say it is, but of course the entire value proposition depends on the value of the use of the extra, harvested cycles. Most investments require ongoing maintenance costs to return on the initial investment.

Re:electricity

[Rosco P. Coltrane]

Does anyone realize that a CPU runs at 100% all the time? It just depends on what it does.

No it doesn't. When it does nothing, it idles. Most, if not all modern OSes explicitely tell the CPU when nothing is being scheduled in the scheduler, and the CPU puts itself in low-power idle mode as a result. Look inside the Linux scheduler, in the idle thread code, if you don't believe me.

Most programs in an underused computer are waiting either for interrupts (which happen all the time, but for much less compounded time than idling), and for other programs to wake themselves. Therefore the idle thread is called often, and the CPU goes idle most of the time.

What you say was true in the DOS days, which was essentially doing polling in a loop when nothing was happening.

Re:electricity

[johansalk]

At 100% my fan draws 2 watts, at 100% my HD draws 12 watts, at 100% my cpu draws 89 watts.

CPU cycles are *A* if not *THE* major power burner.

Re:electricity

[seanadams.com]

You forgot the bit where you sell the cluster, and then lease it back from the company you sold it to - that way it comes out of the monthly current budget, and not the capital account!

Re:electricity

[Jeet81]

I very much agree with you. With summer electricity bills soaring I sometimes think of shutting down my PC at night just to save a few dollars. With higher CPU usage comes more electricity and more heat.

--
Free Credit Report

Re:electricity

[woah]

Yeah, but a lot of it is down to politics. Persuading the powers that be to set budget aside for a new cluster may be difficult. But sneakily (or semi-sneakely) running around with a usb key, installing stuff is easy.

Besides the pointy haired types wouldn't have a clue about such things as power consumption.

"Look, it's a screensaver _and_ it's doing something useful!.

Re:electricity

[arivanov]

Besides wasting more electricity you also drastically increase the speed at which the system deteriorates:

• On a cheap white box systems without thermally controlled fans the power supply fan is usually driven of non-stabilized voltage prior to it being fed into the 12V circuit. This voltage is higher when consumption is higher and the fan runs at higher revs and dies faster. The more power the system eats the quicker the fans dies. Result - dead computer and possible fire hazard.
• On more expensive "branded" systems with thermally controlled fans the speed of all fans is proportional to the power dissipation in the case. As a result on some brand machines the fan dies in less then 6 months at 100% CPU (Compaq P3 DeskPro) or the CPU is thermally throttled (Compaq P3 Prosignia and many P4 Evos). Result - performance at around 20% of the expected or computer requiring repair in around a year or less
• Nearly all modern motherboards have 20+ high quality electrolitic capacitors. If these blow up the bus gets noisy and the motherboard becomes useless. This is especially pronounced on miniITX and other small factor systems which tend to heat up very quickly to 45-50C inside. Running them at 100% round the clock causes the capacitors to start leaking in 6-9 months and the motherboard is a dead hunk of metal in a year or so.
• Ad naseum

If you add up all the numbers using spare CPU from desktops on an average campus does not make sense. You lose on the average 150+£ or so per system per year in electricity, repairs due to thermal failures and accelerated depreciation. Once you add helldesk and IT staff hours caused by the failures the numbers add up to 200£+. There is no way on earth you can get 200£ per year worth of computing power back so the numbers do not add up (at least for Compaq desktop gear).

Re:electricity

[dongshu]

Taking into account that (a 2 Gig Hz computer with 7200 rpm hard disk, a 100 M Hz bus etc) a keeping a computer up and running for a month at 100% CPU costs $10.00 USD, if I run my CPU at 10% saves me 9 dollars and getting my job done with 100% CPU faster (say by multi threading the app or something) or getting more jobs done faster fetches me 100s of dollars in revenue or wages...I would go for running a CPU at 100% :-)

Re:electricity

[Gumber]

You do realize that simply fabricating a CPU takes a lot of energy? There are energy efficiencies in reducing the need to buy additional systems.

Re:electricity

[SpikeSpiff]

Yep, for the Athlon 64s, the difference is published by AMD because of their PowerNow program. Peak thermal load is limited in the socket/motherboard spec, at 115 watts. The processors are right at the limit under 100% utilization. When running idle/powersaving, the CPUS run about 30 watts. It's a pretty dramatic saving.

Re:electricity

[blackSphere]

Just a couple questions:

What is the power draw at 100% load on a single computer as compared to the same computing power on multiple computers at lower loads?

In other words, how many CPUs are needed at 10% load to achieve the same computing power as a single CPU running at 100% load, and how does the power draw compare?

Now, if you were to run the same computation on one computer at lower load levels (say let something run overnight if available) how does the power draw compare?

I have a hypothosis: Power draw increases with the number of operations (the transitions between states are where the highest power draws occur) so I would speculate that you are using the same amount of power overall, unless you take into account the need for higher heat dissapation at higher transitions.

Can you point me to some references, or give me an answer on my previous questions.

Thanks

Re:electricity

[Anonymous Coward]

What would be amusing was if global warming research was being done with the 'spare' cycles

Putting your joke aside, look here
http://climateprediction.net/

Re:electricity

[AtrN]

Typically large organizations spend millions on mainframes to do I/O not compute and trying to move those types of things to PC clusters doesn't work without (a) adequate network infrastructure and (b) a distributed I/O system that scales. Some tasks can move, e.g. the obvious example is Google but they have rather unique constraints that make it possible, i.e. trivially parallelizable, no need to guarantee total correctness and a willingness to expose details of the distribution to applications (ref. GoogleFS API, MapReduce and Pike et al's language atop MapReduce).

People who do need lots of compute cycles need problems and, more importantly, solutions that are amenable to distribution over a cluster made from the office PCs, with its relatively slow and high latency comms. I.e. you need tasks with large amount of compute over relatively small amounts of data which don't need to communicate with each all that much. Not everything falls, or can be pushed, into that model.

Re:electricity

[buckymatters]

That's okay. I don't pay for electricity at my apartment.

Re:electricity

[TERdON]

What you write is very informative - but you miss one important point. The system may detoriate a bit faster, but they should have been constructed for that workload from the beginning - a computer should be able to run for full load for prolonged periods of times (gaming, workstation work containing heavy computations, simulations, etc).

Your points just more or less show that some engineer at the constructing company either is incompetent, making constructions that won't last, or the companies are purposefully making imperfect products. And I actually don't know which of the implications is worse...

Re:electricity

[dkf]

With some experience, even $250k is on the low-side for a serious compute cluster. Assuming you put in reasonable interconnect (needed for many interesting problems).

Re:electricity

[Bastian]

Right. But it doesn't use anywhere near 10% as much power when the CPU is only running at 10%.

And it consumes a hell of a lot less power than one desktop running at 10% and a second high-powered computing cluster.

If it leads to making better use of the computers the world already has rather than building new ones, noxious chemicals and all, I'm all for it.

Re:electricity

[Kris_J]

The extra electricity directly due to running the CPU at 100% is not the main issue. The main problems running a modern PC's CPU at 100% are;

• Noise from the fans that have to run faster
• Heat from the CPU

I tested the Distributed.Net client on a couple of student labs. I had to remove it because it was straining the air-con and making the rooms way too noisy. If I run it on my current laptop, the fan kicks in. If I run it on my old laptop it overheats.

The air-con in the server rooms is a whole different grade of product to the offices. I've already had to remove hardware from my office because the air-con can't cope. If all of our staff PCs ran at 100%, we'd have to do a very expensive air-con upgrade. (Of course, that wouldn't actually happen, we'd just all get hot more of the year and morale would fall a little more.)

Re:electricity

[kbjnash]

So someone prove me wrong, what's the electricity cost of running a CPU at full cycles for a year vs. running at typical load? I don't know. I turn mine off at night, but then I'm a cheap bastard.

Re:electricity

[AaronGTurner]

We are designing a Campus Grid (we have the middleware ready to go). Part of our cost calculations include energy costs. Rhys Newman at Oxford University has been doing some work in this area running jobs on PCs running cycle scavenging software, connected via metering boxes to the mains so that the usage can be monitored.

Given that most PCs in offices are left on and running (if idle) 24-7 then the typical utilisation of an idle PC is around 125W and one doing batch computing about 250W. The relative cost from normal practice would be different if the PCs were suspended, hibernated, or powered off overnight, of course.

Play fair on the resources

[Mattygfunk1]

I think it's great as long as they're careful not to impede on the user working. Done badly these applications get annoying if they are too pushy about beginning their processing before a reasonable user timeout.

Google's desktop search is one example where the timing and recovery back to the user is really done well.
__
Laugh daily funny adult videos

Re:Play fair on the resources

[Mattygfunk1]

I should add that I didn't mean to imply that Google's desktop search is doing a similiar style of mass-computing job as this grid will be used for, but it does do a similiar thing using processing cycles that would not be used for it's local indexing.

__
Laugh daily funny adult videos

Re:Play fair on the resources

[lorque]

If these applications are anything like Folding@Home , IBM's World Community Grid etc, there shouldn't be any problem. I haven't noticed any slowdowns since installing the latter, running Win XP. If the application is placed on 'low' priority, you should be able to do whatever you normally do withouth the application disturbing you.

Re:Play fair on the resources

[Misagon]

Yes, I have had that problem a couple of times. The campus workstation and network was overloaded with work and all my apps were really slow ... during scheduled lab hours even.
It annoyed me so much that I got up and got home to do the work there.

Re:Play fair on the resources

[InvalidError]

Setting thread priority to 'idle' is a simple way of keeping intensive tasks from abusing the CPU(s) whenever the user is doing something. The only time where this may be insufficient is for tasks with huge data sets that would cause major swapping whenever the idle threads wake up.

For tasks that are IO-bound by nature, the likes of NCQ will help a little but for moderately sized compute-intensive tasks (like multiple SETIs), having processes running at idle/nicest priority makes nearly no measurable difference in most benchmarks.

GridMP is a commercial distributed computing impl.

[ReformedExCon]

There are several non-commercial distributed computing systems, so the GridMP system isn't anything particularly new or groundbreaking. However, in companies that run very resource intensive applications and simulations, such a distributed system that uses unused CPU cycles has some serious applications.

However, the most critical aspect of this type of system is not just that the application in question is just multithreaded, but that it be multithreaded based on the GridMP APIs. To do such would require either a significant rewrite of existing code or a rewrite of it from scratch. This is not a minor undertaking, by any means.

If the performance of the application and every cycle counts, then that investment is definitely worth it.

Sure about that?

[brwski]

REusing idle cycles? Really?

Re:Sure about that?

[Anonymous Coward]

Sure, just imagine the karmatic synergy, not to mention product entropy, attainable from both using waste cycles and recylcing them!

Re:Sure about that?

[St. Arbirix]

REusing idle cycles? Really?

I had high hopes about this, until I realized they misused the term.

I was hoping they meant that I could give cycles to various projects and they'd keep track of how much I donated so that when I wanted to do something CPU intensive I could use their systems.

I'd expect something like, for everyone 1000 cycles I donated to their project they'd give me 100 cycles at 10 times the speed. That would be kind of handy if I were a 3D graphics artist and I only spent a few hours out of the week actually rendering images.

Imagine a world where companies like Pixar or Industrial Light and Magic let you run their modelling software on your computer so long as you donated cycles to their grid. Then, since the image rendering software is the same, you get back you 1/10th by sending your image specs to their systems.

If I recall correctly POV-Ray already has a system a little like this where you run a client that renders parts of other people's files for them. I think the big render houses and processing centers could get a lot of gain out of something like that.

Re:Sure about that?

[1tsm3]

Cool... now I can insert porn into a Disney animated movie!! :P

Re:Sure about that?

[SnprBoB86]

Yea, that headline confused the hell out of me. If a CPU is idle, cycles aren't being used.

For something to be REused it is generally a requirement that it have been used at least once pior ;-)

Spambots

[HermanAB]

are harvesting spare cycles all the time. I don't think there are much cycles left over anymore!

If I could only use this to improve rendering time

[waif69]

in Final Cut Express or Shake or DVD Studio by using the PCs that I have on the network to do the heavy lifting required.

So Is This The New Enron....

[Anonymous Coward]

storing our unused CPU cycles in a data warehouse somewhere so that, when a user pays enough, they can buy the needed cycles?

Like water, like electricity, like natural gas and shares of Enron stock?

Lousy title.

[Ziviyr]

Okay, I'm gathering up lots of unused CPU cycles. But if I were going to reuse a cycle, I'd probably want to reuse a cycle that did something...

Who wants to pay me for the past three years of my computer idle time?

Hey, pay for my idle time too, arcades aren't getting cheaper.

"Compute" should only be used as a verb.

[Anonymous Coward]

"Compute" as an adjective is just weird. Keep your creepy clustering terms to yourself kthx

Re:"Compute" should only be used as a verb.

[Stauf]

Can the computes compute how long it will be computing to compute the time for the compute's computing to compute? Or is it too compute-intensive for the compute resources to correctly compute?

Electricity vs cost of more machines and labor

[G4from128k]

Does anyone realize that running a CPU at 100% takes more electricity than running a CPU at 10%?

This is a very insightful post, but has two crucial counterarguments

1. Does anyone realize the cost of buying extra computers to handle peak computing loads?
2. Does anyone realize the cost of idle high-tech, high-paid labor while they wait for something to run?

The proper decision would balance these three (and other factors) in defining a portfolio of computing assets that can cost-effectively handle both baseline and peak computing loads. Idle CPUs aren't free, but then neither are idle people or surplus (turned-off) machines.

Re:Electricity vs cost of more machines and labor

[Alwin Henseler]

"The proper decision would balance these three (and other factors) in defining a portfolio of computing assets that can cost-effectively handle both baseline and peak computing loads."

You're probably right, but oh what a beautiful line of marketing-speak... If you happen to work in management or sales somewhere, write this baby down!

Re:Electricity vs cost of more machines and labor

[oliverthered]

but idle people or surplus (turned-off) machines don't contribute to global warming.

Re:Electricity vs cost of more machines and labor

[ScrewMaster]

If you can truly tell if he's right, or probably right, then it doesn't really qualify as marketing-speak, whose sole purpose is to make you think you know what is being said without realizing that you don't because in reality nothing was. But yeah, it's a nice line of verbiage, no kidding.

Re:Electricity vs cost of more machines and labor

[Anonymous Coward]

Unless you've gone off "toast" and "hot water" you're also contributing to global warming.

Re:Electricity vs cost of more machines and labor

[richy freeway]

But idle machines will still use power. Then the new cluster they have to build to do the job uses power. All the extra machines took power to build and used various nasty chemicals. All this will contribute to damaging the environment in some way. I'd rather just use the facilities I already have.

Re:Electricity vs cost of more machines and labor

[Poltras]

Worse, they contribue to global entropy, thus reducing universe lifetime... What are we doing?

Re:Electricity vs cost of more machines and labor

[Anonymous Coward]

but idle people or surplus (turned-off) machines don't contribute to global warming.

I beg to differ, idle people are generally lazy people. Lazy people would more likely drive somewhere than walk or cycle, and therefore they contribute to global warming.

And as to the machines, if not using a surplus machine causing a new one to be bought instead the production of that new machine also causes global warming.

Re:Electricity vs cost of more machines and labor

[bcrowell]

There are costs that fall on the person who's donating the cycles, and costs that fall on the person who's getting the benefit of them. Unless both people are in the same organization, operating under the same budget, it's not just a question of minimizing the total cost. In the typical situation, the cost to the donor needs to be almost zero, otherwise the donor isn't going to do it. Even in a university environment, one department may have a separate budget from another department. Or electricity may be provided from the campus without a budget charge to the departments, but other costs, like paying sysadmins, may be specific to the department.

Personally, I ran the SETI@home client and the Golomb ruler client for a while, but stopped because of a variety of factors:

1. It makes my configuration more complicated, and any time I buy a new computer or do a fresh install, it's one more chore to take care of.
2. I ran SETI@home for a while at work (on my own desktop hardware I brought from home, hooked into the network at the school where I teach), but I got scared when I heard stories about people getting fired for that kind of thing at other institutions. The network admins at my school are very uptight about this kind of thing, and don't have the same ethic of openness and sharing that most academics have.
3. If I run it at home, I'm paying for the extra electricity.
4. Most of the clients are closed source. I'm very reluctant to run closed-source software on any machine I maintain. You might say that the people who wrote the clients are trustworthy, well-known academics, not malicious Russian gangsters, but in my experience, most academics are actually pretty piss-poor, fly-by-night programmers. What if there's a security hole? Sure, the client described in TFA is supposed to be sandboxed, but how sure can I be that the sandboxing is really secure? I'm not normally particularly paranoid about security, but the rational approach to security is to weigh costs and benefits, and here the benefits to me are zero.

I think if grid computing is ever going to take off, it needs to become a capitalist enterprise. If someone would pay me a few bucks a day for my spare cycles, and the client was open-source, and there was close to zero hassle, I'd gladly do it. Remember, one of the good things about a free market is that it tends to be an efficient way to allocate resources.

Re:Electricity vs cost of more machines and labor

[slazar]

STOP GLOBAL ENTROPY NOW!!!

Re:Electricity vs cost of more machines and labor

[ThousandStars]

and here the benefits to me are zero.

You may want to consider using idle CPU power for something that will help, then, such as Stanford's Folding@Home or one of the cancer distributed projects.

Your other concerns are still valid, and I'm not trying to belittle those, but you may be able to provide more help to others and to yourself than you think.

Re:Electricity vs cost of more machines and labor

[haakondahl]

Huh? Whatever, sign me up for two.

Re:Electricity vs cost of more machines and labor

[Arren]

"...and don't have the same ethic of openness and sharing that most academics have."

?!?

That's some dry wit my friend.

Re:Electricity vs cost of more machines and labor

[seanvaandering]

...If someone would pay me a few bucks a day for my spare cycles, and the client was open-source, and there was close to zero hassle, I'd gladly do it.

Me too! However I don't think the powers that be, would allow anyone to be looking at the source code especially if they are paying.

i.e. "Oh look at this!"

if cycles>1000000 then cent=cent + 1

"Wait a minute... let me adjust this to oh... 10"

3. Profit!!!!

and so it goes..

Re:Electricity vs cost of more machines and labor

[Anonymous Coward]

That's just if they are running Windows!

Re:Electricity vs cost of more machines and labor

[ultranova]

You may want to consider using idle CPU power for something that will help, then, such as Stanford's Folding@Home or one of the cancer distributed projects.

I tried running Folding@Home. For whatever reason, running it even in batch priority (with patches from Con Kolivas against Linux 2.6) makes the system feel slow (which, of course, means that there is a bug in the batch patch, since it's supposed to make the process idle-only). Furthermore, if you try to kill the process, it keeps spawning new ones.

For these reasons I've stopped running Folding, and won't take part in another grid computing project anytime soon.

Re:Electricity vs cost of more machines and labor

[chefren]

Most of the clients are closed source.

The reason is that this way they can prevent cheating. If they open source it they can no longer trust YOU..

any similar fre solutions that can be implimented?

[BipinG]

---> Does anyone realize that running a CPU at 100% takes more electricity than running a CPU at 10%? ---
is it? i'm WAITING FOR OTHERS TO ROAR!

bty: do we have some OSS software of similar kind... that can be used for similar network (of frends... & within companies & its branches?)

I haven't seen any SIMILAR free & flexible solution that can be imprimented by anyone willing to...

Imagine

[Progman3K]

Imagine a cluster of those -
never mind...

Reused???

[LemonFire]

Does anyone realize that running a CPU at 100% takes more electricity than running a CPU at 10%?
"wasted compute cycles" aren't free. I would assert they're not even "wasted".


And neither are the computer cycles reused as the slashdot article would have you believing.

How can you reuse something that was never used in the first place?

Re:Reused???

[chris_eineke]

I have mod points, but whatever... you have to look at it on a grand scale.
Take the processor running in this box I am typing on -- an athlon64 3400+. Now say that we get about 3.4 GHz worth of cycles each second (that's what AMD tells you), running 24 hours a day, 365 days a year, for 2 years:

3.4 * 10^10 * 60^2 * 24 * 365 * 2 = 2.14 * 10^18 cycles.

Your CPU will only live so long. They usually break when the warranty has gone void. ;)

Re:Reused???

[codeguy007]

Now say that we get about 3.4 GHz worth of cycles each second (that's what AMD tells you)

You should have used your mod points and not made a fool of yourself.

An Athlon64 3400+ does not run at 3.4GHz but 2.2GHz. Thus you're whole calculation of computer cycles is wrong. 3400+ is a PR rating comparing the performance of the Athlon64 to a Pentium4 of 3.4GHz.

Re:Reused???

[stecoop]

You should have read the last part of his sentence. See the "that's what AMD tells you". The 2.2GHz in an Athlon64 3400+ dosn't mean that my 2.4Ghz P4 can calculate more then the Athlon64. Each clock cycle can execute a certain numbe of instruction "steps" so the calculation the grandparent is good enough for a rough estimate.

Re:Reused???

[ZosX]

Who's the fool now? The 3400 rating is actually based upon the performance of a 1ghz Thunderbird IIRC. So a 2000 would be roughly the equivavelent of a 2ghz Thunderbird, NOT a 2ghz Pentium 4, even though the processor is actually running at 1.6ghz.

But don't take it from me. From the horses mouth:

Section 2 The Model number

The model number is fairly straight forward the numeric code of the Core ID will give you the model number. In the case of the newer Athlon XP's it will be the PR rating of the CPU. For example the AMD Barton 3200+ would have 3200 as its model number and not its operating MHz. The older CPU's such as the Thunderbird and the Duron which do not have PR ratings will have their operating speed in the model number section. A Thunderbird 1.4Ghz will have a model number of 1400.

Re:Reused???

[codeguy007]

No that's not what AMD tells you. AMD tells you that a there 2.2GHz Processor performs about the same as a 3.4GHz P4. That makes the cycles calculation completely bogus as cycles from one processor don't compare with cycles from another.

Re:Reused???

[codeguy007]

That does not say that they base performance on comparison with a 1GHz Thunderbird anywhere. Thus you have provided nothing.

And I can guarrantee you there's a corrolation with the Performance Rating of 3400 and how it performs compared to a 3.4GHz P4. Whether AMD admits it or not, there's a reason why an Athlon64 3400+ Performs about the same or better than a P4 3.4GHz. AMD wants you to know that an Athlon64 3400+ runs about the same speed as a P4 3.4GHz so that the end user more easily compare these AMD Apples to Intel Oranges.

Re:Reused???

[ZosX]

A Thunderbird 1.4Ghz will have a model number of 1400. See above.

From the goddamned wikipedia:

With the demise of the Cyrix MII (a renamed 6x86MX) from the market in 1999, the PR rating appeared to be dead, but AMD revived it in 2001 with the introduction of its Athlon XP line of processors. The use of the convention with these processors (which are rated against AMD's earlier Athlon Thunderbird cpu core) is less criticized, as the Athlon XP is a capable performer in both integer and FPU operations, and manages to out-perform an Intel Pentium 4 at a PR rating equalling the P4's mhz. The Athlon XP (as well as the Athlon 64) PR rating scheme is not intended to be anything more than a comparison to the same family of processors, and not a direct comparison to Intel or any other company's processor speeds (in raw MHz) which most skeptics say isn't true.

Don't believe me now?

You sir are a troll in the worst way. I've already explained how their PR system works, but I think at this point you are just looking to create an argument. You are totally goddamned right there is a correlation. In an awful lot of cases, the AMD chip outperforms the P4 with a similar PR number. These two numbers have little to do with each other in reality though other than for strict marketing purposes.

Clock for the clock AMD has the P4 beaten hands down. Also, clock for clock, the Pentium III/M will beat the pants off of anything out there.

Re:Reused???

[codeguy007]

So what makes the wikipedia the be all and end all of information? A collection of user contributed information. I would hardly use it as proof in an argument.

Suffice to say however AMD calculates it's PR rating really doesn't change the fact that it's to provide a comparison between Athlons and P4. I can guarrantee you if intel released a P4 processor that changed that correlation, AMD would change their PR rating on new processor to match it. Of course now that Intel itself is going to a PR rating of sorts that all changes.

Re:Reused???

[Anonymous Coward]

Isn't the warranty usually based on the MTBF, which is based on a certain usage duty cycle? Isn't it likely that exceeding that duty cycle will cause the processor to fail before the warranty expires? The converse seems likely also. So, it really depends on how often you want to be sending in your processor for warranty replacement.

Re:Reused???

[weierstrass]

IIRC, a GHz is 10^9 cycles/sec, not 10^10.

Re:Reused???

[CrackedButter]

ZosX, (just out of interest) why the relationship change? :-)

Re:Reused???

[arete]

"about 3.4 GHz worth of _P4_ cycles each second"
there, happy? It is clearly what they meant...

And it's not a bad way to think about it, because the "3400+" numbers and the P4 numbers are the only numbers that ARE comparable.

Re:Reused???

[codeguy007]

"about 3.4 GHz worth of _P4_ cycles each second"
there, happy? It is clearly what they meant...

That's just wrong. A Cycle is a Cycle and simple commands still run at one Cycle. The advantage Athlon64's have is that they can do complex commands in less cycles than the P4. Some processors also allow you to run multiple commands in parallel as well. But still a cycle is a cycle and in some applications clock speed is still king. Especially ones that need to be run serially and complex piping won't help because it can't maintain the required order of operation.

GridEngine

[Anonymous Coward]

http://gridengine.sunsource.net/

Free and opensource, runs on almost all operating systems.

Spyware, Adware & Malware

[Krankheit]

I thought that was what spyware was for? When you are not using your computer, and while you are using your computer too, let your computer send out e-mail and perform security audits on other Microsoft Windows computers! In exchange, you will get free, unlimited access to special money saving offers for products from many reputable companies, such as Pfizer.

Wrong

[imsabbel]

What you are saying was perfectly correct even 3 years or so ago.

But case in point: My Athlon64 computer doubles its wallplug powerdraw (including everything:PSU, Mainboard, HD, ect) at 100% load compared to idle desktop (ok, cool%quite helps pushing idle power down).

The cpu IS the biggest chunck besides some high-end GPUs (and even those need MUCH less power when idle), and modern cpus need 3-4 times as much power under full load compared to idle.

Re:Wrong

[big tex]

Using the Lap-Burn-O-Meter (TM) as a gauge of overall power consumption with my Powerbook G4, I can definitely say that higher cpu cycle activities (encoding 1hr AAC files, for instance) increase power usage.

I could probably do something fancier by monitoring power draw with it unplugged, but my balls would be fried before I could tabulate the accurate data.

Re:Wrong

[kesuki]

What you are saying was perfectly correct even 3 years or so ago.
Hrm no.
no need to repeat myself

Running cpus at full load has made a huge difference in the cost of operation since the early pentium days. His point is that the cost of the 'electricity' is less than the cost of buying/powering new hardware specifically designed to do the work. Remember the electrical cost of the systems that are idle doesn't go away. those systems are on, anyways. Computer lab access is generally 24 hours a day, so the systems always need to be on, thus they always need to use power.

You are right that running under load can double or even triple electricity consumption (the CPU isn't the only piece of electronics in a desktop that has a 'power saving mode') the motherboard shuts down whatever it can, the PSU especially lowers rotational speeds on fans to reduce power, the PSU itself wastes less power on conversion etc etc.. but all that was just as true 5 years ago.

The fact of the matter is your main savings is on the hardware cost. Even if you consider that a true cluster is going to be more efficient than a distributed cluster, the fact that you're increasing electrical draw by buying said cluster without being able to reduce the number of idle systems is enough to offset the slightly greater electrical draw/mips ratio of distributed computing.

A big cluster has way more fans, and cpus, and many many high power server class PSU's, unless you're running it directly from a DC power generating station.

1st Grid Design: GNU Jet Fighter

[reporter]

Let's do something really interesting with this grid technology. Instead of participating in SETI, let's use this grid to design the first GNU jet fighter (GJF). Our target performance would be the Phantom F-4J, modified with a gattling cannon. We could design and test the GJF entirely in cyberspace. The design would be freely available to any foreign country.

Could we really do this stunt? I see no reason why we could not. Dassault has done it.

Dassault, a French company, designed and tested its new Falcon 7X entirely in a virtual reality. The company did not create a physical prototype. Rather, the first build is destined for sale to the customer.

Cleanup and maintenance costs?

[HermanAB]

Before you can use the idle cycles, you first have to remove all the spambots, spybots, adware and screen savers that are already running on these machines. Also, about ten seconds after the regular user comes back from lunch, the shiny new grid computing app will be broken and all the crap apps will be back, so the maintenance cost of this system will be huge.

CPU power consumption

[ergo98]

http://www.tomshardware.com/cpu/20050509/cual_core _athlon-19.html

60-100W difference between idle and full power consumption. That is not an insignificant amount of power.

Re:CPU power consumption

[Anonymous Coward]

Great link!

FTA: there is something that we can't really tolerate: the Pentium D system manages to burn over 200 watts as soon as it's turned on, even when it isn't doing anything. It even exceeds 310 W when working and 350+ W with the graphics card employed! AMD proves that this is not necessary at all: a range of 125 to 190 Watts is much more acceptable (235 counting the graphics card). And that is without Cool & Quiet even enabled.

end quote.

Bottom line, if you care about energy conservation at all, buy an AMD and don't sweat letting it run full-bore.

Re:CPU power consumption

[Anonymous Coward]

AMD's 90 micron architecture cpu seems to have a very low power consumption at load, near Intels idle power consumption (I dont know how independant test/results are) http://www.anandtech.com/cpuchipsets/showdoc.aspx? i=2249&p=13

Re:CPU power consumption

[Xandu]

But it isn't that expensive. Let's call it 100W extra. 24 hours in a day gives us 2.4 kWh per day. For a year, call it 876 kWh. Approximate cost of electricity (in Texas) is about 10 cents per kilowatt-hour. That's $87.50 per year. Let's assume that you can extract half of the computer's horse-power for your cluster (the rest is lost in overhead of the cluster software etc, and of course, whatever the actual user of the PC does, which is often just word processing, email, and surfing the web). For an extra ~$175 per year you get the equivelant of another computer.

If you wanted to get that computing power in a stand alone system, you'd not only have to purchase the PC (up front capital), but you'd have to pay more for electricity. From the reference link, only about 30% of a computer's power is used by the CPU, the rest is doing nothin'. The computers referenced, at full bore use 185W (best case). That's $162 per year at my 10 cent per kilowatt hour quote. Cheaper, sure, but by the cost of a computer? Not even close.

Of course, there are other (hidden) costs involved in both methods, of which I'm not including in my (overly?) simplified model. And I'll just brush under the rug the fact that this kinda assumes that the average secretary has a top of the line system to surf the web with.

Re:CPU power consumption

[Anonymous Coward]

It would be better to turn those lights off in the other room that you aren't using. Or turn the AC up 5 degrees. My tiny apartment has at least 640 watts just in light bulbs.

Re:CPU power consumption

[ergo98]

It would be better to... ..do every power conservation technique you can reasonably achieve with a marginal impact on quality of life. All of my bulbs not on dimmers are compact flourescent, and nonethless we turn them off when we're elsewhere. I don't run SETI@Home because I don't see those cycles as free, and secondly because I hibernate my PC when I'm not using it.

I'm especially sensitive to these things in the summer because it's 31C outside, and my AC is fighting to keep the temperature down, so every electric device is exascerbating the heat problem and making it that much more of a battle.

Re:CPU power consumption

[RicktheBrick]

If one does not control the computers where the computations are being proccessed than one does even know if any results will be returned. For this reason alot of redundancy has to be built in the system. I think the United Devices use a reduncancy rate of 5. That is they send out the work to computers until they receive results from 5 different computers. They compare the results to determine the correct results. Thus all cost of electricity should be multiplied by 5 to determine the true cost. They also put a time constraint on how long one can work on the problem before they give up on that individual. Because of the length that the computations take it usually means that the computer must be either be over a two giga cycle computer or that the computer be left on for a great deal of time each day. I know because I have a 1 giga cycle computer that does nothing but the protein folding problem and has taken over 10 days to proccess the problem.

Re:CPU power consumption

[Anonymous Coward]

A redundancy level of five is insane. You should never really need more than two, since you just send the unit out again if the returned results disagree and only if they disagree. And if they agree you can be very confidant they are correct

Re:CPU power consumption

[Anonymous Coward]

In New York City power costs 19 cents per kilowatt-hour. :(

Re:CPU power consumption

[Xandu]

In New York City power costs 19 cents per kilowatt-hour. :(

True, but I took numbers for the location in question, namely Texas. But the argument is nearly independant of the cost of electricity. As long as people who don't need super powerful machines continue to have them, and as long as computers consume much (over 50%) of their power regardless of CPU state (as they still do today), it makes sense to use the spare cycles. At least in some situations.

Re:CPU power consumption

[Xandu]

In the case of United Devices, a redundancy level of 5 is not out of the question. They are dealing with totally anonymous computers, and need to verify that their computations 'add up'. However, in the case where you have more direct access to the machines (and therefore better control and trust), as in the case of a University rolling this out (University of Texas in this case), they can probably skimp on the redundancy.

Re:CPU power consumption

[1110110001]

But it isn't that expensive. [..]

Oh come on. Like it's always just about the money. Electricity needs to be generated. It doesn't come magically out of that power outlet.

b4n

Re:CPU power consumption

[Anonymous Coward]

Five levels of redundancy IS stupid! Did you even read what i wrote? If you send out two identical units and they both return the exact same results than you can be 99.999999999% sure that the returned results are correct. It is so unlikely for any two random errors in two seperate machines to produce the exact same incorrect result as to be impossible. If UD is repeating the results 5 times I think thats just because they have TOO much processing power. If I were to design a distributed computing grid I would test each computer by sending out a special work unit out for the first unit they request. It would be designed expressly to test the hardware and if it passes then they get more work units and if they don't the client won't request work. Then each work unit would be sent to two random computers and the results compared (perhaps using a cryptographic hash?). If the results disagree then the unit is sent out to a third computer and the result is compared to the previous two. If it matches one, then we know the other one is wrong and we send out a new test unit to that computer that sent the erronous result to see if the error is repeated. This way we ensure very high reliability with only 200% instead of 500% waste.

Re:CPU power consumption

[Medevo]

Geeze, you are getting killed on power. We get ours for something like 6 or 7 cents CANADIAN per KWh.

Assuming its 7 cents, that works out to roughly $50 USD, a good 57% cheaper.

Nevertheless, I know that we do generate most of that, power with coal-fired generators, so overall not particularly great for the environment

Medevo

Re:CPU power consumption

[AaronGTurner]

These are the sort of calculations we are doing for out Campus Grid.

Re:CPU power consumption

[Xandu]

It varies widly (wildly) in the States. New York is like 19 cents, Kentucky is about 4. It just depends on where you are, how they locally produce electricity, and demand/infrastructure.

Re:CPU power consumption

[Anonymous Coward]

don't forget the cost of cooling (running AC more) that the added heat will cause.

For an extra ~$175 per year you get the equivelant of another computer.

Considering my computers tend to be used for 3-5 years. 3 * $175 = $525, 5 * $175 = $875. I think I could buy a computer for between $525 and $875.

Simple: put the user in control

[antispam_ben]

I think it's great as long as they're careful not to impede on the user working. Done badly these applications get annoying if they are too pushy about beginning their processing before a reasonable user timeout.

Even back in the Windows NT4 days I would put a long-running task to Idle priority and the machine would be as responsive as when the task wasn't running (though I don't recall running a disk-intensive task that way). I've noticed the badly written apps tend to be viruses and P2P software, crap you don't want to be running anyway.

Re:Simple: put the user in control

[Anonymous Coward]

In Windows 2000 and XP there is no such thing as Idle priority, its Low.

Re:Simple: put the user in control

[fa2k]

Yes, there is

Don't invent your own mouse trap

[mi]

It is almost a 'meme' -- when people start on projects like this, they tend to think, off-the-shelf software (free and otherwise) is not for them and they need to write their own...

PVM offers both the spec and the implementation, MPI offers a newer spec with several solid implementations. But no, NIH-syndrom prevails and another piece of half-baked software is born.

Where I work, the monstrosity uses Java RMI to pass the input data and computation results around -- encapsulated in XML, no less...

It is very hard to fight -- I did a comparision implementing the same task in PVM and in our own software. Depending on the weight of the individual computation being distributed, PVM was from 10 to 300% faster and used 5 times less bandwidth. Upper management saw the white paper...

Guess, what we continue to develop and push to our clients?

Re:Don't invent your own mouse trap

[dsci]

It is very hard to fight

Yeah, 'grid' or 'distributed' computing has become a buzzword. Many folks that see this as a panacea seemingly fail to realize:

(1) many problems that can benefit from parallel crunching are not suitable to so-called grid computing; they fail to account for the granularity of the problem and communication latency.

(2) parallel implementation of a problem is not unique; how you implement the parallel mapping to one architecture is not necessarily the best mapping on another. In other words, good, high performance parallel implementation is no 'black box' solution.

(3) many problems have unique requirements for parallel implementation; providing libraries for basic calls to the network layer may be useful, but globally useful parallel routines are probably not globally useful.

Just some thoughts I have every time I see an article about 'grid computing.'

Re:Don't invent your own mouse trap

[Coryoth]

MPI is great. I used to work at a shop that had a lot of Sun workstations. After doing some reading I managed to recode some of our more processor intensive software to run distributed across the workstation pool (automatically reniced to lowest priority) using MPI. As long as you managed to get a large enough workstation pool (which wasn't that hard, given how many people had one sitting on their desk) the distributed version was every bit as fast as standard version running on high performance servers.

In effect, using MPI and a bit of recoding effort, I managed to double the number of available servers.

Jedidiah.

Re:Don't invent your own mouse trap

[Anonymous Coward]

I did a comparision implementing the same task in PVM and in our own software. Depending on the weight of the individual computation being distributed, PVM was from 10 to 300% faster and used 5 times less bandwidth. Upper management saw the white paper...

Guess, what we continue to develop and push to our clients?

Don't take this personally, but is it possible that they chose the inferior approach because you did a lousy job communicating with the managers?

Having a great idea isn't enough. If you can't translate techno-speak well enough to get the message across to the decision makers, part of the responsibility for their bad decisions lies with you.

Re:Don't invent your own mouse trap

[Anonymous Writer]

Yeah, 'grid' or 'distributed' computing has become a buzzword...

Just some thoughts I have every time I see an article about 'grid computing.'

Just look at the post...

• "integrate numerous, diverse resources"
• "comprehensive campus cyber-infrastructure"
• "harvesting unused cycles from compute resources"
• "compute-intensive"

It looks more like a press release from a marketing department full of jargon and hype targeted at the general public rather than the technically minded. Anything that uses the word "cyber" just rings alarm bells. And anyone that says it either isn't into computers but wants to make others think they know more than they really do, or they are being pressured by upper management into using it. I'm surprised they didn't throw in a little "e-commerce", "scalable and robust", and "solutions".

Re:Don't invent your own mouse trap

[iamnotaclown]

I'm surprised no one has mentioned Condor. It can run serial or parallel jobs (PVM and MPI are supported), does checkpointing, scales up to massive compute farms, can talk to the Globus Toolkit, is multi-platform (Windows, Linux, Mac, Solaris, HPUX to name a few) and is open source.

Support contracts are available, but not mandatory.

Not affiliated, just a happy customer.

Re:Don't invent your own mouse trap

[Jeff DeMaagd]

I really don't see that as necessarily the cause. I mean, how hard is it to communicate "A is faster and better than B"?

I think it is more likely that the PHBs simply don't trust their resident techs, or have prejudices that prevent them from accepting this sort of thing.

Re:Don't invent your own mouse trap

[mi]

Don't take this personally, but is it possible that they chose the inferior approach because you did a lousy job communicating with the managers?

My immediate manager agreed with me. He reviewed my comparision 'white-paper' before it was published to the internal mailing list -- the tables in it spoke for themselves. Our in-house solution could only approach (not reach) PVM's performance, when the invidual tasks where very compute-intensive (when the communication delays mattered least).

In addition, our own 'daemons' -- using Java RMI -- are quite chatty even when not doing anything. Which means, large chunks can never be paged by the OS. It is all around much worse than PVM and, probably, any decent MPI implementation.

But the all-trumping counter-arguments were: "Well, this is, what we devised here and will be able to support, whereas nobody knows, what this PVM thingie is all about..." Followed by: "The two guys, who wrote it, are really nice people.."

Re:Don't invent your own mouse trap

[prefect42]

Indeed, Condor (and Condor-G) fits the bill nicely, and has been scaled up to seriously large installations.

The checkpointing doesn't work right with the intel compilers though last I checked, which is a bit of an arse.

Re:Don't invent your own mouse trap

[benjamindees]

I managed to double the number of available servers for quake. (hopefully)

Pre-grid harvesting

[davidwr]

Technology like this is interesting, but there are other ways to harvest spare CPU cycles.

Mainframes have used batch jobs that ran during the night forever.

10 years ago my college let you run batch jobs on any of the hundreds of Unix workstations. The catch is they would go to sleep when someone logged into the console and would only wake up when that person walked out. In theory this improved CPU utilization a lot. In practice it didn't, since there weren't that many people running jobs that took more than a few minutes to finish, i.e. almost everyone sat down at a terminal, logged in, did their thing, and logged out.

Heck, your PC probably does "janitorial services" like like antivirus scanning, indexing, disk-optimization, and other tasks when the CPU is relatively idle.

Re:Pre-grid harvesting

[Ziviyr]

Heck, your PC probably does "janitorial services" like like antivirus scanning, indexing, disk-optimization, and other tasks when the CPU is relatively idle.

CPU is not a big issue for drive based processes like those.

Parallels to the ethanol debate

[mc6809e]

How much energy does it take to harvest the energy?

How many cycles does it take to harvest the idle cycles?

Is the balance positive or negative?

Virtual Dual Processing

[OsirisX11]

I apologize in advance as this question will clearly illustrate my ignorance of these types of programs.

Is it possible for two different CPU-reaping programs, such as Folding@Home and Seti@Home to be doing the same math problems in some instances?

Say, although ludacris, bear with me, they both wanted to use computer to find out 5 * 7, and they both wanted to know 7 * 9. Could one computer run those sets of instructions once, and distribute the result to both programs to do in essence, double the work?

Re:Virtual Dual Processing

[Anonymous Coward]

Rapper Ludacris is working on virtual dual processing? Doesn't really seem like his forte. ... Oh, you meant ludicrous . My mistake.

Re:Virtual Dual Processing

[Anonymous Coward]

No. Also... ludacris?? You are an idiot.

Re:Virtual Dual Processing

[Anonymous Coward]

In theory? Yes, I'm pretty sure it's possible.

However, in practice, it's almost certainly more work than it's worth. You've got to have a LOT of code tracking what program wants what results, when it wants the results, etc. etc. Although it might work if they had large amounts of overlap, in other cases, chances are you'd spend a good deal more CPU power just doing the coordination than the sharing would save.

Re:Virtual Dual Processing

[OsirisX11]

Yepp.. I'm an idiot. :)

Re:Virtual Dual Processing

[OsirisX11]

It was a typo fucker. at least the guy above you was funny about it. Yeah I made a mistake.

Re:Virtual Dual Processing

[mabraham]

It is quite easy for a human to observe when the machines are doing the same computation, as in your example. In part, that is because the human brain is superbly adapted to do pattern recognition. It is very difficult to write an algorithm to beat the human eye/brain combination for recognising patterns. I would expect that a speed-up could only be obtained by storing the results of computations that took a long time to occur, because you have to pay the cost of building the database, storing it, and checking against it. In practice, however, the scientists are unlikely to be doing an identical calculation at a high enough level of abstraction to make it profitable.

Re:Virtual Dual Processing

[OsirisX11]

Thank you for a worthwhile comment.

Distributed computing less efficient

[imstanny]

Everyone is saying that the cost of making a machine to do the same process that can be distributed to a computer is overlooking a very crucial point.

Distributing computing processes to third parties is much more inefficient. The workload has to be distributed in smaller packets, it has to be confirmed & rechecked more often, and the same workload has to be done multiple times due to not everyone runs a dedicated machine or always has 'spare cpu cycles.'

I would agree that distributing the work load is cheaper in the long run, especially with an increase in the amount of participants, but it is not a 1 to 1 cycle comparison, and therefore it is not necessarily 'taht much cheaper', 'more efficient', or 'more prudent' for a research facility to rely on others for computing cycles.

Re:Distributed computing less efficient

[codeguy007]

I would agree that distributing the work load is cheaper in the long run

I think you have that backwards. Grid computing is cheaper upfront because you don't have the expensive of buying an extremely expensive serial supercomputer or a beowulf cluster. But it requires more administration, isn't as efficient powerwise. Thus you can end up spending more in the long run or just get no where near the same performance. (Unless you aren't paying the power bill for all the nodes)

Grid Computing makes sense for things like SETI online or DNETC where the cycles are donated. It also makes more sense if you want to Grid together big clusters because it allows you to leverage the high performance of several clusters for a combined project. However using your secretaries' computers at night really doesn't provide a great solution but it may look good to PHBs and shareholders.

Re:Distributed computing less efficient

[patio11]

I work in the R&D department at a Japanese technology incubator, and mocked up a demo of distributed computing (counting solutions for N-Queens for N=20+, for those that are interested). Our largest test was with 1300 computers. Some real numbers for you: the program running on one computer (Java for logic and the unused networking, one C++ module for the inner loop calculation) is 25% as effective as the world-record C++ code running on the same computer.

Yeah, that sucks. Just use the C++ code, which is GPLed anyhow. But we were able to get 1300 machines in various cooperating organizations which a) were all 2.0+ GHz machines on a problem which is processor bound and b) had an average load factor of less than 1% during the day. The vast majority are off-the-shelf Windows boxes which sit in computer labs that are 80% empty 80% of the time and busy themselves with displaying screen savers (I imagine usage patterns at the typical American university are similar). Even if 75% of our theoretical capacity slipped through our fingers, thats still harnessing many thousands of times more useful work out of those boxes than we were getting before we implemented the computer grid.

Re:GridMP is a commercial distributed computing im

[codeguy007]

It would be interesting to compare costs between a campus wide grid like this and a dedicated beowulf cluster. I believe you will find that the Beowulf cluster will still be more the efficient solution. Of course each situation would be different making it hard to get a true comparison.

Of course the grid will be less money up front, but I think you will find that performance to power consumed will be higher (Especially if you use a water cooled cluster). The Adminstration costs will definitely be higher as a campus wide grid will be a much more complex animal than a straight beowulf cluster.

Now onto the performance issues.

With such a grid you limit yourself in several performance related ways:

• Parallel Code. - To achieve good performance on a cluster with such slow communications (network latency), you need problems and code that lend themselves to lots of parallelization. If your code has too many serial routines in it, your performance will suck.
  
  
• Limited Network Bandwidth. Not only will a grid give you poor network latency which limits MPI performance but it also will provide very poor network bandwidth. This is a problem if you have large data sets to be processed.
  
  
• Heterogeneous Hardware - This is a major issue. Because a campus wide grid is going to be made up of Heterogeneous Hardware, you are limited in how much you can tweak your code for a specific processor. Sure you can create multiple sets of code for different processors and architectures but that takes time and you will never be able to optimize your code for all the different options available.
  
  Most Beowulf cluster are built with Homogeneous Hardware which allows for code optimization.
  
  Also having different speed machines means that results are going to be obtained at differnet intervals so you would have to limit or remove the interdependence between the nodes.
  

sunsource.net

[Jose-S]

This seems to be a new site, right? Found this in their FAQ:

Q: Will Sun make Java Technology Open Source? A: Sun's goal is to make Java as open as possible and available to the largest developer community possible. We continue to move in that direction through the Java Community Process (JCP). Sun has published the Java source code, and developers can examine and modify the code. For six years we have successfully been striking a balance between sharing the technology, ensuring compatibility, and considering the needs of a growing installed base of more than 2.5 million Java developers who depend on us. We are certainly evolving Java through the JCP to a model that works for all involved but that also ensures compatibility. Cross-platform compatibility has always been the key to Java's success and integrity; a notion we feel was protected by Microsoft's agreement in January 2001 to settle the lawsuit regarding Java technology.

I take it that's a 'no.'

Re:sunsource.net

[Anonymous Coward]

What do you mean by "no"??

If you spend a minute or two more to find out more before jumping into the conclusion, you will find:

http://gridengine.sunsource.net/servlets/ProjectSo urce

Condor?

[Anonymous Coward]

Why not use condor?

It seems like this article exists merely to hype the IBM clustering solution.

Re:GridMP is a commercial distributed computing im

[Anonymous Coward]

It is true that the Beowulf cluster would be faster/more efficient for dedicated tasks. However, using free cycles on idle computers is not quite the same thing.

The Beowulf cluster is designed to process the task using as many machines as necessary. The Grid computing system must be able to handle the computation being interrupted at any time and only process on machines that are available.

The two domains are related, but are completely opposite ways of approaching the problem. The Beowulf cluster is designed to tackle the problem head on by throwing sufficient power at the task. The grid system is designed to squeeze out whatever spare cycles are available from existing resources.

Re:If I could only use this to improve rendering t

[SlamMan]

If you have extra Macs, you can with DVD studio and Shake. Look up qmaster.

Heterogeneous Hardware & mathematical accuracy

[mosel-saar-ruwer]

Heterogeneous Hardware - This is a major issue.

The kinds of things that interest high-end computing geeks tend to be extremely sensitive to round-off error.

If you're trying to get accurate results by spreading calculations around among disparate machines that might deploy e.g. IEEE 64-bit doubles, IEEE 96-bit doubles [Intel & AMD], IEEE 128-bit doubles [Sparc], or various hardware cheats [MMX, SSE, 3dNow, Altivec], then trying to make any sense of the results will drive you absolutely bonkers.

PS: A good place to start in understanding the uselessness of e.g. 64-bit doubles is Professor Kahan's site at UC-Berkeley; you might want to glance at the following PDF files:

Matlab's Loss is Nobody's Gain

How JAVA's Floating-Point Hurts Everyone Everywhere

etc

wear & tear associated with running at 100% cy

[espek]

Is there "wear and tear" associated with running a computer at 100% CPU cycles all the time via one of these distributed computing programs like Folding@Home?

Will running these programs make my computer less reliable later? Shorten it's productive life (2-3 years)?

I have a Dual 2.0 Mac that I leave running all the time because it's also acts as my personal web server, and because it's just easier to leave the computer on (not asleep) all the time. I run Folding@home because I believe in the science and research and know that my contributions actually help good science. But the idea of wear and tear of the machine has crossed my mind and want to know what the negatives are to doing this to the machine (besides having to pay for the electricity).

P4 also doubles usage under load.

[Anonymous Coward]

My P4 conumes about 200 watts at the plug while under load, less than 100 while idle. All at a crappy power factor of 0.6.

Re:1st Grid Design: GNU Jet Fighter

[codeguy007]

Remind me to boycott that first customers airplanes. I really don't want to be used to alpha test a new aircraft.

Re:GridMP is a commercial distributed computing im

[Smart Teapans]

Good point. Having to rewrite the application to make use of a parallel MPI can be a pain. Condor is a free full-featured batch system that allows you to run apps on remote machines without having to recompile them

Re:wear & tear associated with running at 100%

[Ziviyr]

Only concern I imagine is when those apps hit the drive (wear and tear on the one important piece of non-solid-state equipment a computer has).

Or if you have an unstable cooling system that extra heat pushes over the edge. The latter would be good to know about anyways. :-)

Green fancy-dancy!

[TheStonepedo]

The voltage from my idle memory cycles goes through a series of capacitors and ICs to make my fancy-dancy lights blink so I won't have to buy new computers and waste power - and all of this is within a 133 MHz underclocked pentium box with 32 MB ram running linux.

I'm saving the world!

Re:1st Grid Design: GNU Jet Fighter

[DigiShaman]

Ya, let's countries such as China and N. Korea have such access to free engineering. After all, we want oppressive regimes to have as much power over their own citizens. I mean, when was the last time YOU could fly your own jet? Such gaps between non-democratic governments and it's citizens make much-needed revolutions that much harder to achive.

Re:wear & tear associated with running at 100%

[ericdano]

Um, no. I've been running Setiathome on my Dual 450Mhz Pentium III server for years. Like 6 years.

I'd never use a G5 for a webserver. What a waste! Go build a CHEAP PC and slap Unix on it, and use that. Cheap PCs are good for that.

I stopped using Setiathome a couple of weeks ago when I tried to use the latest version of FreeBSD 4.11. Boinc, the new client, seems not to run at all. Never connects to the server, nada.....:-(

Re:wear & tear associated with running at 100%

[linsys]

The short answer is Yes. The long answer is YES chances are it will shorten the productive life of your system..

As others have stated aboive running your system at 100% will increase heat etc... computers are like anthing else, run it hard enough and you will shorten the life span.

laptop cores are much better

[steve_l]

I saw some some posters from the fraunhofer institute in germany on the subject of power, with a graph of specint/watt.

0. all modern cores switch off idle things (like the FPU) and have done for some time.

1. those opteron cores have best in class performance

2. intel centrino cores, like the i740, have about double the specint/watt figure. That means they do their computation twice as efficiently.

In a datacentre, power and air conditioning costs are major operational expenses. If we can move to lower power cores there -and have adaptive aircon that cranks back the cooling when the system is idle, the power savings would be significant. of course, putting the datacentre somewhere cooler with cheap non-fossil-fueled electicity (like British Columbia) is also a good choice.

Re:laptop cores are much better

[hazzey]

and have adaptive aircon that cranks back the cooling when the system is idle

You mean like a thermostat?

Re:laptop cores are much better

[scottv67]

You, my friend, would definitely get mod points if I had them today. :^)

Never has a one-line post been more "insightful". :^)

Re:laptop cores are much better

[steve_l]

yes indeed :)

Actually, with some knowledge of CPU load you can do better; cranking up when the workload increases, scaling back when it goes down, and you can also distribute jobs to systems that are under less load.

the way speedstep cores work is that the final 25% of performance comes at the expense of 50% of the power consumption (as the core switches to max voltage). Its a bit like running a car at high rpms -you get power, but it is inefficient.

if your load balancer sends load to cpus running at below 70%-or-thereabouts load, you can do more efficient work.

no, i am not aware of anything that does this (yet), though a lot of the grid work is very power aware.

Re:1st Grid Design: GNU Jet Fighter

[nbritton]

How about we do something that's a little more pratical and useful such as finding new drugs that will cure cancer.

And the Pentium M??

[Grendel Drago]

Err, not precisely. Intel's Pentium M can create a system that draws 132 watts at maximum CPU load, and runs nearly as fast.

I've been buying AMD for about five years, but I think my next system will be a Pentium M. Just as soon as they're a bit cheaper...

--grendel drago

Re:And the Pentium M??

[Anonymous Coward]

Err, not precisely. Intel's Pentium M can create a system that draws 132 watts at maximum CPU load, and runs nearly as fast.

Incorrect. It does decently in gaming benchmarks, but for everything else it falls behind. I believe Anandtech had an informative article on this some time ago.

Another interesting article on rounding error.

[mosel-saar-ruwer]

In addition to Professor Kahan's site, listed above, you might want to read this article over at Sun [which references SPARC's 128-bit IEEE double, known as the "SPARC-quad"]:

Floating-Point Computing: A Comedy of Errors?

Unfortunately, I don't think it lists an elapsed time for the 128-bit calculation [only for the 64-bit calculation].

Re:wear & tear associated with running at 100%

[ergo98]

As others have stated aboive running your system at 100% will increase heat etc... computers are like anthing else, run it hard enough and you will shorten the life span.

I overclocked my Celeron 300a several years back to 450Mhz, requiring me to boost the voltage/power consumption, and thus heat. The common understanding was that the increased running heat increased silicon decay, but it does so at a rate that is irrelevant - e.g. that chip has long been sitting in my garage in a "gotta get rid of this" pile, so whether it would die in 5 more years rather than 10 is somewhat irrelevant.

Sorry. This is hardly news

[Moderation abuser]

Seriously. We're talking about literally a 30 year old idea. By now it should really be built into every OS sold. The default configuration for every machine put on a network should link it into the existing network queueing system that you all have running at your sites.

Re:Sorry. This is hardly news

[FooAtWFU]

By now it should really be built into every OS sold.

Forget operating systems that are sold, how about the ones that are free?

Re:Sorry. This is hardly news

[imsabbel]

Yeah.
Because its SUCH a great idea of having a default pipe for executing remote code without user intervention build into your OS.

You're Missing the Point

[kf6auf]

Your choices are:

1. Use distributed computing to use all of the computer cycles that you already have.
2. Buy new rackmount computers which will cost additional money up front for the hardware and then they have their electricity and cooling costs.
3. Spend absolutely no money and get no more computing power.

Note that the solution in this article is obviously not free due to electricity and other support costs, but it is undoubtedly cheaper than buying your own cluster and then paying for electricity and the support costs.

Cycle-Sharing

[blechx]

Woulnt it be neat to build a system for sharing idle cpu-time?

Just imagine the enormous amount of "wasted" cycles that could be utilized. If everyone that participated in this network installed software that was capable of computing certain things, maybe for rendering or compiling or whatever. Then when you wanted something compiled you just sent out lil chunks of work and the global net of cpus could compute this and send you back the results.

A ratio system could even be used to prevent misuse, like, the more cycles you contribute, the more you can use.

Are there networks like these already?

Yup.

[Anonymous Coward]

Been done before. Nowadays, the big thing is better power management. How 'bout scoring a little more for originality here: "Harvesting and Reusing Idle Computer Cycles with Blue LEDs and Multicolored Sugar Sparkles".

Pentium M Benchmarks.

[Grendel Drago]

Really? Looked like it did fine in media encoding as well; it's just the synthetic benchmarks (PCMark04) that it falls behind in.

--grendel drago

Re:wear & tear associated with running at 100%

[Anonymous Coward]

Several years ago I also bought a 300a and overclocked it to 450. The sales guy told me it would die within a year (probably a scare tactic), but the chip is still running just fine!

In fact, my sister inherited the machine, so with the amount of spyware/malware I'm sure she has installed, it's probably been worked pretty hard over the last several years.

Mmmmm

[Francis85]

Yummah... idle cpus.. mmmmm, i'll take four please, with extra CPU HTL topping, and a bit of thermal paste to the mix (drool)

Re:Mmmmm

[Ravatar]

Would you like some SSE3 with that?

Re:1st Grid Design: GNU Jet Fighter

[linzeal]

Well you are not going to be using much open source software to design it, so what is the point?

I don't "get" grid computing.

[vortex2.71]

Sorry, but I've never really understood the allure of grid computing. I am a parallel computer user (Sciborg at NERSC, the Earth Simulator in Japan, Klondike at ARSC) and I've never had the need for empty processor cycles on a network with a lousy backbone. Except for embarisingly parallel codes that farm out domain space and don't require processor communication, the whole bottle neck in parallel computing is the backbone of the network and not usually the lack or power of the processors. I guess it might be a novel tool for solving some 10-D integral via monte-carlo integration, but it doesn't do much for everyone else (e.g. the majority of us don't run embarisingly parallel codes).

Re:I don't "get" grid computing.

[vanyel]

Granted, the majority of applications can't really benefit from it, but there are still a decent number of applications that can, and as long as there are all these idle cpus that have nothing better to do, why not put them to work on the problems that they can contribute to? Save the tightly coupled parallel processing systems to work on the problems that require them.

Re:wear & tear associated with running at 100%

[Zendra Thon]

There is increased wear and tear associated with running a computer. However - in university environments, this may not matter. At the university where I did my undergrad work, and now at the current one where I work, all general student-use computers in labs are replaced on a three-year basis. At any one time, there is a huge glut of just-barely-not-newest computers to be had. So shortening the lifespan of these machines really won't matter. The lab boxes are on most of the time anyway, and will be rotated out before they break.

Wisconsin Condor

[mrm677]

The Wisconsin Condor Project has been harvesting unused compute cycles for over a decade. The software is free to use and deploy, and is used by various corporations including Western Digital and others.

Some of the big graphics houses

[ScrewMaster]

use all their regular workstations as part of their render farm at night.

Re:GridMP is a commercial distributed computing im

[codeguy007]

Yes but compiling an application to make use of parallel MPI is often necessary to distribute it. If I have an application that is serial, I still can't distribute it's load across a cluster running condor. I can send that specific application to a specific node to run and I can run multiple instances of the application but I cannot make the application run faster than a on a single box with similar hardware as the node. In fact it will run slightly slower because the process has to be moved from the master node to the node running it.

You will however get some advantage from cluster software that can migrate threads from a multithread application across a cluster.

Re:1st Grid Design: GNU Jet Fighter

[Anonymous Coward]

As much as I dislike Chinese and N. Korean governmental systems, I feel that it's important to note that revolutions need not be based on military actions. In fact, two of the most notable military regimes currently in power got there at the hands of a military revolution: Cuba, and China.

Perhaps something like the post-Cold War Velvet Revolution is in order. It's unlikely, but much preferable, compared to the bloodbath of a pitched battle...

Olde Stuffe

[Anonymous Coward]

Ye University of Texica be collecting paypre for the purposes ... what? What good is paper? What you write on it is important.

Wow, I've never heard of this idea before...

[Jugalator]

Hmm, where have I heard about this before again?
Exciting to read a paper on this fanastic new idea.

How about extra GPU cycles?

[davidwr]

When I'm doing pedestrian things - read anything but games, videos, or high-end graphics work - my graphics card is underutilized.

Wouldn't it be cool to utilize it to its full potential?

Even better, when the screen saver would normally in, just turn over the graphics card completely to the background process.

Imagine Seti@home running on your GPU.

PS: Ditto some other processors that aren't being used to their full capacity.

Re:How about extra GPU cycles?

[SETIGuy]

Imagine Seti@home running on your GPU.

We're looking for someone to help us implement this very thing.

Anyone up for the challenge?

Re:How about extra GPU cycles?

[scottv67]

Hey, that's a great idea. Why not submit your question to Slashdot to see what others think?

Oh, wait, it's already been done:
http://developers.slashdot.org/article.pl?sid=04/0 5/09/039204&tid=152

Re:How about extra GPU cycles?

[Duncan3]

Wow, why didn't we do that... oh wait, we did, years ago in fact. (everyone else thought of it too)

GPU don't have real math... yet. So instead of Folding@home you get "tossed on the bed"@home. Which is unfortunately useless.

Stay tuned tho. :)

Re:1st Grid Design: GNU Jet Fighter

[Anonymous Coward]

What a crock of shit. They're telling me that structural and flight testing has been rendered obsolete? And I suppose their manufacturing process is perfect (perfect in the strictest sense of the word)?
Damn marketing bullshit. A little bit of truth bending and obufuscation = megaimpressive press release.
IAAAE (I am an aeronautical engineer)

Electricity doesn't matter

[HermanAB]

If you want to do a certain amount of computing, then you are going to pay for the electricity anyway, whether in desktop use, or mainframe use, or dedicated cluster use. The savings come in the form of deferred capital expenditure, while increased costs will be incurred in maintenance and helldesk support. Electricity use is pretty much immaterial.

The most powerful computer that will ever exist ..

[nosfucious]

.. will be the collective computing power of all the PC's burning nearly-idle cycles just downloading porn.

Anyone that can harness that power, is made.

Could be win-win. Install "clustering software-x", download free porn. Legally.

distributed storage is where it's at..

[university_whore]

If a University really wanted to do something with all the hundreds/thousands of computers sitting around doing nothing for 90% of the day, they should look at distributed storage. There are thousands of highed ed institutions all over the country giving you incredible data redundancy. Most University Computers don't allow much local storage anyways. No more UPS losing your tape backups of financial bank records. The prOn industry has been doing this for years, but a more formalized approach might help.,

Re:distributed storage is where it's at..

[yuriismaster]

The only problem with this is security.

While distributing CPU cycles is anonymous enough (MULT 7 45, ADD 2 98) whatever, data storage is a whole 'nother thing.

You wouldn't want some evil person on the other side of the globe with a 'backup' of your personal financial records?

This would purely have to be in-house, and would kill bandwidth if implemented poorly.

BUAhahahaha... poor suckers...

[jnelson4765]

Our new buses are the exact same - designed in CAD - no prototype phase - first production models were sold.

And they are shit.

Flimsy, awkward, handle like a drunken whale, weak brakes, and parts you *physically cannot get to*.

There is a very good reason for prototypes - you get to see what breaks *before* you invest in production tooling and large material and parts purchases.

They're gonna lose their ass on that...

GPU cycles in Java or .net

[Anonymous Coward]

I'd love to have a Java vm or net runtime that runs on the GPU in my video card.

That would let me run stuff both on the main cpu + the gpu since the GPU does almost nothing 99.999% of the time (I don't play FPS anymore).

Turning cycles back in to electricity

[Anonymous Coward]

Hi!

This relates to my research paper, "Turning Processor Cycles Back In To Electricity" where I've showed that it's possible to output electricity from a computer through a judicious use of NOP's.

So if you don't save the cycles to begin with, we can always turn them in to electricity!

future movie plot

[MegaFur]

I can see some campus-based grid network thing being used in a future sci-fi "the-computer-became-self-aware-and-tried-to-kill- us" plot.

I am a sinner

[exp(pi*sqrt(163))]

So a while back our company shut down. For the last couple of months a bunch of us worked 3 days a week on making a graceful shutdown. During that period we had about 1500 2-3GHz CPUs sitting idle. I had about 2 days spare to work on writing code, and even on the days I was working there wasn't much to do. At the start of the shutdown period I thought "Wow! A few teraflops of power available for my own personal use for two months. And the spare time to utilize it. I could write the most amazing stuff." And what did I do? Nothing. I am a sinner. I have some excuses: I had to look for a new job 'n' all that. Even so, I could have done something.

So what should I have done with that CPU power?

Re:I am a sinner

[MontyApollo]

Until you decided on your own project, you could have donated the cycles to Folding@Home or some other project.

Re:I am a sinner

[Duncan3]

Actually, if you only havea day to contribute, none of the real science based projects (Folding etc al) have work units that small.

Re:I am a sinner

[exp(pi*sqrt(163))]

Now that I do have an excuse for. These were privately owned machines used to process sensitive data owned by our parent company. Running an application from an external source that communicates with a remote server would have been against company policy.

Re:I am a sinner

[toddestan]

So what should I have done with that CPU power?

Waited until no one was looking and tossed a few into the trunk of your car.

Re:I am a sinner

[exp(pi*sqrt(163))]

Asset tags everywhere. Well...almost everywhere...

Re:If I could only use this to improve rendering t

[ToasterofDOOM]

BURP is a project of a similar concept built off of BOINC. I'd link to it but I don't have it. Just Google it.

A marketplace for PCs

[G4from128k]

There are costs that fall on the person who's donating the cycles, and costs that fall on the person who's getting the benefit of them. Unless both people are in the same organization, operating under the same budget, it's not just a question of minimizing the total cost. In the typical situation, the cost to the donor needs to be almost zero, otherwise the donor isn't going to do it. Even in a university environment, one department may have a separate budget from another department. Or electricity may be provided from the campus without a budget charge to the departments, but other costs, like paying sysadmins, may be specific to the department.

Absolutely spot on! Something needs to align the costs and values of the exchange of CPU electricity and CPU value. For operations like SETI, the motivation to donate CPU time seems to be combination of coolness, charity, and some degree of apathy that residential electricity users have with respect to usage.

I think if grid computing is ever going to take off, it needs to become a capitalist enterprise. If someone would pay me a few bucks a day for my spare cycles, and the client was open-source, and there was close to zero hassle, I'd gladly do it. Remember, one of the good things about a free market is that it tends to be an efficient way to allocate resources.

Very insightful. Anything that has certifiable value (i.e., people can agree on what it is being offered and received) and can be "delivered" to others will end up in a marketplace if the transaction costs are low enough. With PCs and broadband, one could argue that this is readily possible. In fact, at some level, we already have a market for "spare cycles" in the form of the trade or renting of spammers' botnets. In the commercial world, IBM and Sun are working hard to make this a reality, too.

Of course, if anyone can "sell" their PC's cycles, then apartment owners will have to stop offering "utilities-included" leases. The day will come when some landlord discovers an apartment with a couple dozen PCs slurping down the kWh and raking in the money. If CPU time sells for just $0.25/hr/CPU (1/2 to 1/4 the resale value), a couple dozen PCs can bring in more than $50k/year.

Re:A marketplace for PCs

[Neoncow]

Of course, if anyone can "sell" their PC's cycles, then apartment owners will have to stop offering "utilities-included" leases. The day will come when some landlord discovers an apartment with a couple dozen PCs slurping down the kWh and raking in the money. If CPU time sells for just $0.25/hr/CPU (1/2 to 1/4 the resale value), a couple dozen PCs can bring in more than $50k/year.

And then all landlords will be millionares, apartments will require smokestacks to vent excess heat, and all apartment owners will be homeless. =P

Alan Greenspan will rebuke thousands of landlords for their irrational exuberance. I anticipate the great CPU Cycle crash of 2012.

Re:Heterogeneous Hardware & mathematical accur

[greg_barton]

How JAVA's Floating-Point Hurts Everyone Everywhere

That presentation was done in 1998.

That'd be seven years ago...

Ever heard of java.lang.StrictMath? Didn't think so. Been around since Java 1.3. Current version is 1.5.

Re:wear & tear associated with running at 100%

[Anonymous Coward]

hitting the hard drive may not be a bad idea... those deskstar/maybe travelstar failures were apparently due to fluid gumming up the head from extended idle operation without the head moving around. Apparently some later firmwares exercise the head to shake off excess fluid.....

CPU-sucking fancy Screensavers

[billstewart]

Some screensavers use almost no CPU, but many of them really scarf down the horsepower. Displaying random JPEGs is pretty cheap, and having them float around a bit can be done with large or small amounts of CPU, but applications like fractal-generators typically draw as fast as your CPU can handle. If you're burning the CPU anyway, might as well calculate the data that you want instead of pretty decorative stuff - especially if your grid-computer application runs at a low enough priority that the screen-saver outranks it, which is often the case.

This used to be a real annoyance when I had a Sun lab a decade or so ago - the server was a big SparcStation 2 with a great screen on it, and most of the clients were IPXs or Sparcstation 1s, and if somebody had running one of the really cool fractal screensavers as their X environment, and logged in to the server to do their applications, the whole network could dog out if they went to lunch or went home for the night without logging out first. The equivalent I put up with these days is that I run some Protein-Folding-At-Home variant on my desktop, and while I use a display-a-picture-every-few-minutes screensaver, my wife runs some flying-starfield screensaver, so if she was the last one to log on, there are fewer CPU cycles available to fold proteins with.

By the way, NEVER run grid-computing applications on laptops - they aren't made for the constant heat load, and obviously if you're using them on batteries, you'll find that they're not made for heavy frequent battery drain either (I learned this the hard way commuting by train back when I was running the GIMPS Great Internet Mersenne Prime Searcher....)

hmmm

[plbg32]

gee i wonder if it could work on my home network? it would make one heck of a gaming machine hehe!!!

How much are YOU willing to pay?

[dkf]

If these people were serious about grid computing, they'd offer a way for people to sell cycles. Altruistic donation may be good for some people, but the rest of us like to see real money for this sort of thing. If nothing else, covering some of the costs (and not just electric power) of having a machine would be nice.

Do read the article

[roystgnr]

This isn't just about parallel computing - in fact if you'll read the article you'll see that they're using MPI for handling parallelism! Grid computing isn't about reinventing inter-node communications - it's more about inventing inter-node scheduling.

Your cluster - is it so fast that you're never stuck waiting for jobs to finish? If not, then you could probably benefit from being able to borrow time on someone's larger system. Is your cluster so well-utilized that the load's always around 1? If not then you've probably got spare capacity that someone else could benefit from. The fact that both you and those others are using MPI is necessary but insufficient to allow you to cooperate.

Re:1st Grid Design: GNU Jet Fighter

[Anonymous Coward]

So they haven't even built one yet? What makes you think they planned on having ANY customers?

does that mean I can get subsidies

[Anonymous Coward]

from the USDA (or CAP, for those in Europe)? After all, I'm "harvesting"

Re:If I could only use this to improve rendering t

[Huntster]

http://burp.boinc.dk/
It is still very much in Alpha stage, but it is moving along nicely, and from my perspective does a very nice job. Not to mention, when it does move to production status, you have a built in userbase to render projects. Check it out, and post any questions you might have in the Message Boards. The owner, Janus, is very user-friendly.

Windows and ParallelKnoppix

[Quentusrex]

I would like to see a windows based program that gave MPI-like access to idle processor cycles. This way you could boot a program like ParallelKnoppix somewhere on the network and either run C, Fortran, Lisp based computational applications. This would be very useful for most IT staff.

When I was in school....

[fumcr]

we harvested cpu cycles by playing Quake3.

Re:1st Grid Design: GNU Jet Fighter

[hcdejong]

Right, as if CPU cycles are the limiting factor in designing new aircraft. Dassault may not have built a prototype, but they did have rooms full of aeronautical engineers who actually know what to do with all that computational power.

Dassault isn't alone in skipping the prototype, by the way; this is pretty much par for the course for commercial aircraft. When there are no significant new developments (hardware, aerodynamics [1]) to test, the only use for a prototype would be to validate the manufacturing process, which can be done in computers instead.

For military aircraft (where the technology changes much more between generations), prototypes and even experimental aircraft (X-planes, EAP) aren't uncommon.

1: yes, new aircraft do use new technology, but in commercial aircraft it's all incremental so the effect is predictable enough that you don't need a prototype.

Idle cycles aren't free

[jgoemat]

A lot of people probably don't know this, but a processor under load uses much more energy than an idle processor. One of my computers uses 100-105 watts when idle, but jumps up to 150-160 watts running a CPU intensive application. If electricity costs 10 cents a kilowatt-hour, that would be an extra $3 a month. If you don't mind donating that much money to your favorite distributed computing project, then go ahead, but know the facts. Actually shutting down your system when not in use will help much more, it only uses a few watts when turned off.

Really, no. A cycle is not a cycle is not a cycle

[arete]

First, I completely agree with you that it does depend a lot on what you're doing. For instance, last I heard _cycle for cycle_ the Pentium is still the king of integer - such as chess. But the crown is different for flops... Raw clock is meaningless and you are highly misguided. Furthermore, MANY operations are multicycle, and I guarantee you they are used on anything mathmatically intense enough to be worth sending out over the network.

Interestingly, you don't have to leave Intel to see this: the Celeron, "vanilla" P4, Xeon and Pentium M have a lot of good differences just within the current Intel x86 line. The Pentium M is awesome, for instance.

But I'll provide just a few examples of how a cycle is NOT a cycle.

Many of these only help you if you compile for that architecture or it does something fancy in the background to compensate - but you could certainly distribute a mixed exe that ran the appropriate binary for the platform.

- First, bitwidth:
64 bit addition requires 1 cycle on a 64 bit cpu, but at least about 3 on a 32 bit. 64 bit multiplication is MUCH worse on a 32bit machine. Similarly, 128 bit vector math is much cheaper on a G4 ("altivec") than on a CPU limited to 64 or 32 bits in that arena.

- registers: A CPU can only actually DO operations on values in registers. If you have more registers you can do much more complicated (longer-chained) operations without having to go to RAM or cache. This is intensely true on highly serial but complicated math and amazingly significant if the operation data actually fits in registers in one CPU and not in another.

- branch prediction and shorter pipeline depth. All other things being equal you want the shortest pipeline possible because it means you have the lowest branch prediction penalty. Coupled with the quality of your branch predictor, this makes a big difference. (Of course, things _aren't_ equal, and longer pipelines make it easier to physically build faster CPUs) Even if branch prediction is meaningless, the pipeline depth is still important.

- parallelization: _All_ modern computers let you run some multiple commands in parallel using multiple CPUs, cores, hyperthreading and/or multiple processing units. Many computers come with two CPUs. Some newer CPUs comes with two cores. Hyperthreading decreases the process switching penalty. Modern CPUs have separate integer and flop units, often more than 1. Clearly the quantity and efficiency of these multiple units would make a big difference.

At an absolute minimum, all of these things help you run the OS without interfering too much with your actual work. But since we're talking about stuff that's already being distributed over a wide network to multiple computers, on some level this work is clearly parallelizeable. Even if your second core can't help on your first 'chunk' you could likely be executing two chunks at nearly the same speed (barring other constraints listed here)

- cache(L1/L2/L3), cache prediction, RAM, bandwidth, chipsets. I'm not going to go into all the details, but suffice to say that the cores need data and code to function and unless your entire process fits in registers, they have to get it from somewhere. The arrangement of memory has a big impact on 1) how much work the CPU has to do to get information and 2) how much the CPU has to wait for that information.

- I/O - I know this is out of our case, but the CPU efficiency of IDE has increased dramatically, but there is still some variance from system to system and driver to driver. Furthermore, different network cards/drivers use significantly different amounts of CPU time to send large amounts of data. This is true even if the speed of execution is not I/O bound - it still takes some main processor clocks and the quantity varies.

Furthermore, this arbitrary driver code and any OS code - for instance - is definitely susceptible to traditional branch prediction, cache hits, etc - even if your main crunching loop did fit in registers.

I'm sure there's more, but I'm done for now.

Re:much-needed revolutions...

[benjamindees]

I mean, when was the last time YOU could fly your own jet?

Umm, never? Hopefully free software can fix this ;)

Seriously, though, why is it that bureaucrats wet themselves at the thought of Russian engineers being out of work, who could be snapped up by terrorists, but unemployed Americans can go fuck themselves?

Even more seriously, though, I've got an idea. Design a generic robot arm that can be used for manufacturing. I'm sure the patent for such a thing has expired. In fact, I'm sure prior art exists :)

Then we'll see what thousands of people churning out blueprints and prototypes can come up with. If the first plans are for windmills, it could be damn well self-replicating. It's high time that open source moved into meatspace.

Re:Really, no. A cycle is not a cycle is not a cyc

[codeguy007]

Yeah I know all that stuff but you missed my point. My point is that his calculation for the number of cycles produced by AMD Athlon64 3400+ is incorrect. And that's still the case. No matter what you say his calculation still uses 3.4GHz instead of 2.2GHz. It is therefore wrong period. Trying to justify it by saying that he mean something that he didn't say is complete BS. An Athlon64 3400+ does not produce 3.4x10^9 cycles per second but 2.2x10^9 cycles per second. So the free number of cycles that he calculated is incorrect. Justifying his vodoo math with the claims you made doesn't make it correct.

Also electrically if you want to get down to it --- A cycle is a cycle is a cycle. It's a pulse of electricity with leading and falling edges that provides timing to a digital circuit. The difference between processors is what they are able to do during a cycle or multiple cycles.

No.

[arete]

I love that you're not smart enough to notice that the OP started with 3.4 10^10 - which should clearly have been 10^9. So all of his math is off by a factor of 10.

But you're still troll enough to nitpick over the difference between 3 and 2 (a factor of 0.5) even though he clearly stated the units he was using.

I probably shouldn't feed the troll any more than that, but:

To focus more on the original post:
a) 10^18 is clearly the important part of his math. Changing the 2 to a 1 is really not that important to math that demonstrated going from 10^10 to 10^18.
b) he probably didn't want to go look up how many Ghz his Athlon was.
c) OP stated, VERY clearly, that he knew those weren't "real" cycles. Since it was not the important part of his post and it was a reasonable approximation, he left it at that.

the OP said:

"about 3.4 GHz worth of cycles each second"

He said "worth of cycles" not "cycles" which you've chosen to simply ignore. He did not say "worth of P4 cycles" which would have been more specific but was still clearly what he meant - because it's what AMD compares to.

I stipulate, with the support of my gp post that "a cycle" is (between CPU classes) a useless unit of measurement of "worth". Furthermore, in your parent post, you apparently agree with all of my supporting points.

I further stipulate that "a P4 cycle worth of work" is an imprecise unit of measure but much LESS useless than "cycle"
Just in case you think it's appropriate to disregard what his text said and only look at his figure, I'd like to also point out that: