Answers to: Reasons for not using F@H.

[Ivoshiee]

There is a thing that a lot of people who said folding or using electricity is a environemental hazard and contribute to the greenhouse gases and that is how electricity is produced. Where I live, 95% of all the power generated by Hydro-Quebec is coming from hydroelectricity and the rest is from windmills and a single nuclear reactor. This kind of production is environmental-friendly so the argument of environment hazard is useless. The only reason they wanted us to save power is because it cost a lot of money to build new dams but if I pay for power, I decided what I want to do.

Go poke those who love car racing since they are producing a lot more noxious gases in the air

Ha! That is the funniest thing I've heard. I would ask them what they would choose if it was to be between saving a stupid peguin's home and their mother's life one day.

Same crap with stem cells and genetics. People think it's wrong to "play god" or whatever until they or someone they love needs a heart or something. Then you'll see how "green" and "virtuous" they really are!

And by the way, building a dam disrupts the ecology of the habitat and kills fishes.

In the grand scheme of things there is no difference if the dam preventing natural flow of a river will exist couple hundred of years or not - nature will still prevail.

[lshurr]

It comes as no surprise that a thread so entitled should devolve into a flame fest, though really tame compared to some I've seen. I did misc.headlines back in the day, and this is nothing.

Rather than contribute further in that direction, I'm going to return to the discussion of operating temperatures. I have two desktop machines which for my own purposes are already running 24/7 but have lots of idle time, some of which I gladly donate to F@H. There are personal reasons for the donation in addition to the simple fact that I can. As for the "some" qualifier, it so happens that I use the CPU Usage Percent slider on the Advanced tab of the Control Panel to limit CPU usage to 60%. Not out of a desire to withhold anything, but rather, in this way, to both manage the core temperature of the CPU's and the interactive responsiveness of my machines. Continuously running the CPU's (Athlon 64's) flat-out raises the core temps substantially, but at 60%, they are only slightly higher than at idle. Furthermore, though you may not agree that it is so, I notice a slight, but annoying sluggishness in responsiveness if F@H runs at 100%.

[sneakers55]

It comes as no surprise that a thread so entitled should devolve into a flame fest, though really tame compared to some I've seen..... Continuously running the CPU's (Athlon 64's) flat-out raises the core temps substantially, but at 60%, they are only slightly higher than at idle. Furthermore, though you may not agree that it is so, I notice a slight, but annoying sluggishness in responsiveness if F@H runs at 100%.

It was a lot more sluggish on the C2D 6400 until I cut F@H back to 95%, and it would BSOD occasionally. YMMV.

[CrimsonEclipse]

As for power consumption, I sold my desktop A64 4400+ and 23" LCD and bought a Vostro 1500.
It folds faster and used less than 1/4 the power. I like the idea of laptop folding.
Laptop <40watts
Desktop: well over 300 watts.

And my fan makes almost no noise

CE

[ppetrone]

Thank you for your post, John.

Needless to say, F@H researchers we have a big responsibility to our donors, to make good use of their time and resources.
Donors and scientists as a team, we all together push science an inch further.

Thank you for your trust and your encouragement and support.

Paula

[John Naylor]

Meep *is not quite sure what to say* Thanks I guess

[dricks]

Could we have a global ernery consumption World Map just to know how much this project does waste in term of energy?

Something like :
'USA : 3.2GW'
'Europe : 4.1GW'
...

And translate those into, for exemple, nuclear wastes by year?
CO2 emittings?
Just a thought.

Because the way this project works, each time a CPU is added to fold it's in fact a full computer that is running behind. With a graphic card, a network card, hard drives, optical drives and so on.
In a Mainframe, you would only add CPUs if you needed more horse power.
And the GPGPU initiative is even worse : graphics card consume a LOT of energy when pushed at 100%.

So, how many indirect desease because of this?

This project is perhaps the widest one, but is also one of the less energy efficient one

[John Naylor]

This project is perhaps the widest one, but is also one of the less energy efficient one

I think most people on here would argue that the enormous speed boost afforded by using GPUs (and PS3s) to fold would offset the extra power used, as we are speeding towards results at a much faster rate now then we were before.

[Ivoshiee]

Could we have a global ernery consumption World Map just to know how much this project does waste in term of energy?

Something like :
'USA : 3.2GW'
'Europe : 4.1GW'
...

And translate those into, for exemple, nuclear wastes by year?
CO2 emittings?
Just a thought.

Because the way this project works, each time a CPU is added to fold it's in fact a full computer that is running behind. With a graphic card, a network card, hard drives, optical drives and so on.
In a Mainframe, you would only add CPUs if you needed more horse power.
And the GPGPU initiative is even worse : graphics card consume a LOT of energy when pushed at 100%.

So, how many indirect desease because of this?

This project is perhaps the widest one, but is also one of the less energy efficient one

The Folding@home Wikipedia entry does have some wild speculations about the power usage. You should research a bit and add the country/continent numbers there. I am sad to inform you that computational biology, chemistry, engineering, ... can not escape power consumption. That is sad truth. Maybe it is possible to outsource all that computing to some far away country where a lot of people have no work and let them organise full time "human calculation" jobs. Before computers there were similar jobs, but we are likely not seeing anything similar any time soon. So all that talk about power usage is pretty much pointless.

[dricks]

I've already read the Wikipedia tiny lines about energy consumption of F@H, and found it at best funny but mainly ridiculous.

It's assumption about power consumption calculation, based on the most efficient computer is irrealistic.
His statement is that you could divide the global processing power by the avergate MFlops/watt of the most energy efficient computer in the world, which actually is the Blue Gene/P.
Let me think : is there anyone here using a Blue Gene/P solution for F@H? Or even, anyone in the World?

So, i don't think that using the World's most energy efficient mainframe to estimate energy consumption of this project is appropriate. At best, a joke.
And thus, that's funny because i was specifically speaking about energy efficiency and waste of energy, and my point was that while mainframes ARE energy efficient, the cluster of computers used by this project was far from it.

So, how efficient are traditionnal computers? 10times less than blu gene? 100 times less? must be somewhere near that.

So while Wikipedia's statement is 2.8MW/PFLOP i would say 280MW/PFLOP, which, using his number, "[40,000] 4 000 000 standard house light bulbs (between 60 and 100 watts each), or the equivalent of [0.5-3] 50-300 electrical wind mills depending on their size."

As you said, "computational biology, chemistry, engineering, ... can not escape power consumption" but at least, they can avoid wastes, while this project can't. (I say this project but i should say this kind of project, as seti@home had the same basis).

"So all that talk about power usage is pretty much pointless."
Hopefully, while the power consumption and energy efficiency does not seems to be of interest for you, it actually is for mainframe designers.
IMB set the energy efficiency of his blu gene solution ranked #4 in the top 10 reasons about buying their solution : "#4 Low power (~4-10X) — (green500.org), smallest footprint, lowest TCO**"

PS:
I hope it wasn't a stanford student that wrote the energy consumption part on wikipedia.

[bapriebe]

So while Wikipedia's statement is 2.8GW/PFLOP i would say 280GW/PFLOP...
And the GPGPU initiative is even worse : graphics card consume a LOT of energy when pushed at 100%.

It would help if you got your facts straight. WIKIPEDIA claims at least 2.8 MEGAWATTS per petaflop for a PC platform, not gigawatts. And 151Mflops/watt for an average PS3. This works out to 6.6megawatts/petaflop for a PS3, not gigawatts.

My main folding platform here is a 3Ghz Core 2 Duo E6850 with an ATI 3850. With the monitor off, it''s currently pulling 225W (according to my UPS) running a CPU client and GPU2 client simultaneously with both cores and the 3850 at 100% utilization.

I would be surprised if the entire project consumed much more than 50MW. For comparison purposes, that would be barely enough to run 300,000 TV screens tuned to Sex and the City. Which use of 50MW do you suppose is more beneficial to mankind?

[dricks]

So while Wikipedia's statement is 2.8GW/PFLOP i would say 280GW/PFLOP...
And the GPGPU initiative is even worse : graphics card consume a LOT of energy when pushed at 100%.

Kindly get your facts straight. WIKIPEDIA claims 2.8 MEGAWATTS per petaflop for a PC platform.

Right, my mistake it's not 2.8GW but 2.8MG, edited that.

But again, a blu gene is not what i would call a common PC Platform for F@H users, unless today's medium computers do have thousands of CPU....
http://upload.wikimedia.org/wikipedia/c ... P_rack.jpg

[dricks]

Based on the above numbers, it would appear the whole project consumes something like 60MW absolute max. Consider the general utility of this project compared with another use for 60MW such as, for example, 300,000 TV screens tuned to Sex and the City.

Perfect, you're right.
I'll throw away my lithuim-ion batteries in rivers, and garbages in forests. After all, there already are companies doing this, so little more is no hurt.

Kind way of thinking.


Instead, you could see this way :
This project is like if there was 300 000 TV screens turned on, but only 3000 people actually watching TV.
That's my point.

[bapriebe]

This project is like if there was 300 000 TV screens turned on, but only 3000 people actually watching TV.
That's my point.

Some people would consider 297,000 fewer idiots parked in front of the boob tube to be a great leap for civilization.

[dricks]

This project is like if there was 300 000 TV screens turned on, but only 3000 people actually watching TV.
That's my point.

Some people would consider 297,000 fewer idiots parked in front of the boob tube to be a great leap for civilization.

Others would consider that you could turn off 297 000 TV while still entertaining 3000 peoples. That's what i mean.
Using real designed mainframe to do this calculation would be 100 times more efficient, but actually it's just a mess.
A nice proof of concept about clusters but that's all. And having more and more people involved in it using F@H client is just more and more a waste. Can't you realize it?
Will you have to wait having 1 000 000 TV on for 10 000 spectators? duh.

I dont say that this project should stop, but that it should evolve to a more efficient one. GPGPU is a path to it? OK, but then there should be a limit to the number of CPU allowed to folding based on the number of GPU working to keep efficiency up.
Is there any decision on this way?