Myths about folding and why they are untrue

chileman

Gawd
Joined
Jun 15, 2004
Messages
793
Don't know if something like this has been done before but I want to make a list of common myths and misconceptions about folding and how we can combat them when explaining F@H to someone. I know I'm having a hell of a time. If you don't have answers to certian ones I bet someone else will, so post it for discussion.

Mods (moose/dr_k13), if something like this already exist I'm sorry, just point people that way. I looked and searched around and couldn't find one of our own at least.


Myth: WIll it cost me a lot money in power:
Truth: At most a folding boxen will take around $5/month to run, but most people already run their PCs 24/7 so at most it will cost an extra $1/month. $1 is not much for all the great scientific research you computer is doing.

Myth: Folding could install spyware on my machine
Truth: After the initial client is installed, all Stanford will send you is a work unit that is only a few kilobytes in size, everything is encrypted and checked to be sure the content is from Stanford.

Myth: Folding will take up all my resources and it'll slow down my PC
Truth: Folding only uses spare cycles that you aren't using. It's CPU priority is usually idle (unless you set it otherwise). Your mouse and keyboard have more priority over your CPU than folding@home does. As for RAM, as long as you aren't running QMD's and other large memory hogg WUs, F@H is very easy on system memory, my machine I surf on only has 384mb and still crunches big WUs and I see no issues on it at all.

QMDs are very RAM demanding, these are usually recomended for dedicated folding boxen with at least a GIG of RAM, and currently only run on P4s. You must opt in for these WUs with the options WUs > 5mb and -advmethods both turned on.

Myth: My one computer can't make that much difference
Truth: Every computer helps more than you'd think. It would take the world's largest super computer hundreds of thousands of years to crunch all the folding work that needs done. But your computer can help take that load down. Just think if every computer on the planet was folding? We'd be done much quicker, we'd be that much closer to our goal...curing disease that claims thousands of lives each year. (someone might want to work on that one, it's the best I can explain it)

Additions from other people:

Myth: Running a CPU at 100% for weeks on end will "wear it out". (submitted by HTPC Rookie"
Truth: Explain to him that an electrical interface cannot wear out, it can only become damaged by extreme voltages, heat, or physical causes. I had a P3700MHz folding for 24/7, and it's still folding. Many people here have even older machines that still run. I think i'm near the top of the "I have an old crappy computer lying around" from the Mac "Lisa" in my garage that STILL RUNS. The only thing F@H can theoretically shorten is a drive lifespan, but considering how often the windows indexing service crawls the drive, if you disable that rarely-used feature, F@H will have no negative effect. (submitted by Shadowchild)


Anyway it's a start. Try and post people concern you'd ran into about folding and how you've helped them to overcome those concerns. If you have a suggestion for any of my myths/truths let me know. This would be a good threat to pin once it matures.

Fold on people!

 
As a correction, most work units are more than a few kilobytes. Some of the bigpackets can be a few meg. But the overal bandwidth usage is usually not that high especially if you are not running bigpackets. The client only connects to Stanford to download new work units and send back completed results.

 
One I have yet to convince my boss isn't true...that running a CPU at 100% for weeks on end will "wear it out".
 
HTPC Rookie said:
One I have yet to convince my boss isn't true...that running a CPU at 100% for weeks on end will "wear it out".
Explain to him that an electrical interface cannot wear out, it can only become damaged by extreme voltages, heat, or physical causes. I had a P3700MHz folding for 24/7, and it's still folding. Many people here have even older machines that still run. I think i'm near the top of the "I have an old crappy computer lying around" from the Mac "Lisa" in my garage that STILL RUNS. The only thing F@H can theoretically shorten is a drive lifespan, but considering how often the windows indexing service crawls the drive, if you disable that rarely-used feature, F@H will have no negative effect.
 
The claims about the cost are not true.

Even if you have yoru PC up and running 24/7 anyway, with Prescotts or AMD64 processors you will easily take up an extra 60 watts per busy CPU.

60 watts at 24/7 ends up being about $60/year, in addition to the cost of being up 24/7, if you don't do that anyway.
 
I have a pII that's still folding. about 2 years after we got i had it doing SETI, then (~1 yr ago) i started it folding. that's..idk...5+ years of DC work?
 
Great topic guys. It makes me proud to be in this with you all. Lets keep this bumped to it shows up on the main forum page. Maybe we can get one more person to read it, and come on in.

 
Martin Cracauer said:
The claims about the cost are not true.

Even if you have yoru PC up and running 24/7 anyway, with Prescotts or AMD64 processors you will easily take up an extra 60 watts per busy CPU.

60 watts at 24/7 ends up being about $60/year, in addition to the cost of being up 24/7, if you don't do that anyway.

The cost has actually been measured by a Kill-A-Watt and rated by kw/hr costs before. Here is the farm spreadsheet, which gives the cost per year in electricity at average national power rates. Note that this is for a whole system running folding, not the difference between idle and full. I believe that difference is somewhere around 6-10 dollars per year, depending on the hard drives, memory power, processor speed, etc. etc. Pentium 4's are about 40 dollars higher per year at full load, but even then, the cost is fairly negligible. The most expensive system to run max core utilization per year is a Xeon 3.6GHz, at $133.47 per year. Now, math time! 133.47 / 12 months = $11.13 per month, $11.13 per month / 30 days = $0.38 per day. All values are rounded up. That's 38 cents per day, based on national average. That's less than half the cost of a coke bottle (12-16 fl. oz.) at a vending machine, and less than a fifth of what some people pay for a cup of coffee at Starbuck's. Now, I'll admit that it adds up when you have many boxen running, but if you could power four boxes completely (no monitor) for sacrificing a Starbuck's coffee for a cup at home, or skipping it altogether, it all of a sudden seems like a lot less. So, if you're borging an office with 100 machines, it'll cost an average of $7,500 per year to run them with folding, and, ballparking a figure, $6,800 per year to do so without. The costs are hit when you have large farms that need to have A/C, which is where that 60W of energy starts to hurt. Needless to say, though, if you're in a position where you can afford 100 machines, 900 bucks is almost a write-off, and it's far far less than what most companies donate per year to community programs. So, yes, the cost is fairly negligible, but it does add up. Eventually.

Edit for another response:
SmokeRngs said:
As a correction, most work units are more than a few kilobytes. Some of the bigpackets can be a few meg. But the overal bandwidth usage is usually not that high especially if you are not running bigpackets. The client only connects to Stanford to download new work units and send back completed results.

Bandwidth analysis has been done before, and the few megabytes F@H sends back every few days (Most computers, I believe, take at least 2 days for a double-gromacs 600 pointer, which I believe is at 5.5 MB or so), is approximately equivalent, month for month, of opening up the CNN homepage once per day for one month. As stated above, things like this add up when you have many many machines, but the cost could easily be lowered by, say, blocking the CNN homepage or using a caching proxy.
 
That sheet is bullshit and yes I have a powermeter and use it.

65 watts for dual 3.06 GHz Xeon, with CPU load? The dual 3.06 Xeon (130 nm core) I had in the office took 129 watts idle, 190 one CPU busy, 228 both CPUs busy.

The cost difference between idle and busy will typically be $50-$60 per year.
 
Shadowchild said:
Bandwidth analysis has been done before, and the few megabytes F@H sends back every few days (Most computers, I believe, take at least 2 days for a double-gromacs 600 pointer, which I believe is at 5.5 MB or so), is approximately equivalent, month for month, of opening up the CNN homepage once per day for one month. As stated above, things like this add up when you have many many machines, but the cost could easily be lowered by, say, blocking the CNN homepage or using a caching proxy.

I was just referring to the specific WUs at the one time download or upload. Over time the bandwidth is virtually negligible but for the time you have to sit there on dial-up for some of these units is a pain. I was just pointing that out. I don't want people to think it only takes a second to upload or download a WU on a 28.8 connection.

 
Martin Cracauer said:
That sheet is bullshit and yes I have a powermeter and use it. 35 watts for a modern PC? In who's dreams?

The cost difference between idle and busy will typically be $50-$60 per year.
Since I've gotten back to my house, we've added a computer to the house, (A64, two HD's (60GB and 200GB, DVD+-RW, audio, 65 watt speakers, monitor from old computer (CRT), GeForce6600GT, powered by a 450W power supply), and replaced a P3733 with a P43.2 with HT(S775, DVD-ROM, 80GB HD, 10 watt speakers, 17in. LCD) a router, cable modem, and a cable box for the TV. Our power bill, total per month, including costs of AC, has gone up 50 dollars per month from the winter My computer, the first month it was here (no folding for a full month, burning in and such to make sure the stability problems I'd been having were fixed) accounted for about 30 dollars. The AC accounts for approximately 15 dollars per month, which was calculated last year before I had my comp back here, and with the old P3-733, which was not folding. I'd like to know how 5 dollars for two computers per year, plus networking gear and a few extra peripherals, comes out to 50-60 per year per computer.

And, if you're doubting my figures, I'd be more than happy to get my Fluke back and get some solid ratings for my PC only, no monitor. I'd also consider plugging some interconnects directly to my breaker and monitor my whole house usage for a few months and give you the graph readout for folding/not folding periods, and total cost of folding for a home user with 2 computers and a laptop.
 
SmokeRngs said:
I was just referring to the specific WUs at the one time download or upload. Over time the bandwidth is virtually negligible but for the time you have to sit there on dial-up for some of these units is a pain. I was just pointing that out. I don't want people to think it only takes a second to upload or download a WU on a 28.8 connection.
I understand, I was just throwing accurate information out to go into the list. Plus, personally, anyone still on a 28.8 connection deserves the pain induced by transferring files >.>
 
id like to fold but honestly have no idea what i need to do can someone point me in the right direction please?
 
Can we get some more Myths up here? I'm trying to pitch this to my principal and would like a nice solid list so that I may debunk anything he may conquer with.

Thanks!

/resurection
 
Myth # 15: This post really WAS on topic.

This message has been deleted by AtomicMoose. Reason: Off-Topic

MooseEdit: Do not attempt to impersonate me. Mkaythxbye.
 
Myth # 15: This post really WAS on topic.

This message has been deleted by AtomicMoose. Reason: Off-Topic

MooseEdit: Do not attempt to impersonate me. Mkaythxbye.


not having a good posting day mary03?
 
Myth # 15: This post really WAS on topic.

This message has been deleted by AtomicMoose. Reason: Off-Topic

MooseEdit: Do not attempt to impersonate me. Mkaythxbye.

He wasn't impersonating you, he was telling a new myth that you deleted an on topic post of his......
 
Huh? Kinda hard to impersonate someone when in the upper left of a post is ummmmm who posted it....
 
I think we better write it up anyway just to clarify things.

Myth # 15: Marty's post really WAS on topic.
Truth: You're kidding right?

on a more serious note there is probably a myth related to the actual usable science being generated by this project so:
Myth: Folding proteins really doesn't have much scientific benefit
Truth: Scientific advancement is pretty hard to measure however there is a good link here: http://folding.stanford.edu/results.html to some of the results the folding project has come up with.
 
Back
Top