[Scotttheking]

I've been working on calculating how much power we have on each project, and total.
Courtesy of ars, I've got an idea of what a 1GHz Athlon can do each day.
All GHz is in athlon GHz.
All data was taken from our ranks score.

ubero = 103GHz
seti = 175GHz
rc5 = 363GHz
folding = 31GHz
total = 672GHz

Note that we seem more powerful because of the G4s on rc5.
I'm thinking we need to push this number up to at least 1THz.

[cdhostage]

We could improve this number by movng ALL G4s to RC5, but that would leave other pojects open. Ubero needs its many partifcipants or TCP will come up from behind.

WHy wouyld we measure ourselves in terms of Ahtlon=MHz anyqway? Let's ust keep going the way we have it. We're number one in Ubero and climbing in other projets. Good enough for me.

[Scotttheking]

Originally posted by cdhostage:
<STRONG>We could improve this number by movng ALL G4s to RC5, but that would leave other pojects open. Ubero needs its many partifcipants or TCP will come up from behind.

WHy wouyld we measure ourselves in terms of Ahtlon=MHz anyqway? Let's ust keep going the way we have it. We're number one in Ubero and climbing in other projets. Good enough for me.</STRONG>

It'd be great to move all G4s onto rc5, and all non G4s off rc5, but it won't happen.

I used athlons because it's a good processor for benchmarking, I have a bunch of em , and that's what ars used to benchmark all the big teams, so I wanted a comparable number.

[Grozni Majmun]

&lt;----LOOK, now I have 2 stars!!!

Originally posted by Scotttheking:
<STRONG>
I used athlons because it's a good processor for benchmarking, I have a bunch of em , and that's what ars used to benchmark all the big teams, so I wanted a comparable number.</STRONG>

So how does our number compare to ars or their individual teams?

[Grozni Majmun]

&lt;bump&gt;

So how does our number compare to ars or their individual teams?

Did you have info about ars strength somewhere? If you don't I don't mean that you should compile it, but if you already have it, please post it. Would be interesting to see.

thx

[Shaktai]

Originally posted by Grozni Majmun:
<STRONG>&lt;bump&gt;



Did you have info about ars strength somewhere? If you don't I don't mean that you should compile it, but if you already have it, please post it. Would be interesting to see.

thx</STRONG>

Ars Strength = Unbelievable. Thankfully they spread it out among many different projects so that we mere mortals stand a chance. Their sheer numbers of participants are staggering. However, as SETI Germany is proving, they are not invulnerable.

[reader50]

Most of Ars' strengths are in SETI, followed by Genome. They are vulnerable in all of the other projects. This is because much (most?) of Ars' power comes from corporate farms, where an Ars member has permission to run the client on 100+ computers. Typically, said member only has permission to run a particular client. Usually that means SETI@home, or less often, Genome. Those corporate farms could not be switched to another project to reinforce against us or other threats.

We beat them in Ubero, and could beat them in RC5 if it went on long enough. We have been running around 60% of their RC5 production, and could certainly outcrunch them if we wanted to. The main problem is their huge lead, RC5 will end before we could close the gap.

In Folding@home, we are at ~25% of their production, and we are just getting started. In time, the picture could change. We are not officially in distributed Folding yet, but it's another project where we might presently challenge them.

[Shaktai]

Originally posted by reader50:
<STRONG>Most of Ars' strengths are in SETI, followed by Genome. They are vulnerable in all of the other projects. This is because much (most?) of Ars' power comes from corporate farms, where an Ars member has permission to run the client on 100+ computers. Typically, said member only has permission to run a particular client. Usually that means SETI@home, or less often, Genome. Those corporate farms could not be switched to another project to reinforce against us or other threats.

We beat them in Ubero, and could beat them in RC5 if it went on long enough. We have been running around 60% of their RC5 production, and could certainly outcrunch them if we wanted to. The main problem is their huge lead, RC5 will end before we could close the gap.

In Folding@home, we are at ~25% of their production, and we are just getting started. In time, the picture could change. We are not officially in distributed Folding yet, but it's another project where we might presently challenge them.</STRONG>

Fascinating. That is good to know. <font color = red>We are not officially in distributed Folding yet, but it's another project where we might presently challenge them." </font> Did I read that sentence correctly or were you meaning to refer to something else, like Genome?

[Scotttheking]

If I ever get around to bugging the DF guys about altivec tweaks, we can move into it.
But right now it's damn slow on a mac, as far as I can tell. (slow as in slower then just the lower mHz)

Ars tried calculating theirs, and have somewhere around 2.6tHz?.
I'll look for it later.

[Shaktai]

Originally posted by Scotttheking:
<STRONG>If I ever get around to bugging the DF guys about altivec tweaks, we can move into it.
But right now it's damn slow on a mac, as far as I can tell. (slow as in slower then just the lower mHz)

Ars tried calculating theirs, and have somewhere around 2.6tHz?.
I'll look for it later.</STRONG>

Aahh! Now I understand. I didn't realize there was another project called Distributed Folding that was apart from Folding@Home. I found there website though. -- Never a day without learning something new!!!

[reader50]

Originally posted by Shaktai:
<STRONG>Fascinating. That is good to know. <font color = red>We are not officially in distributed Folding yet, but it's another project where we might presently challenge them." </font> Did I read that sentence correctly or were you meaning to refer to something else, like Genome? </STRONG>

Distributed Folding. There is a Mac OS X client (CLI only). I tested it, it works and is stable. It also has a quirk or two. Not sure how the Mac client compares to the other platforms, I don't have a non-Mac for comparison testing.

I started a team for us, in order to look over the stats situation. You can see it here if you like. The units shown are the ones I crunched while testing the client. Along with a few Scott tacked on after finding my "secret" testing.

I'm partial to setting up the site pages for dFold, along with all our usual stats. The main problem? There are missing resource pages on Ubero / Folding@home, which need to be written. On the other hand, not one person has asked for dFold, or even asked for information about it. If no one asks, I'll set up our official support after writing some of the missing resource pages.

If anyone wants it sooner, start yelling. Outyell the people waiting for resource pages on Folding / Ubero, and you will get the dFold sections much faster.

[Scotttheking]

Originally posted by reader50:
<STRONG>Not sure how the Mac client compares to the other platforms, I don't have a non-Mac for comparison testing.</STRONG>

It looks to be much slower.
I couldn't get it working on my linux boxes, but I checked with some other guys and it seems to be much slower on my Ti550 then on their systems, even accounting for the mHz difference.

[Shaktai]

Originally posted by Scotttheking:
<STRONG>

It looks to be much slower.
I couldn't get it working on my linux boxes, but I checked with some other guys and it seems to be much slower on my Ti550 then on their systems, even accounting for the mHz difference.</STRONG>

Fascinating. Just for the heck of it, I powered it up on my iBook 600 (now that I have the Athlon on folding) just to see how it does overnight. I may benchmark it later when I have time, and then compare it against my 1 ghz Celeron. Of course I am crunching for Team MacNN. Team MacNN is currently #196 out of approx 250 teams. After only a few minutes (less than 30), I contributed 843 structures.

I estimate my initial rate to be 100 structures every 2.5 to 2.75 minutes. Of course that is based on a very short run. I am going to let it run overnight, and then switch it back to folding tomorrow AM.

As far as resources go, l think we should focus on the projects at hand first. I may play with DF some, but really want to focus on Folding@Home right now.

[reader50]

Do not quit the dFold client with Control-C, Control-X, or by closing the Terminal window.

Always use "Q" as they suggest, and allow the client to shut down normally. If you quit this client with Control-C (or other forcible interrupts), you will lose all local work completed. Everything that hasn't been uploaded yet will be gone.

That was the main quirk that I ran across while testing.

[ 04-04-2002: Message edited by: reader50 ]

[Shaktai]

Originally posted by reader50:
<STRONG>Do not quit the dFold client with Control-C, Control-X, or by closing the Terminal window.

Always use "Q" as they suggest, and allow the client to shut down normally. If you quit this client with Control-C (or other forcible interrupts), you will lose all local work completed. Everything that hasn't been uploaded yet will be gone.

That was the main quirk that I ran across while testing.

[ 04-04-2002: Message edited by: reader50 ]</STRONG>

I did read that, and have used Q to quit so it uploads completed work before quitting. I am one of thos people who actually reads the read me files sometimes.....

My few short hours of effort using my slowest machine, my ibook 600 (roughly 6-7 hours), I turned in 16,104 results and Team MacNN moved from 196 to 190 in the team rankings. How does this compare with the results you got?

I have now moved the iBook back to F@H, so have a total of 3.2 ghz working on that. Not certain, but I think they may have introduced some new proteins. My Athlon got something called "proteinAnat" and it looks like it is going to take about 20 hours to process. I couldn't find it in the current project list.

[ 04-04-2002: Message edited by: Shaktai ]

[ 04-04-2002: Message edited by: Shaktai ]

[reader50]

Originally posted by Shaktai:
<STRONG>I did read that, and have used Q to quit so it uploads completed work before quitting. I am one of thos people who actually reads the read me files sometimes.....

My few short hours of effort using my slowest machine, my ibook 600 (roughly 6-7 hours), I turned in 16,104 results and Team MacNN moved from 196 to 190 in the team rankings. How does this compare with the results you got? ...</STRONG>

My results were built up over several days of on-off testing. I'm not sure how the rates compared because I was looking more at useability than particular speed. My tests were regularly interrupted by running assorted stats code, other clients, and of course, games.

As for the warning, I read the instructions also. But habits are hard to break, I lost perhaps an hour of work finding out why you should not use Control-C.

[ 04-04-2002: Message edited by: reader50 ]

[Shaktai]

Originally posted by reader50:
<STRONG>
My results were built up over several days of on-off testing. I'm not sure how the rates compared because I was looking more at useability than particular speed. My tests were regularly interrupted by running assorted stats code, other clients, and of course, games.

As for the warning, I read the instructions also. But habits are hard to break, I lost perhaps an hour of work finding out why you should not use Control-C.

[ 04-04-2002: Message edited by: reader50 ]</STRONG>

I think when I get a chance, I will try DF on all of my different machines to see what the performance differences are. Maybe I will do an all night crunchfest, where I can compare times directly between machines. It would be good info for if we finally decide to move in that direction. -- I know what you mean by habits. I nearly did the same thing, but caught myself just in time.

[slipjack]

Ok Reader, I'm officially asking for DF pages

Whenever you have time.

[mikkyo]

I put one machine on dFold just for fun.

[Scotttheking]

Originally posted by mikkyo:
<STRONG>I put one machine on dFold just for fun.</STRONG>

just one?
not one hundred ?

[Shaktai]

Originally posted by Scotttheking:
<STRONG>

just one?
not one hundred ?</STRONG>

Now you know that we really need mikkyo on Ubero and all those other projects! (as Shaktai fervently hopes that he can hang onto his meager lead for a little longer)

[reader50]

Originally posted by slipjack:
<STRONG>Ok Reader, I'm officially asking for DF pages

Whenever you have time. </STRONG>

You got it.

[slipjack]

Originally posted by reader50:
<STRONG>

You got it. </STRONG>

Now that's responsive team leadership!

[ 04-07-2002: Message edited by: slipjack ]

[SkiBikeSki]

I seem to be moving up the our DF team ranks pretty quickly. I'd love to be able to see my speed comapred to others, and see when i'll pass or be passed. Thanx reader.

[reader50]

Yesterday Active, and Yesterday Changes are now up. Hope that works for you SkiBikeSki, the Personals that you want will not be possible for another 3 days without a little more hacking. At that time we will have the one week of data on the project that are needed for Personals.

One of these days, I'll rewrite the Personals code modules so they don't need a full 7 days to become accurate. The rewrite needs to happen anyway to make personal Graphs and production tables possible. But there are a bunch of resource pages missing still, and personal graphs are not possible anyway because of the current server config (no GD library). I was going to tackle that rewrite after the next server PHP upgrade, which I have not managed to talk them into yet. Gotta hone up on my sales skills.

I'll probably try to hack up Personals sooner, but no promises.

[reader50]

We have fast service here.

SkiBikeSki, you're welcome.

Translation: Personals and Weekly stats are up.

[Shaktai]

Originally posted by reader50:
<STRONG>We have fast service here.

SkiBikeSki, you're welcome.

Translation: Personals and Weekly stats are up.</STRONG>

Thanks from me too.