View Full Version : whats better to -nonet with?
fizler
14th July 2001, 09:54
I'll be going nonet for a while, and i have the following systems to run:
1x600mhz amd
1x700mhz amd
1x800mhz amd
my question is, should i nonet with a 70+ or should i stick with something small? I nonet'ed a 8 for a day on the 600mhz and cranked out alot of genes and i got one with 57 genes, but not alot of wu's..anyone got an answer for me? thanks
fizler
Ridcully
14th July 2001, 10:59
Jed from our team (Picard) ran two different Work units on a dual PIII over the same period of time and his results were:
CPU1 did 4 x 99aa = 117 units
CPU2 did 34 x 21aa = 73 units
So I would have to say go with the biggest unit you can get.
phil
14th July 2001, 11:03
I agree, the bigger the better...over 99aa and you start to get slightly more credit than under the old stats weighting.
dnar
14th July 2001, 11:12
Originally posted by Ridcully
Jed from our team (Picard) ran two different Work units on a dual PIII over the same period of time and his results were:
CPU1 did 4 x 99aa = 117 units
CPU2 did 34 x 21aa = 73 units
So I would have to say go with the biggest unit you can get.
I assume the stats are now weighed more evenly, but, with small units, the CPU is wasted more often, to perform re-seeding and filtering. Add to this, when -nenet, the client STILL DOES SOMETHING with the network. I have not fully sussed out what it's doing, but it does stil try to talk with something, sometimes........
Stay tuned.:D
dnar
14th July 2001, 11:13
Originally posted by phil
I agree, the bigger the better...over 99aa and you start to get slightly more credit than under the old stats weighting.
In my experience, the 98aa and 99a units are the only ones I see crash....
Dave S
14th July 2001, 11:59
Originally posted by dnar
In my experience, the 98aa and 99a units are the only ones I see crash....
Had a 56 go titsup on me:(
ohms18k
14th July 2001, 12:23
I agree with dnar the 98's and 99's crash more. but i have crashed more often running in -nonet mode with various sizes.
dnar
14th July 2001, 12:26
Originally posted by ohms18k
I agree with dnar the 98's and 99's crash more. but i have crashed more often running in -nonet mode with various sizes.
That will be due the random number re-seeding..... The client must have a bug.....
fizler
14th July 2001, 12:32
i just upped 1 day of work from 3 computers.. took me down 19 places and im almost at 4,000 units.. i am going to get a big 80+ and nonet it for a while...
dnar
14th July 2001, 12:36
Has anyone had a WU bigger than 99aa yet?
verT
14th July 2001, 13:10
Personally I have not seen anything larger than the 99 but have heard of 120's but that is just heresay
LBaker
14th July 2001, 13:18
Originally posted by dnar
Has anyone had a WU bigger than 99aa yet?
I had a 102 the other week. Yesterday I got a 12 and a 16. I'm running the 16 on a old p75 so that should work out good.
LB
dnar
14th July 2001, 13:21
Originally posted by verT
Personally I have not seen anything larger than the 99 but have heard of 120's but that is just heresay
I tend to think your right. Tonight I setup an additional Linux client, a quick bit of scripting had it loading units continously. I saw units from 12aa -> 99aa.
fizler
14th July 2001, 13:33
how do i tell how many aa's a unit has?
Dave S
14th July 2001, 14:35
Originally posted by fizler
how do i tell how many aa's a unit has?
HiYa, when it starts it scrolls down....
1
2
3
4...etc:)
That’s how you can tell, I think!!
Hemi
14th July 2001, 15:21
Is that true that the number of positions is what you are looking for? If so, I have a 92 on one of my machines now...
Mule
14th July 2001, 18:06
Spot on Hemi.
You can also open input.inp in any text editor the first number you see is the number of amino acids
HTH
Mule
siggy
14th July 2001, 20:24
Also if you do not have the ability to scroll on the DOS window, you can count the number or letters that are listed with sequence. It will also equal the number of aa's.
Martyn
14th July 2001, 20:57
Or you can download electron microscope and it'll tell you, as well as configutr the switch for you etc.......
http://www.thegenomecollective.com/12aa.gif
DL from here:
http://homepages.waymark.net/~lperry/weather/download/em2.zip
Ridcully
14th July 2001, 21:13
I didn't find that the nonet switch worked every time with em2, so I just go with the wordpad method. :)
viperdog
14th July 2001, 21:36
I have had a 120 and it crunched forever on the athlon 1.35 but boy did it rack up units when it uploaded..
verT
15th July 2001, 00:33
OK so I stand corrected, if I take viperdog at his word and I will because he is an upstanding team member the 120aa beast does exist :D
siggy
15th July 2001, 08:46
Originally posted by Ridcully
Jed from our team (Picard) ran two different Work units on a dual PIII over the same period of time and his results were:
CPU1 did 4 x 99aa = 117 units
CPU2 did 34 x 21aa = 73 units
So I would have to say go with the biggest unit you can get.
I guess they are trying to stop -nonet on small genes.
The 21aa should have been worth 6.3 times 34 would have been 214 WU in the old world.
Guess we need to stay away from the small ones.
fizler
15th July 2001, 09:20
i picked up a 99, 85 and 99 last night and im going to nonet with them for 2 days ...anyone else want to go nonet and have a big upload with me?
Hemi
15th July 2001, 09:25
I'm -nonetting on an 85, 88, and 92. I'll probably do the same and dump it in a few days... :)
Dave S
15th July 2001, 13:17
I’ve been nonetting for 2.5 days, was going to dump @18:00GMT on Monday – could go a bit longer? Name it guys? Wot time & wot day?
fizler
15th July 2001, 13:35
lets all do a dump at like 9 tuesday night, thats 3 days for crunching!
Dave S
15th July 2001, 13:40
Originally posted by fizler
lets all do a dump at like 9 tuesday night, thats 3 days for crunching!
OK is this GMT?
ohms18k
15th July 2001, 13:43
I'm in,
is that 9 est
Dave S
15th July 2001, 13:46
Originally posted by ohms18k
I'm in,
is that 9 est
Err.....that’s 2 or 3am for me...OK i will try to stay awake!!!!
pelligrini
15th July 2001, 14:14
Ya'll might want to read this from Stephan (http://groups.yahoo.com/group/genomeathome/message/2493), From the Yahoo messageboard Tue Jun 26, 2001:
I would say that for the sake of the project, it's best to not run any one unit cached for more than a few days, since we've started to run a greater diversity of projects and we need to keep the wheels turning by getting results back for the latest proteins we send out. On the other hand, a stats race is a stats race, and sneakernetting endlessly is within the "letter of the law", though perhaps not the spirit.
Dave S
15th July 2001, 14:20
Originally posted by pelligrini
Ya'll might want to read this from Stephan (http://groups.yahoo.com/group/genomeathome/message/2493), From the Yahoo messageboard Tue Jun 26, 2001:
Thanks for letting us know pelligrini; i can’t upload over the weekend….only from Monday to Friday:(
pelligrini
15th July 2001, 14:34
My office machines are the same way. Sometimes longer if don't feel like going in for a week ;)
From what I understand it's not that big of a deal to -nonet for a while, but I don't see a reason to do it if you don't have to.
Dave S
15th July 2001, 15:00
Originally posted by pelligrini
My office machines are the same way. Sometimes longer if don't feel like going in for a week ;)
From what I understand it's not that big of a deal to -nonet for a while, but I don't see a reason to do it if you don't have to.
I here U mate & thanx i will have to check-out the Yahoo messageboard a bit more:)
vBulletin® v3.7.4, Copyright ©2000-2012, Jelsoft Enterprises Ltd.