DC Vault

[alan2308]

We just had a huge bump in SZTAKI points, which should be good for the Vault.

If you compare our team score with the total points for the WUs we've finished, they don't add up. Apparently the admins have been going in and manually assigning points to people that worked on the huge WUs but didn't get points because too many other people had errors. That's how I shot up from 1000 points to 4000 points, and I think the same thing happened for alan2308. I'm still not sure how everything is supposed to add up, but I'm not complaining about the extra points. (I could probably figure out more if I could read Hungarian. )

BTW, they seem to have fixed a lot of issues as well, and most of the SZTAKI units I've gotten recently have been relatively small.

That must be it. I got a nice bump in total credit without losing any pending credit. A quick spot check of my pending points looks like all those units are still out there being crunched. So I should still have ~700 more points sooner or later.

Right now I'm killing the SIMAP units while this batch lasts, and I see nomad8u is killing them even harder. Hopefully we'll see a good jump there in the near future unless other teams are thinking the same way.

 

[alan2308]

lol set majestic-12 to use my entire connection.. im catching up to you razor should catch up to you here in the next couple days.. doing about 1.8 million urls a day

I'm glad someones hitting majestic-12 pretty hard. I have enough reasons to worry about getting a letter from Comcast.

 

[sirmonkey1985]

Overall Team Position: (Up 2 places) 37


lol i have comcast too.. and they can bite me if they think they can cancel my internet.. until they setup a bandwidth meter they cant do anything about it.. they can claim its my responsibility all they want because it wont do them any good.. its their responsibility to have a way to monitor total in/out bound traffic.. its impossible for me to monitor all my traffic let alone the fact that i avg about 2 TB a month of traffic just on my home network..

 

[sirmonkey1985]

ok all systems are up and running and i ended up putting seventeen or bust on my sempron 3100+ @ 3.1ghz and im getting a little over 4 million cEMs/sec which is way better then my be-2350 was getting at 3.05ghz on 1 core..

 

[alan2308]

Overall Team Position: (Up 2 places) 37


lol i have comcast too.. and they can bite me if they think they can cancel my internet.. until they setup a bandwidth meter they cant do anything about it.. they can claim its my responsibility all they want because it wont do them any good.. its their responsibility to have a way to monitor total in/out bound traffic.. its impossible for me to monitor all my traffic let alone the fact that i avg about 2 TB a month of traffic just on my home network..

Tell that to everyone who's already been cut off. Especially those who were cut off before Comcast would even tell you what the top secret cut off point is. I don't like it any more than you do, but its their network, and they get to make the rules.

And no, it's not impossible to monitor your bandwidth usage.

 

[Kendrak]

ok all systems are up and running and i ended up putting seventeen or bust on my sempron 3100+ @ 3.1ghz and im getting a little over 4 million cEMs/sec which is way better then my be-2350 was getting at 3.05ghz on 1 core..

How is that? The Semprons are slower chips right?

 

[alan2308]

How is that? The Semprons are slower chips right?

Not necessarily. Depending on which model it is, Semprons may just have reduced cache and/or have x64 mode disabled. Socket A Semprons were just rebranded Durons and later rebranded Athlon XP's.

And not all projects are as cache dependent as F@H.

 

[sirmonkey1985]

How is that? The Semprons are slower chips right?

yeah and it didnt last long it was a glitch or something.. it spiked up to 4 million for a little bit and then started slowly dropping.. its been sitting steady at 2.2 million cEMs/sec.. guess my happiness didnt last to long damn L2 crippled processors.. maybe ill throw in my athlon 64 3600+ i got for free the other day.. i just dont feel like taking that system apart though..

Not necessarily. Depending on which model it is, Semprons may just have reduced cache and/or have x64 mode disabled. Socket A Semprons were just rebranded Durons and later rebranded Athlon XP's.

And not all projects are as cache dependent as F@H.

its a socket 754 sempron 3100+ on a DFI lanparty UT NF3 250Gb

 

[sirmonkey1985]

Tell that to everyone who's already been cut off. Especially those who were cut off before Comcast would even tell you what the top secret cut off point is. I don't like it any more than you do, but its their network, and they get to make the rules.

And no, it's not impossible to monitor your bandwidth usage.

east coast has been screwed from day one with comcast.. most of the people you see that lost their internet live on the east coast.. very rarely do you ever see some one lose their net on the west coast.. even during that sandvine thing.. we weren't effected over here..

on a good month i average about 600GB of total bandwidth.. bad month maybe 400GB.. been like that for going on 2 years now..


and your right its not impossible.. but unless comcast wants to buy me a router that has a bandwidth monitor built into it.. im not going to waste another dime on a router just to make them happy.. ive already gone through 5 routers in the last 3 years.. plus the amount of money comcast gets from us they should just keep looking the other way..


woohoo i might be getting my hands on a amd athlon 64 3200+ clawhammer to replace my sempron 3100+.. yippy!!!

 

[Razor_FX_II]

Yippie! I have been chosen as the The Lattice Project user of the day!
My profile will be featured on the The Lattice Project website for the next 24 hours.
http://boinc.umiacs.umd.edu/

p.s. - I'm running out of money to bribe all these projects.

 

[Razor_FX_II]

We now have a team [H]ard|OCP at EON.
I have submitted the new team to the DC-Vault.
Cheers!

 

[alan2308]

Just a random thought, when they said it would be a while before proteins@Home showed up on our Vault page, just how long were they talking about? Its been over a week now.

 

[Razor_FX_II]

Just a random thought, when they said it would be a while before proteins@Home showed up on our Vault page, just how long were they talking about? Its been over a week now.

Yeah I was wondering that to. He didn't say how long, just that sometimes it takes awhile.
Maybe we should request it again?

 

[alan2308]

boincsimap 2/3/2009 3:47:00 PM Message from server: (Project has no jobs available)

It looks like we've already picked them clean. Are you kidding me?

 

[Razor_FX_II]

boincsimap 2/3/2009 3:47:00 PM Message from server: (Project has no jobs available)

It looks like we've already picked them clean. Are you kidding me?

Hmm don't know what to make of it.
http://boinc.bio.wzw.tum.de/boincsimap/server_status.php

Edit: Maybe grab the last of the 900 left of The Lattice project
http://boinc.umiacs.umd.edu/

 

[alan2308]

Hmm don't know what to make of it.
http://boinc.bio.wzw.tum.de/boincsimap/server_status.php

Edit: Maybe grab the last of the 900 left of The Lattice project
http://boinc.umiacs.umd.edu/

Guess they must just hate me.

And as for The Lattice Project, I have it set to grab anything but GARLI units. All my boxen are daily drivers for someone, so I can't run those even with 2GB in every box. It looks like GARLI is all they're giving out since I'm not getting anything right now.

 

[Razor_FX_II]

Looks like The Lattice Project is out of work to!!

 

[sirmonkey1985]

lol you can blame me for simap.. i set up my client to grab 6 days worth of work.. so each of my systems has over 50 projects each in queue..

 

[Razor_FX_II]

lol you can blame me for simap.. i set up my client to grab 6 days worth of work.. so each of my systems has over 50 projects each in queue..

Great job! Someone on our team need to get the rest of them.
I seen there was 900 left so I set my work fetch to 10 days but they were all ready gone when it went to get them.

 

[sirmonkey1985]

lol i gotta make up for the last 5 days of january that i wasnt doing any work..

welp ive finally finished all the simap projects..


btw anyone know what time primegrid updates.. decided to start running it last night to help out since we are starting to lose places in there and the 181.20 nvidia drivers do not play nice with my overclock and F@H or RC5-72 but seem to run a little smoother with boinc..


MJ12node : v1.6.8 (.NET 2.0)
Platform : Win32 specific running on Microsoft Windows NT 5.1.2600 Service Pack 3
Total URLs : 2,604,510 (100.0%)
Successes : 1,697,795 (65.2%)
Not found : 63,025 (2.4%)
Timed out : 75,258 (2.9%)
Disallowed : 24,857 (1.0%)
Banned : 14,815 (0.6%)
DNS errors : 674,114 (25.9%)
Conn errors : 12,412 (0.5%)
Forbidden (403): 3,163 (0.1%)
Other : 57,791 (2.2%)
Retries : 51,026 (1.5%)
Uptime : 1 day 18 hours 31 mins 41 secs
Memory usage : 75 MB
GZIP requests : 579,833 (22.3% of successes)
GZIP saved data: 22,482 MB (35.3% of total)


not bad i guess.. now i just wish i had 100mbit id friggin make a killing on majestic-12 maybe ill set up majestic-12 on one of my dedicated servers.. see how pissed off my server host gets..

 

[alan2308]

lol you can blame me for simap.. i set up my client to grab 6 days worth of work.. so each of my systems has over 50 projects each in queue..

I really wish I would have thought to do that when the work became available. We made a nice jump with this batch of work, but its going to be a long climb when we can only crunch a few days every couple months.

 

[sirmonkey1985]

yeah you would think especially after this batch how fast it went that they would release more sooner.. but oh well guess thats how it goes.. suprising so many people stick with a project that is rarely updated..

time to install the 181.22 drivers see if it fixes my stability problem when using gpu clients..

 

[alan2308]

That link to the SIMAP server status sounded like there was still a pile of work ready to go, but according to the project front page:

Project status
The project has currently no work (see below).
The similarity matrix of the SIMAP project is currently up to date. All workunits from the last calculation period are distributed, thanks to all users for chrunchig them.
The calculation of new simap workunits containing the novel proteins from February 2009 (simap app) and their domains (hmmer app) will start around March 1st.

I bet if SIMAP got dropped from the Vault, all the work wouldn't get finished in a day and a half next time.

 

[Jathanis]

And as for The Lattice Project, I have it set to grab anything but GARLI units. All my boxen are daily drivers for someone, so I can't run those even with 2GB in every box. It looks like GARLI is all they're giving out since I'm not getting anything right now.

Ditto here - I tried it a bit, even just running them overnight, and I kept getting some error about waiting for memory allocation... I'm waiting fro those MARXAN units, so I can grab a weeks worth...

lol you can blame me for simap.. i set up my client to grab 6 days worth of work.. so each of my systems has over 50 projects each in queue..

I tried that, but already had a bunch of SHA-1 (damn those download a bunch at a time!), so I only got about 30 total between all three of my boxen. You just running the 3 desktop systems again, or the lappies, too?

I bet if SIMAP got dropped from the Vault, all the work wouldn't get finished in a day and a half next time.

Heh, I bet if we hadn't hit em so [H]ard, it wouldn't have been out in a day and a half this time. What say we aim for less than 24 hours to clean 'em out next time? If we all do like sirmonkey1985 did, the servers won't stand a chance...

 

[sirmonkey1985]

Ditto here - I tried it a bit, even just running them overnight, and I kept getting some error about waiting for memory allocation... I'm waiting fro those MARXAN units, so I can grab a weeks worth...


I tried that, but already had a bunch of SHA-1 (damn those download a bunch at a time!), so I only got about 30 total between all three of my boxen. You just running the 3 desktop systems again, or the lappies, too?


Heh, I bet if we hadn't hit em so [H]ard, it wouldn't have been out in a day and a half this time. What say we aim for less than 24 hours to clean 'em out next time? If we all do like sirmonkey1985 did, the servers won't stand a chance...

2 desktops running boinc seti/primegrid/WCG/SHA-1/cosmology both are dual cores in my sig..and also running a c2d 1.5ghz laptop but only 1 core due to temps same projects as the desktops.. sempron 3100+ desktop and a 1.7ghz pentium M are running seventeen or bust.. was going to run boinc on another C2D 2ghz laptop but the fans annoying as hell with its spin up and spin down.. plus its using vista 64bit so i didnt want to restart it to disable the driver signing so i could run rightmark cpu clock utility to undervolt it.. and drop the full load temps some..

 

[SazanEyes]

So the temperature outside is 28F and it's calling for snow, but my window is open just to keep the room from getting too hot. I don't think my condo's heat has turned on at all today, thanks to my new 1000W F@W "heater". I'm not looking forward to summer.

Windows XP x64 is working great, I've got all 8 GPUs folding and four Wieferich@home processes running. We should overtake Team Japan in W@H sometime tonight.

 

[sirmonkey1985]

im still having issues with windows stability and using gpu clients.. causes everything to lag like a mofo..



you know i was sitting here thinking about random stuff and an idea popped into my head that maybe we should work on creating a list of projects that shows the avg completion time for lets say a q6600 or something as the comparision processor.. i could atleast do it for a few projects but then again my systems are beyond slower then a q6600.. just one of those random ideas that popped up in my hollow head of mine....

 

[Tobit]

And as for The Lattice Project, I have it set to grab anything but GARLI units. All my boxen are daily drivers for someone, so I can't run those even with 2GB in every box. It looks like GARLI is all they're giving out since I'm not getting anything right now.

Bummer. You may want to try them again in the future though. The most recent batch released on the 2nd showed vastly improved memory management and performance. I think I did at least 20 of them and none of them used more than 512M of memory each. Unfortunately, we are out of work again but more work is coming.

To all - Speaking of Lattice, Adam just changed the credit-granting formula to BOINC's two_credit method.

two_credit()

Useful for 2 valid results.

It will average them if the claimed credits are close together. If they aren't, then it will compare each claimed credit against that computer's historical granted credit per CPU sec to see which one is claiming closer to their historical average. It will grant credit to both computers with the claimed credit that is closer to the historical value. This helps grant the most appropriate credit for situations where there is a computer that always claims too low or a computer that always claims too high.

 

[sirmonkey1985]

damn my pentium M 1.7ghz laptop keeps overheating running SoB.. runs about 10C hotter then doing boinc..


i also added milkyway@home and ufluid to my boinc manager.. so when my client updates we should get a good jump.. this computer has 17 waiting to update and my other system has 6 but its been stuck doing a clean energy project for the last 6 hours.. i have my boinc clients set to switch projects after every 1000 minutes..

 

[Tobit]

i set up my client to grab 6 days worth of work..

How do you do this? Is it a manual config file tweak you have to do? I'm not seeing any options in the manager that would accomplish this for me.

 

[Tobit]

Lattice Update - currently over 4K new GARLI work units available. These are only using ~520M of memory each and are taking me ~12H each to crunch @ 3.5GHz. Currently crunching 8 at a time.

 

[sirmonkey1985]

How do you do this? Is it a manual config file tweak you have to do? I'm not seeing any options in the manager that would accomplish this for me.

in your boinc client go to preferences then to network usage.. then under additional work buffer change it between .1 and 10 days.. default its set to 0.00 days but its actually defaulted to 1 day of work.. its really for people that have dialup and dont get online very often.. ..

 

[Razor_FX_II]

SazanEyes - and anyone else who wants to run EON try this: http://www.pubquizcentre.nl/DigiK-oz/dceon.aspx

 

[Jathanis]

you know i was sitting here thinking about random stuff and an idea popped into my head that maybe we should work on creating a list of projects that shows the avg completion time for lets say a q6600 or something as the comparision processor.. i could atleast do it for a few projects but then again my systems are beyond slower then a q6600.. just one of those random ideas that popped up in my hollow head of mine....

I started that for a lot of my projects to see what was best for my three systems, so I can contribute whatever data i have, too...

 

[Jathanis]

Lattice Update - currently over 4K new GARLI work units available. These are only using ~520M of memory each and are taking me ~12H each to crunch @ 3.5GHz. Currently crunching 8 at a time.

More update - They seemed to have released a handful of MARXAN units, and my work machine grabbed 4, but by the time I noticed & went back to grab more, they were gone - looks like a beta test batch... Anyone else seen some of these? The only unit to finish for me so far showed 0% progress & 13:50 to finish right up to the point it completed @ 18:04. As a bonus, it only used about 6.5 MB memory - no 1.2 GB requirement here!

edit: And I got a whopping 3.53 credit for it! Whohoo... I'll be blazing up the Lattice ranks at this rate...

 

[Mr Flibble]

If you look at the changes file in the doc folder for the new rc5 beta it lists the addition of an ATI stream core. I can't find a core for it in the app, but if you look up bug#4099 it has an ATI client that runs. For the life of me I can't find a way to upload/flush the finished results.

 

[metallicafan]

Did you get it to actually run on your ATI card? Or are you saying that it will run on the card just not flush the results?

I spent about an hour last night trying to read and research how to run the alpha MW@home ATI app but I couldnt quite find enough info to figure out how to do it. I have a couple ATI cards that I would like to run but . . .

I bet with 800 shaders an ATI core could tear through some rc5!

 

[Razor_FX_II]

More update - They seemed to have released a handful of MARXAN units, and my work machine grabbed 4, but by the time I noticed & went back to grab more, they were gone - looks like a beta test batch... Anyone else seen some of these? The only unit to finish for me so far showed 0% progress & 13:50 to finish right up to the point it completed @ 18:04. As a bonus, it only used about 6.5 MB memory - no 1.2 GB requirement here!

edit: And I got a whopping 3.53 credit for it! Whohoo... I'll be blazing up the Lattice ranks at this rate...

I got 1

 

[SazanEyes]

Razor, thanks for the EON link. I've just been trying to set up multiple clients, and that tool solves a lot of my problems. Are you using the suggested proxy settings?

Everyone, EON is another one of these projects that can easily run side by side with other jobs. There's rarely a lot of work, so most of the time the clients are idle. However, each WU is only worth one point, so when work is available you need to have a bunch of clients running to grab it. The tool linked above lets you run as many clients as you want, and has its own proxy to grab work and distribute it to the clients.

BTW...

36!

 

[Mr Flibble]

Did you get it to actually run on your ATI card? Or are you saying that it will run on the card just not flush the results?

I spent about an hour last night trying to read and research how to run the alpha MW@home ATI app but I couldnt quite find enough info to figure out how to do it. I have a couple ATI cards that I would like to run but . . .

I bet with 800 shaders an ATI core could tear through some rc5!

The app I downloaded from the bug4099 page ran on my 4670 card just fine but I couldn't get it to flush..