Rosetta@Home - 03/20: Stats & milestones

[BlackMountainCow]

:beer: Please welcome Brad to [TA]R@H :beer:


[TA]R@H is now 267 members strong. That number excludes some additional 30 people combined in TeAm Enterprise and about 11 in the Rebel Alliance.




Need some stats? Take a look at Free-DC, BOINC-Stats and BOINC-Synergy

Need a general overview over Rosetta? Take a look at this video with lots of background info and interviews.

Got some questions? Try the Official R@H FAQ (work in progress).

Want to know something about the medical relevance of Rosetta? Take a look at this article.

Interested in alpha testing for Rosetta? Take a look at RALPH.




Yesterday's highest overall climber: Kaasbytes (3,688 spots)

Yesterday's highest in TeAm climber: rockettman (2 spots)


Yesterday's production for TA: 92,583 (day before: 88,984 = +3,599)

TA's RAC 86,599

TA's total score 41,569,416

TA's world rank is: #4

Yesterday's production rank: #2




Milestone Makers:

Superself passed 5,000 creds

BigShroom passed 5,000 creds





Yesterday's top 10 producers:

• uallas5 (13,284 creds)
  
• TeAm Enterprise (11,435 creds)
  
• BadThad (6,518 creds)
  
• Dennis (6,203 creds)
  
• Mark_F_Williams (4,163 creds)
  
  
• Spongebob_Fan (3,634 creds)
  
• TA_TheReaper (2,856 creds)
  
• aaronpriest (2,569 creds)
  
• Fardringle (2,078 creds)
  
• fusuikan (1,904 creds)
  
  
• Neurodog (1,852 creds)
  
• thehennegroup (1,786 creds)
  
• teriba (1,675 creds)
  
• deerslayer (1,661 creds)
  
• Thehive (1,579 creds)
  
  
• Pokey (1,565 creds)
  
• strider7 (1,464 creds)
  
• 3chordcharlie (1,421 creds)
  
• Rebel Alliance (1,392 creds)
  
• Climb4 (1,268 creds)
  
  
• TA_JC (1,139 creds)
  
• BigShroom (1,063 creds)
  
• rockettman (1,060 creds)
  
• jw.middleton (1,040 creds)
  
• OhioDude (984 creds)
  
  
  

SitRep for the TeAm and Rosetta News:

• Welcome to our TeAm, Brad! :beer:
  
• Yesterday, we had 84 active members (31.34%).
  
• Congrats to yesterday's milestone maker(s)! :gift:
  
• Official News: Rosetta@home is now at version 5.54. This version fixes the RNA mode to follow your CPU run time preferences, rather than always producing 30 structures. There are also some small fixes in the graphics. For more details, see this post.
  
  For those of you who are new to Rosetta (=our new members), there's no need to take any actions. Rosetta will download the new application automatically once your old work units are finished.
  

:beer::beer:

BMC with thx to CyGoR

 

[deerslayer]

Thanks for the stats BMC!

 

[BadThad]

Thanks BMC!

Anyone else notice that the new BOINC client seems to actually use your preferences? After upgrading all my machines to the new client, the "connect every 5 days" preference seems to be working because I have machines that are almost totally depeleted of work before they transmit. It's kind of annoying because my daily numbers vary a LOT. When my Core 2 systems dump, I get a huge daily surge, then it goes to crap for a few days.

Anyhow, I changed my preference to 3 days to try and smooth that out....at the cost of potentially running out of work during an outtage because the client will cache less work. Personally, I don't like how they did this in BOINC. Just let us specify the amount of work to cache and let the dang clients transmit everyday! :disgust:

 

[Rudy Toody]

Welcome to the TeAm, Brad!

Congratulations, Superself and BigShroom!

Thanks for the stats, BMC!

 

[Pokey]

Welcome to the TeAm Brad

Congratulations to the MM 's .

Thanks BMC. :thumbsup:

Also, welcome to all the new members I missed the past week or so.

While I may come and go, my herd just keeps on going, and going, and going...............

 

[Rattledagger]

Originally posted by: BadThad
Anyone else notice that the new BOINC client seems to actually use your preferences? After upgrading all my machines to the new client, the "connect every 5 days" preference seems to be working because I have machines that are almost totally depeleted of work before they transmit.

Short answer: "Too large cache-size"...

Long answer:
v5.8.xx has a new way of deciding if a wu is in deadline-trouble.
Computational deadline = report deadline - (Work Buffer size + 1 day + "switch between projects every N hours")
if Computational deadline < 0.9 * report deadline => deadline-trouble.

In a project there atleast 1 wu is in deadline-trouble, two things happen:
1; Projects wu's is run before other projects wu.
2; Project is blocked from asking for more work until not in deadline-trouble any longer, or idle cpu...

With Rosetta@home's 10-day deadline, the formula means max usable cache-size is 4.24 days.
If "needs" larger cache, run multiple projects.


BTW, if the purpose of a large buffer is to handle unexpected outages, it's always been a bad idea to set cache-size larger than 1/2 the deadline. BOINC now enforces this fact, but adds a little extra safety-margin, in case work takes longer than estimated...

 

[BadThad]

Originally posted by: Rattledagger

Originally posted by: BadThad
Anyone else notice that the new BOINC client seems to actually use your preferences? After upgrading all my machines to the new client, the "connect every 5 days" preference seems to be working because I have machines that are almost totally depeleted of work before they transmit.

Short answer: "Too large cache-size"...

Long answer:
v5.8.xx has a new way of deciding if a wu is in deadline-trouble.
Computational deadline = report deadline - (Work Buffer size + 1 day + "switch between projects every N hours")
if Computational deadline < 0.9 * report deadline => deadline-trouble.

In a project there atleast 1 wu is in deadline-trouble, two things happen:
1; Projects wu's is run before other projects wu.
2; Project is blocked from asking for more work until not in deadline-trouble any longer, or idle cpu...

With Rosetta@home's 10-day deadline, the formula means max usable cache-size is 4.24 days.
If "needs" larger cache, run multiple projects.


BTW, if the purpose of a large buffer is to handle unexpected outages, it's always been a bad idea to set cache-size larger than 1/2 the deadline. BOINC now enforces this fact, but adds a little extra safety-margin, in case work takes longer than estimated...

Thanks Rattledagger!

 

[uallas5]

Holy Cr@p! How did I get into 1st place? Must be a combination of a few not sending for a few days as Rattledagger describes combined with the few new machines I added. Unfortunately one of the servers I've been crunching on is about to go live, and that machine is my #1 cruncher

Thanks for the stats BMC.

 

[BadThad]

Originally posted by: uallas5
Holy Cr@p! How did I get into 1st place? Must be a combination of a few not sending for a few days as Rattledagger describes combined with the few new machines I added. Unfortunately one of the servers I've been crunching on is about to go live, and that machine is my #1 cruncher

Thanks for the stats BMC.

Man, sorry to hear you're losing a cruncher! You've been really pouring out the work, I've been waiting for you to pass me for quite some time. Anyhow, the TeAm appreciates what a great cruncher you are!