Message boards : Number crunching : Rosetta v Seti times v credit
Author | Message |
---|---|
Ebola Bob Send message Joined: 12 May 07 Posts: 22 Credit: 19,770 RAC: 0 |
|
dcdc Send message Joined: 3 Nov 05 Posts: 1832 Credit: 119,860,059 RAC: 1,141 |
Hi Celeron's (except the 256kb tualatin ones) don't get particularly high scores on rosetta because it's quite cache and FPU dependant, which are two areas the Celeron isn't great at. I believe SETI can make use of SSE(?) which improves the thoroughput of the P4 family quite a lot I believe. At the moment, SSE doesn't help rosetta, although it may in the future as there are people working on it, but then there's a lot more code to rosetta than SETI. Having said that, that particular task did score quite low. There rosetta tasks aren't all the same though so scoring on the jobs varies due to a number of factors such as how big the task is and how well it fits into the CPU cache. Many other jobs will score much better on that same machine - here's one of mine that's identical to yours- sometimes it gets quite a bit more than claimed. Your Athlon should do quite a lot better than that here though as it's got more cache and a much stronger FPU. I think Rosetta gives less credit than seti even on like-for-like though. The positive is that the research is a lot more useful. Not that I think SETI is a waste, but as there's currently a limited CPU resource, and SETI has such a large share, rosetta will make use it much more effectively. Danny |
Ebola Bob Send message Joined: 12 May 07 Posts: 22 Credit: 19,770 RAC: 0 |
Thanks for the reply Your Athlon should do quite a lot better than that here though as it's got more cache and a much stronger FPU. Would you believe that the athlon is the machine I cannot get to download any wu's! I have adjusted the settings all over the place and not managed to get a single unit. Cheers anyway. Click below to see my werewolf |
dcdc Send message Joined: 3 Nov 05 Posts: 1832 Credit: 119,860,059 RAC: 1,141 |
that's strange! the daily quota is still 100/day so that's fine. What does boinc say in the messages tab? Also, you might already know all this, but does either of these in BOINC Manager do anything?: 'Retry Communications' under 'Tools' 'Update' under the Projects tab? |
Ebola Bob Send message Joined: 12 May 07 Posts: 22 Credit: 19,770 RAC: 0 |
It connects ok and seti and rosetta run fine on the celeron box and its 6GB HDD. In the messages I get: 19/05/2007 10:29:39|rosetta@home|Sending scheduler request: To fetch work 19/05/2007 10:29:39|rosetta@home|Requesting 172800 seconds of new work 19/05/2007 10:29:44|rosetta@home|Scheduler RPC succeeded [server version 509] 19/05/2007 10:29:44|rosetta@home|Message from server: No work sent 19/05/2007 10:29:44|rosetta@home|Message from server: (there was work but you don't have enough disk space allocated) 19/05/2007 10:29:44|rosetta@home|Message from server: Not enough disk space (only 81.2 MB free for BOINC). Review preferences for minimum disk free space allowed. 19/05/2007 10:29:44|rosetta@home|Deferring communication for 1 hr 0 min 0 sec 19/05/2007 10:29:44|rosetta@home|Reason: requested by project I have looked at other forums to try sort this but had no luck so far. any ideas? Click below to see my werewolf |
FluffyChicken Send message Joined: 1 Nov 05 Posts: 1260 Credit: 369,635 RAC: 0 |
It connects ok and seti and rosetta run fine on the celeron box and its 6GB HDD. You just need to creat more room on the harddrive OR if it's just for crunching then go to the Rosetta or Seti BOINc General Preferences and set that computer up to use more hardrive space, I tihnk by default it's set to only use 75% of free space. Increase that. I see your using 5.8.16 boinc, so just set it in local preferences within the client as it's easier and less confusing normally. Team mauisun.org |
FoldingSolutions Send message Joined: 2 Apr 06 Posts: 129 Credit: 3,506,690 RAC: 0 |
What you have to do is to go to your general preferences when you log in on the rosetta website, then go to edit preferences at the bottom. Then on the "disk and memory usage" section, change the option which says "leave at least(disk space)" to a very small number. I set mine to 0.005GB of space, but the smaller the better, this way you don't have to worry about the problem again. I run BOINC from a memory stick so I had this problem before too. |
Ebola Bob Send message Joined: 12 May 07 Posts: 22 Credit: 19,770 RAC: 0 |
What you have to do is to go to your general preferences when you log in on the rosetta website, then go to edit preferences at the bottom. Then on the "disk and memory usage" section, change the option which says "leave at least(disk space)" to a very small number. I set mine to 0.005GB of space, but the smaller the better, this way you don't have to worry about the problem again. I run BOINC from a memory stick so I had this problem before too. I shall try this now, I have had no luck so far and am about to give up. Click below to see my werewolf |
dcdc Send message Joined: 3 Nov 05 Posts: 1832 Credit: 119,860,059 RAC: 1,141 |
Bob - as FC posted, it's a BOINC thing to do with disk space: (there was work but you don't have enough disk space allocated) You just need to free up space on the boinc drive or change your preferences and then update BOINC, and then rosetta should run fine. HTH Danny |
dag Send message Joined: 16 Dec 05 Posts: 106 Credit: 1,000,020 RAC: 0 |
According to the stats site, Seti provides 1.084 times the credit Rosetta provides, comparing 22000 computers that are crunching both. That's not a large difference, but if your computer is light on FPU and cache, you will do worse on Rosetta. dag --Finding aliens is cool, but understanding the structure of proteins is useful. |
Mike Gelvin Send message Joined: 7 Oct 05 Posts: 65 Credit: 10,612,039 RAC: 0 |
According to the stats site, Seti provides 1.084 times the credit Rosetta provides, comparing 22000 computers that are crunching both. I do not think this is accurate. I believe the web page you got the figure from is based on information contained in the Hosts.gz file provided by Rosetta. In that file, the specific information (<credit_per_cpu_sec>) has not been update by Rosetta for a LONG time, hence the stat is probably wrong. I’m actively seeking an answer to this in another tread. |
Paydirt Send message Joined: 10 Aug 06 Posts: 127 Credit: 960,607 RAC: 0 |
Mmmm Sausage... Anyhow, I get less credit with Rosetta, and the project inspires me more than SETI, so I crunch Rosetta. I get totally different credit with Folding@Home. I use two graphics cards to crunch for them. One thing that always annoyed me about SETI@Home is that they would use up to 5 computers on one WU. This is an indication they don't really need the horsepower. Whereas, I THINK, Rosetta@Home uses 2 if they return a valid result and the same result. |
Mod.Sense Volunteer moderator Send message Joined: 22 Aug 06 Posts: 4018 Credit: 0 RAC: 0 |
I THINK, Rosetta@Home uses 2 if they return a valid result and the same result. No, Rosetta just uses one. It is very rare for a task, with the same random number seed that drives the specific models to crunch, to be sent to more then one machine. Basically if the first machine fails to return result on time or if first machine gets an error processing the task. But otherwise, you are the ONLY one crunching those models. This is part of what makes issuance of credit on Rosetta a bit of a challenge. It is also why there will always be variability in credit vs. time expended, because different models will require slightly different calculations and time to explore. Rosetta Moderator: Mod.Sense |
Message boards :
Number crunching :
Rosetta v Seti times v credit
©2025 University of Washington
https://www.bakerlab.org