| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: Quickest WUs? |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 11
|
|
| Author |
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Now that Genome has finished, which (on average for an 'average' machine), is the fastest project WU/ Does anyone have a sample time for all of the remaining?
|
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Sekerob has a nice graph here. Looks like HPF2 for now, but Dengue fever & FAAH round 2 are starting soon also, and the upcoming cancer project too.
http://www.worldcommunitygrid.org/forums/wcg/viewthread?thread=13733 |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Just the job, many thanks.
|
||
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3715 Status: Offline Project Badges: 
|
Now that Genome has finished, which (on average for an 'average' machine), is the fastest project WU/ Does anyone have a sample time for all of the remaining? Since I have finished my Genome WUs I am trying to determine it by monitoring my machine. And I am puzzled: so far I have found that FAAH units go faster in my BOINC agent, and HPF2 in my UD agent?!?Also HPF2 units seem to jump several % at once between each cycle in UD while the percents are slowly increasing in a linear fashion for the same project in BOINC. Is this a known phenomenon or is it just pure coincidence? Cheers. Jean. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
The graphs show the average figures. Now, because batches of work units aren't shared between the two agents (batches are assigned to one agent or the other), and the average size of batches may vary (one may be for the exosite experiment, and another for the sidechain experiment, for example) the behaviour you are seeing is perfectly normal.
And the jumping percentages? I think that's a known detail, too. Explaining what causes it is quite complicated (I will try to explain it if you ask, but I'm not 100% sure of all the details). Sometime, though, you may see the %complete actually jump backwards. The short version: this is what happens when you are doing non-deterministic computing. The end time is just a guess. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Hello Scribe
Based on the previous post it seems that which work unit is the fastest could depend on which agent you are using ie BOINC or UD. As I only have the one computer I can only comment on BOINC, in my case just like JmBoullier, I find FAAH work units run faster that HPF2 units. I would be interest in seeing what Sekerob's graphs would show if he graph on only UD agent results versus only BOINC results, could be UD versus BOINC result differences have never come up before. |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
No such split data exists on the public side. Sometimes I ask a tech to give me one or the other extract, but here there is little point as batch durations vary all the time. The only thing that is actually visible since a few days is the GC per WU time rise. BOINC has deadlines of 9 days, effectively the client always working to complete in 8, but that's a different story, BOINC is done with GC. What now becomes evident is the lower end machines that were employed on UD, which has 2 months deadline to return a WU. The average time hiked to 3 hours. Like any previous project WU's will be trickling back pretty much up to 2 months ofter close.
----------------------------------------For upcoming Dengue there is no comparison as it will only run on BOINC for phase I, similar as HCMD Phase I only running on UD. Maybe I'll ask knreed to set up a script for daily export. Lots easier than scraping web pages.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 1 times, last edit by Sekerob at Jul 8, 2007 11:09:55 AM] |
||
|
|
retsof
Former Community Advisor USA Joined: Jul 31, 2005 Post Count: 6824 Status: Offline Project Badges:
|
FAAH is about 1/3 faster than HPF2 on these BOINC systems. The big advantage to FAAH is the tiny transfer files. They don't tie up a dial-up phone connection much.
----------------------------------------
SUPPORT ADVISOR
----------------------------------------Work+GPU i7 8700 12threads School i7 4770 8threads Default+GPU Ryzen 7 3700X 16threads Ryzen 7 3800X 16 threads Ryzen 9 3900X 24threads Home i7 3540M 4threads50% [Edit 1 times, last edit by retsof at Jul 8, 2007 2:04:29 PM] |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
As I asked in the Boinc forum I thought it would be obvious which agent I was referring to!.
I also find that FAAH is faster than HPF....which is NOT what the graph shows....anyone know why the graph is so far out? |
||
|
|
JmBoullier
Former Community Advisor Normandy - France Joined: Jan 26, 2007 Post Count: 3715 Status: Offline Project Badges:
|
I also find that FAAH is faster than HPF....which is NOT what the graph shows....anyone know why the graph is so far out? Hello Scribe!I am glad that you and dseto confirmed what I have observed for Boinc. And I think the fact that it is just the opposite with UD agents might probably explain why the graph does not reflect what can be seen under Boinc. There are still so many UD agents running in WCG (BoincStats shows only 51000 users for WCG) that their effect on the average (raising it up) might be what we see on the graphs. Not forgetting that these UD machines are probably the slowest ones since the owners of the powerful multicore ones are probably all running Boinc already. This is just reasoning and, short of having separate figures per agent, it will not be easy to confirm. Cheers. Jean. |
||
|
|
|