| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Completed Research
Forum: FightAIDS@Home
Thread: this is a really long work unit |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 149
|
|
| Author |
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
1. I think you presume wrong. The batch generation has been performed 100s if not 1000s of times without that particular parm discussed kicking in. Why sample?
----------------------------------------2. knreed explained the technical hurdle and the code that needs and will be implemented by Berkeley to send these monsters to appropriate hardware. That "appropriate hardware" would than probably also eliminate to certain extend any disparity due slow meeting fast (but fast with 64bit OS will still claim hi on the integer part of the benchmark) as brought to table once more in the "just an observation" post. 3. Point 2 for FAAH is moot anyhow as we are in the kick over phase to go to quorum-less validation and credit claim/grant calculation. The integrated "reference" mini work-unit benchmark in each job will be part of how much a client claims. So far seeing this methodology in practice on DDDT, this project jumped 9 points per hour average on the whole fleet (from about 75 - pre conversion - to 84 on half Sunday, giving 4 days production in new setup with 36% of original work unit daily). This is even exceeding RICE which lies about 7 per hour above the WCG average. More days of data needs collecting if this credit level holds.... win some loose some, but so far under the bottom line my WCG average has been on the increase for the Quad and C2D. 4. Well, I hardly know anything, but hope to have understood a bit more after reading all the posts contributed by the technical staff recently, from my comfy chair  PS Per today, the monsters are positively NOT prevalent. Still only getting the faah42xx versions. Average run time even dipped a bit for the project.
WCG
Please help to make the Forums an enjoyable experience for All! |
||
|
|
mclaver
Veteran Cruncher Joined: Dec 19, 2005 Post Count: 566 Status: Offline Project Badges: 
|
Ok, so now I am confused!
----------------------------------------The good news is one of my long running units completed at 62 hours with a claim credit of 821.5. The Bad news is that my Grant credit was only 422.8 becasue they Quarm result only took 20 hours and its claimed credit is 422.8. How can that be. This is an AMD 9500 Quad and the only thing running on it was four FightAids units all running flat out. Are these units somehow interacting with one other and that is why they are taking so long. Although the first one ended at 62 hours the other three still have 10-20 hours more to go, one is projected to take more than 80 hours. I thought the objective is to give fair credit for a fair amount of work. 422.8 credit does not seem fair for 62 hours worth of work. Although claimed credit and grant credit varies, I have never lost this much credit before.    |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Ok, so now I am confused! The good news is one of my long running units completed at 62 hours with a claim credit of 821.5. The Bad news is that my Grant credit was only 422.8 becasue they Quarm result only took 20 hours and its claimed credit is 422.8.<snip> This is exactly what I posted about earlier in this thread and then got a response of "Big Mistake" by Sekerob when I suggested kicking the WU to the curb. Subsequently, we've see post after post conveying information about WU's timing out and others who do not get full credit for the large WUs. What a waste. My WU is now at a CPU time of 58:46 and is 94.1% complete. I'm not sure what I'll do if I get the same results, or lack thereof. But, my electric bill has doubled over the past year and computers sure take a lot of juice when running 24/7. And I'm sure not incurring the cost for nothing. |
||
|
|
mclaver
Veteran Cruncher Joined: Dec 19, 2005 Post Count: 566 Status: Offline Project Badges:
|
Thats it for me!!!!!!
----------------------------------------I have now had five ot these canceled after 60 hours with no credit. I am only reciveing 50% credit for these work units when they do finish. I have been a significant contributor to WCG with 19,889,679 Points and this has never hapened before. I have stopped contributing to FightAids and have aborted all of my waitng units. I would like to know if anyone is going to give us credit for the CPU time we donated, that WCG (BOINC) aborted. I have lost 300 CPU hours of those uncredited contributions. I have my logs to document this for 5 of 6 canceled units. I would also like to know if someone is going to fix why I had workunits that went 60 hours but only got half credit. 8/2/2008 11:04:40 AM|World Community Grid|Aborting task faah5009_1hvk_1bv7_00_0: exceeded CPU time limit 201447.246158 8/2/2008 11:04:45 AM|World Community Grid|Computation for task faah5009_1hvk_1bv7_00_0 finished 8/2/2008 3:54:46 PM|World Community Grid|Aborting task faah5012_1qbt_1hpx_01_0: exceeded CPU time limit 177736.732098 8/2/2008 3:54:51 PM|World Community Grid|Computation for task faah5012_1qbt_1hpx_01_0 finished 8/3/2008 9:28:14 AM|World Community Grid|Aborting task faah5013_1qbt_1izi_01_1: exceeded CPU time limit 198871.268434 8/3/2008 9:28:20 AM|World Community Grid|Computation for task faah5013_1qbt_1izi_01_1 finished 8/3/2008 1:51:30 PM|World Community Grid|Aborting task faah5012_1k6p_1hpx_00_1: exceeded CPU time limit 246826.206698 8/3/2008 1:51:31 PM|World Community Grid|Computation for task faah5012_1k6p_1hpx_00_1 finished 8/3/2008 2:01:46 PM|World Community Grid|Aborting task faah5013_4fiv_1hxw_01_0: exceeded CPU time limit 246826.206698 8/3/2008 2:01:47 PM|World Community Grid|Computation for task faah5013_4fiv_1hxw_01_0 finished - Mitch |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
mclaver, I asked you yesterday to wait until Monday for the techs to respond. I also suggested you abort any work you thought may not finish.
Nothing has changed since yesterday. |
||
|
|
Sekerob
Ace Cruncher Joined: Jul 24, 2005 Post Count: 20043 Status: Offline |
Results that validate are not a waste of time and obviously for those who the job bombs due to time exceed or substantial reduced credit wont be happy, but as is normal, those short sold will be reporting and those who get what they expect don't even look up. Hopefully coming week the techs have some data to share on these faah50xx results and as earlier suggested in this thread a review provides some statistical data.
----------------------------------------now regarding: Are these units somehow interacting with one other and that is why they are taking so long. Although the first one ended at 62 hours the other three still have 10-20 hours more to go, one is projected to take more than 80 hours. With mitch his result of 62 hours on a quad versus 20 hours by the quorum partner, it could be that 4 concurrent monsters may be part of the long processing times cause! I know a project that if run on my quad it causes 2 concurrent to run slower than 2 sequential and 3 concurrent in fact doubling run time not to speak of 4 at the same time. A quad is a quad but if there is resource contention they will clock CPU time, but no real progress. We've seen with HCC early versions that some inordinate run times were incurred which got fixed through science app code improvements. It may thus pointing to a technical issue becoming more pronounced with these very complex jobs at very high resolution.
WCG
----------------------------------------Please help to make the Forums an enjoyable experience for All! [Edit 2 times, last edit by Sekerob at Aug 3, 2008 9:04:05 PM] |
||
|
|
mclaver
Veteran Cruncher Joined: Dec 19, 2005 Post Count: 566 Status: Offline Project Badges:
|
I appreciate that, and will see what the techs say tomorrow. I do not know what work units will finish and which ones won't as I have had some finish at over 60 hours and some abort at 49 hours. WIth your post yesterday I did not understand that I was not getting fair credit for the ones that did finish. 422.8 granted credit on a work unit that took 62.4 hours and claimed 821.5 credit does not seem to me that FightAids is cranting fair credit for the work done.
----------------------------------------I have stopped accepting any new work from FightAids and have aborted all wiating work units. I still have four in process with under 50 hours estimated so I will let them complete. I have the equivlant of 33 CPUS on 13 computers all committed 24x7 to WCG so I this is now a small part of my total. As you know, this whole expirience with FightAids, due to these work units has been disappoint to me becasue my contribution to WCG/BOINC has been reduced by almost 50% the last couple of days. - Mitch |
||
|
|
mclaver
Veteran Cruncher Joined: Dec 19, 2005 Post Count: 566 Status: Offline Project Badges:
|
I am running four Quad machines, eight dual, and one single. I have looked back on WCG and I see no indication of this much difference between claimed an granted on any work I have done on Quads with the exception of the Faah50xx units. In some cases, my granted credit slightly exceeds my claimed credit on my quads, for other work units. You may be right, that this is an additional issue with how these Faah50xx units were set up.
----------------------------------------- Mitch |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
Well, it completed. 62.3 CPU hours. 710 claimed points.
The other cruncher's WU has not yet reported ... and, something tells me it won't for a while. For those of you who have one of these WUs, I recommend suspending all other WUs that your machine is running except for the big one. Let it run its course, then resume the other WUs. |
||
|
|
Former Member
Cruncher Joined: May 22, 2018 Post Count: 0 Status: Offline |
I feel sorry for people who had their machines set up to download large work queues! Mine are set for a 2 day queue but some machines ended up with 15+ calendar days worth of work due to the error in the WU size estimation. I should be able to sneak them in under the deadline, so for now I'll just let things take their course.
As for the credit debacle, things seem to be normal on my end. Yes my x64 machines get less than they claim but they always have in the past too. It's just more shocking to see 800 claimed and 600 granted vs 80 claimed and 60 granted. Percentage wise, it's the same though. |
||
|
|
|