Message boards :
Number crunching :
Declining RAC
Message board moderation
Author | Message |
---|---|
Dennis Send message Joined: 26 Jun 07 Posts: 153 Credit: 15,826,319 RAC: 0 |
On all of my SETI dedicated rigs, the RAC is declining slowly since the last scheduled down time or so. Each so far has declined 150 to 500ish, except one that is a few weeks old, dennis3c, still going up. My pending has raised a few K but I do not know if that is enough to off set the decline. Is this my problem or others doing the same? Thank you for any answers in advance. Hope you are all having a grand 4th of July weekend. |
John Clark Send message Joined: 29 Sep 99 Posts: 16515 Credit: 4,418,829 RAC: 0 |
Exactly the same issue I am mentioning in this thread, I think. It's good to be back amongst friends and colleagues |
Dennis Send message Joined: 26 Jun 07 Posts: 153 Credit: 15,826,319 RAC: 0 |
Exactly the same issue I am mentioning in this thread, I think. I read your post just after you originally posted it, and again this morning and I maybe convinced myself it did not apply to my question since I have not played around with the different AK-8 applications. SSSE3 is running on the 66/6700's and 4.1 on the 9450's and SSE3 on the AMD's since they came out. |
Fred J. Verster Send message Joined: 21 Apr 04 Posts: 3252 Credit: 31,903,643 RAC: 0 |
Exactly the same issue I am mentioning in this thread, I think. I'm not still sure, if it's has something to do, with a 'bunch off very short WU's, though. And some errorous WU's, wich won't easily UPLoad. And I have many pending WU's. Time will tell. I've been watching my RAC, for the last week and it rises and falls, but, for now, it is declining a little. |
ka4iqd Send message Joined: 19 Jul 99 Posts: 16 Credit: 6,009,002 RAC: 0 |
On all of my SETI dedicated rigs, the RAC is declining slowly since the last scheduled down time or so. Each so far has declined 150 to 500ish, except one that is a few weeks old, dennis3c, still going up. My pending has raised a few K but I do not know if that is enough to off set the decline. Is this my problem or others doing the same? Thank you for any answers in advance. Hope you are all having a grand 4th of July weekend. I've been watching a steady decline in my rac since June 17th and wondering what is going on too. I have three different computers that all started downhill in the same 24 hour period. My combined rac was to almost 21,000 that day and since have been dropping steadily and was at about 18,500 just a little bit ago and the graph is still nose diving. No changes have been made in the apps, the computer usage, hardware, or the overclocking on any of the three computers. I've been hoping it was just normal ups and downs but there have been no ups in that time. Maybe tomorrow. :-) -- Steve ka4iqd |
John Clark Send message Joined: 29 Sep 99 Posts: 16515 Credit: 4,418,829 RAC: 0 |
Looks like it's just ups and downs (I hope). I have lost 550 average on 1 rig and 100 on another sofar. It's good to be back amongst friends and colleagues |
Terror Australis Send message Joined: 14 Feb 04 Posts: 1817 Credit: 262,693,308 RAC: 44 |
I've noticed that my "big" crunchers have been struggling to hold RAC but the slower ones (< 500 RAC) have actually been doing slightly better than normal. What I've also noticed is that most of units being handed out are only scoring around the 50 credit mark. Out of the last 200 or so units crunched on one computer only 6 scored above 60 and only 2 of those above 70. On a second box it was 8 out 200 above 60 with again only 2 above 70 There was also something posted on another thread that the "correction factor" applied on units in this score range was on the low side of the "S" curve resulting in a lower credit per CPU second value than at the top and bottom of the range. Maybe this has something to do with it. Brodo |
John Clark Send message Joined: 29 Sep 99 Posts: 16515 Credit: 4,418,829 RAC: 0 |
The Penny still seems to be falling. Now below 8,200, when not a week ago it was at 8,750. Baaah! It's good to be back amongst friends and colleagues |
Fred W Send message Joined: 13 Jun 99 Posts: 2524 Credit: 11,954,210 RAC: 0 |
The Penny still seems to be falling. Now below 8,200, when not a week ago it was at 8,750. We're all dropping together from looking at relative positions in the top_hosts listing. So it must be the mix of work coming out ATM. F. |
Geek@Play Send message Joined: 31 Jul 01 Posts: 2467 Credit: 86,146,931 RAC: 0 |
I believe that anytime there is a mix of short time duration work units being sent out, the RAC suffers. I have observed this for some time and with all my boxes. Boinc....Boinc....Boinc....Boinc.... |
Gavin Shaw Send message Joined: 8 Aug 00 Posts: 1116 Credit: 1,304,337 RAC: 0 |
What timing. I upgraded from Boinc 5.4.11 to 5.10.45 on the 12th of June. Since then I have seen a constant drop in RAC for that machine. At the moment it is down about 200 points (approximately 10%). It just seems funny with the timing of upgrading and the lower paying work. I was beginning to wonder if the newer version of Boinc had something to do with it... Never surrender and never give up. In the darkest hour there is always hope. |
Fred W Send message Joined: 13 Jun 99 Posts: 2524 Credit: 11,954,210 RAC: 0 |
What timing. I don't believe so. I think it's just your lousy timing (again? ;) F. |
John Clark Send message Joined: 29 Sep 99 Posts: 16515 Credit: 4,418,829 RAC: 0 |
I would agree with Fred. I have made no change to my BOINC Manager and the RAC still goes down and down, even when it looks like settling, it starts to fall again. In relative positions, I seem to be falling marginally faster than those surrounding me. I am now out of the top 20. My drop has been made worse since I swapped from Alex's SSE4.1 code to the SSSE3x. I am on a longer term test to see which is the faster, using time and RAC as my guide. I am beginning to wonder whether to return to the 4.1, but need to give it another week or more (really). It's good to be back amongst friends and colleagues |
Fred W Send message Joined: 13 Jun 99 Posts: 2524 Credit: 11,954,210 RAC: 0 |
I would agree with Fred. I would stick with the SSSE3x for now. I am following you down (in fact I have disappeared entirely from the top_hosts list ATM but this seems to happen occasionally - should be back at the next update) and I have not changed my App recently. I am sure that the change of App is not a contributary factor in your fall in the rankings and you need to wait for things to stabilise before coming to any conclusion. F. |
Dennis Send message Joined: 26 Jun 07 Posts: 153 Credit: 15,826,319 RAC: 0 |
I look at my teams RAC and since Tuesday I see a constant daily small decline, and the same on the Navy, our direct but respected rival. Both teams primarily do SETI crunching. If the earned cobblestones have been reduced or the AR's or short or long WU's atm cause it, that is all fine. I see nothing announced here or elsewhere RAC's may be adjusted or go down, so I tend to wonder if my rigs have a problem. I may not notice this drop as much, but just starting up another Q9450 rig OC'd to 3400ish a few weeks ago and still seeing my RAC decline has made it more noticeable to me. I seem to have answered my own questions if the RAC is dropping and if it is my problem or not. Yes it is slowly dropping and no it is not a problem with my rigs. I should have waited a while before posting the questions. |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
Well, the multipliers reported in the stderr.txt task outputs still say 2.85, so that's not adjusted (yet) AFAICT. The next likely suspect is work distribution. If the apparent shift keeps up I might need to collect some data on current WU AR distribution relative to previous months, to aid in determining if tweaking any of the profile guided optimisation weighted datasets used to prepare the optimised applications might help. I would want to see if current trends stay with us for a few weeks first though, as it may be transient due to seasonal telescope use. Jason "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
Fred W Send message Joined: 13 Jun 99 Posts: 2524 Credit: 11,954,210 RAC: 0 |
Well, the multipliers reported in the stderr.txt task outputs still say 2.85, so that's not adjusted (yet) AFAICT. The next likely suspect is work distribution. If the apparent shift keeps up I might need to collect some data on current WU AR distribution relative to previous months, to aid in determining if tweaking any of the profile guided optimisation weighted datasets used to prepare the optimised applications might help. I would want to see if current trends stay with us for a few weeks first though, as it may be transient due to seasonal telescope use. @Jason, I'll see if I can knock out a quick AR distribution from my log later on today. F. |
Fred J. Verster Send message Joined: 21 Apr 04 Posts: 3252 Credit: 31,903,643 RAC: 0 |
Well, the multipliers reported in the stderr.txt task outputs still say 2.85, so that's not adjusted (yet) AFAICT. The next likely suspect is work distribution. If the apparent shift keeps up I might need to collect some data on current WU AR distribution relative to previous months, to aid in determining if tweaking any of the profile guided optimisation weighted datasets used to prepare the optimised applications might help. I would want to see if current trends stay with us for a few weeks first though, as it may be transient due to seasonal telescope use. Saw my RAC, falling also and had a look at my >35K pendings, look at this one : This WU (Here is the whole list off pendings and found a few strange things). PENDING Credits One off my wingmans, has a problem, maybe me too??? For now, just have to wait! |
Fred W Send message Joined: 13 Jun 99 Posts: 2524 Credit: 11,954,210 RAC: 0 |
@Fred The links you posted just link to the Pendings of whoever clicks on them (i.e. when I click on them I see my Pendings). F. |
jason_gee Send message Joined: 24 Nov 06 Posts: 7489 Credit: 91,093,184 RAC: 0 |
Thanks Fred. When you have this TweakMaster Joe may be the best person to determine whether adjusted PGO builds *may be* worthwhile (or not) depending on any apparent distribution change and its likelihood of persistence. (IIRC Joe was predicting a change about now, but may have said at the time it could be temporary, though that piece of information was lost amid the clouds of development for the initial AKv8 releases). In the meantime we have some cosmetic fixes to look at so that'll keep us occupied either way. Jason "Living by the wisdom of computer science doesn't sound so bad after all. And unlike most advice, it's backed up by proofs." -- Algorithms to live by: The computer science of human decisions. |
©2024 University of California
SETI@home and Astropulse are funded by grants from the National Science Foundation, NASA, and donations from SETI@home volunteers. AstroPulse is funded in part by the NSF through grant AST-0307956.