Slow wingmen & growing pending accounts
27 Sep 2008 20:45:40 UTC
Topic 193955
(moderation:
I seem to have just lost the crunching use of my QX6,700 quad, using XP Pro (32 bit) at 3.1GHz.
We had a power dip which affected this PC, but not the others in my small farm (Penny, a Xeon Prestonia or P3).
During trials to reboot, I proved that 2 power cords for my quads had gone at the same time (sems to be fuses). Neither would power the Prestonia. But, fortunately, I found 2 more (one never used) and the newer quad is back up crunching.
No matter what I do with live power cords to the older quad it will not come on. Zilch, Nada, Kaput, dead.
I can only assume the PC has blown a fuse (say on the power block - it is a 450 Watt one) or the MoBo is gone or it's something else.
Can anyone guide me where the fuses are to be found beyond the power cord fused plug (which is OK), and what are they rated at?
For example, is there a fuse in the PC power pack? Are there other small fuses I need to be aware of.
The MoBo is an Asus P5B, and when it is running RealTemp 2.60 says the cores are at 58C, 58C, 57C, 57C and steady.
I am hoping it is just a fuse which needs replacing, and not more serious.
I hope to access skills and tools to check for blown fuses or worse, Monday.
Shih-Tzu are clever, cuddly, playful and rule!! Jack Russell are feisty!
Slow wingmen & growing pending accounts
)
Hmmm...
Well my suggestion would to be to start in with a DMM at the DC power out connectors and work back from there. First check them no load (all DC connectors popped off, then plug the MB connectors back and check at least the 'biggies' (+5, +12, and -12 volt) to make sure they are regulating under load. You might even be able to check some of the other ones if you can find documentation about test points on the MB. If your probes have fine enough and/or long enough tips, you can usually do this via the 'backdoor' on the main MB connectors when they are plugged in. Physical clearance problems are the typical show stoppers when trying this.
You are correct in suspecting there are overload fuses inside the Power Supply. In the old days you used to have access to the AC line protection one from the back panel (some still do). However, it is more common nowadays to have to crack the case of the supply to get to them.
Alinator
Hi, When a mb of mine
)
Hi,
When a mb of mine will not boot I try removing the power cord and mb battery for a half hour or so Then plug in the power cord and try to boot without the battery. If it boots I then reseat the battery and get back to crunching. This sometimes works.
Joe B
RE: You are correct in
)
And then be very careful when poking in the insides of it due to possible residual power buildups.
RE: RE: You are correct
)
...unless your name is "Horace Pinker"...
I will try out JoeB's
)
I will try out JoeB's suggestion, when I play with the Quad tomorrow afternoon. If it starts I may be able to get in some crunching.
However, the quad, when I apply a live power cord shows absolutely no lights anywhere. I suspect the PSU has bombed, so will need to confirm this Monday and swap out another one.
No crunching this week end then.
There is little point in fiddling within a PSU, and trying to avoid a very nasty belt, when they do not cost large amounts to swap out.
Shih-Tzu are clever, cuddly, playful and rule!! Jack Russell are feisty!
It was the PSU, but,
)
It was the PSU, but, fortunately, nothing else borked.
Repaired with a heavier duty (as wattage) and better build quality PSU and back crunching,as at this time (for Malaria)
Shih-Tzu are clever, cuddly, playful and rule!! Jack Russell are feisty!
Yeah your old PSU was a bit
)
Yeah your old PSU was a bit too small for your machine.
That happened to me with a dual core so I replaced it with a Ultra X2 750 watt ATX
Didn't cost that much and it is a nice one that come with more cables than you need (of course you put those in your pc stash)
Fast and easy to replace with these nice cables too.
And-now-for-something-complet
)
And-now-for-something-completely-diferent.
Looking at my pending, I see my WU numbers are still rising and the wingmen are very very slow to report.
This was mentioned in another thread earlier, but this seemed to discuss a problem which new and the large pending credit claims were likely to be cleared quickly.
Now, it seems, the problem is continuing and is likely to be the norm.
Any thoughts as to how long the slow wingmen issue is likely to last?
Shih-Tzu are clever, cuddly, playful and rule!! Jack Russell are feisty!
RE: And-now-for-something-c
)
First and foremost, it is not "a problem". The oldest pending task I can find for your systems was generated on September 19th. The deadline is 18 days. 18 days from September 19th was yesterday. The second replication was issued 4 days later, which is not all that bad. It could be better, but even if it was reduced to only a 1 day lag between first and second issuance, that means that if that were the case, in effect you would be (bolding to stress) complaining that someone ran over their deadline by a few hours and are demanding that something be done about it.
Think about that for a few seconds... Now consider that the remaining pending credit situations that you have, even your original deadline has not expired!
OK. Back during S5R2, the deadline was cut from 21 days to 14 days. This caused slower / and less allocated hosts to have BOINC (not Einstein, but BOINC) enforce "Earliest Deadline First" scheduling. People griped and complained that Einstein@Home was "hogging their CPU".
Due to seeing all this, I asked the project to consider an increase in the deadline from 14 days to 18-21 days again. They took some time to consider it, then they did bump the deadline up to 18 days.
As it stands right now, with the Windows application disadvantaged, I don't think it would be a good idea to reduce the deadline back to 14 days. I think you'll just change out the people complaining and there will still be people upset. I would say that a determination on reducing deadlines back down to 14 days should wait for the data to come back on how much improvement the power users application for Windows provides. It should be a pretty decent increase in performance and should impact a very large segment of the user base if it can be promoted to the stock application.
Like I said, getting the Windows app up to par with the Linux app will help many things. I'm used to being scoffed at by now, so I stick to what I said before...
I am sorry the comment came
)
I am sorry the comment came across as a complaint, which it was not intended to be. On this point I appear to be acting a little stupid, and I now understand the rising pending account issue (which it is not).
What I have, from your clear comment, is that given time (18 days) things will clear up. So, I am not advocating a return to the shorter reporting period.
I do have a P3 Coppermine host which takes 42 hours to crunch a S5R4 WU, but I keep my cache down to between 0.5 and 1.0 days which solves the problem of reporting late.
I want to crunch on another project ATM. But, on return will be trying the new SSE2 client. Unfortunately my P3 will not feature here, and it will continue using the SSE detection and client use.
Shih-Tzu are clever, cuddly, playful and rule!! Jack Russell are feisty!