| Index | Recent Threads | Unanswered Threads | Who's Active | Guidelines | Search |
 |
World Community Grid Forums
Category: Support
Forum: BOINC Agent Support
Thread: Weirdness in WUs processed |
| No member browsing this thread |
|
Thread Status: Active Total posts in this thread: 11
|
|
| Author |
|
|
BobbyB
Veteran Cruncher Canada Joined: Apr 25, 2020 Post Count: 638 Status: Offline Project Badges: 
|
A little background:
I have an AMD 4 core machine which I suspect has a problem with the CPU. The CPU I can replace. In determining my problem I let it run memtest86 to see if it is just the memory. With memtest86 using 4 cores it freezes after a good while. I see this same behaviour when running Xubuntu 18.04.5(GUI) and BOINC 7.9.3. Memtest86 using the 1 core will go all night. Freeze = no response to RPC, ping, mouse, keyboard, and hardware reset. The light on the front panel flashes as it does when in sleep mode. Power off and on is the only solution. I thought about heat but am not convinced. While waiting for a CPU, I let it run at 82% of CPUs (=3) Xubuntu no GUI and it has not frozen in days. (fingers crossed) Now I expect all the stats for that device to be about 75% of what is used to be. But not. Whereas it use to hover around 20-30 results returned it now does about 6-9. Points and Run Times reflect the same respective lower values. This does not compute and this is the object of this post. How come? The hardware problems, if any, I can fix. |
||
|
|
BobbyB
Veteran Cruncher Canada Joined: Apr 25, 2020 Post Count: 638 Status: Offline Project Badges:
|
In all the diagnosing I did it seems I tampered with the BIOS settings to do with CPU timings in an effort to see if heat was a problem... and I forgot to reset them.
I see a small rise in WU reported this morning but will wait 1 whole day and see. |
||
|
|
Falconet
Master Cruncher Portugal Joined: Mar 9, 2009 Post Count: 3315 Status: Offline Project Badges: 
|
Hopefully it's solved, then.
---------------------------------------- - AMD Ryzen 5 1600AF 6C/12T 3.2 GHz - 85W - AMD Ryzen 5 2500U 4C/8T 2.0 GHz - 28W - AMD Ryzen 7 7730U 8C/16T 3.0 GHz |
||
|
|
BobbyB
Veteran Cruncher Canada Joined: Apr 25, 2020 Post Count: 638 Status: Offline Project Badges:
|
Yes, kind of. I am still experimenting while I wait for a CPU.
|
||
|
|
BobbyB
Veteran Cruncher Canada Joined: Apr 25, 2020 Post Count: 638 Status: Offline Project Badges:
|
It seems it was heat after all.
After a lot of experimenting for nil, I got my hands on a CPU fan with copper tubes and installed. It has now run smoothly for 24 hours with 4 CPUs crunching and way below the max degrees of 70C. It use to hover at just below that at 66C. Now it is ~46C. The specs say it's sweet spot is 59-62C. Lesson learnt from this: When you have an older machine which just did general home processing and convert it to crunch here at WCG then make sure the cooling is correct. |
||
|
|
AgrFan
Senior Cruncher USA Joined: Apr 17, 2008 Post Count: 396 Status: Offline Project Badges: 
|
Which AMD CPU are you running?
----------------------------------------
[Edit 1 times, last edit by AgrFan at Aug 30, 2020 4:32:41 PM] |
||
|
|
BobbyB
Veteran Cruncher Canada Joined: Apr 25, 2020 Post Count: 638 Status: Offline Project Badges:
|
The one in question is an AMD Phenom(tm) II X4 955. It was an AMD Athlon II X2 245 before I upgraded it but I left the same cooler in the machine. This was my error. The thermal design power on the X4 is 125W. Much less on the Athlon II X2: 65W.
----------------------------------------I have another Intel machine I will upgrade soon and will keep my eye on the heat. [Edit 1 times, last edit by BobbyB at Aug 31, 2020 3:29:49 PM] |
||
|
|
AgrFan
Senior Cruncher USA Joined: Apr 17, 2008 Post Count: 396 Status: Offline Project Badges:
|
Yeah, that would do it. AMD Phenom II x4 955 is a furnace at 125W. Especially if it was overclocked.
----------------------------------------You might want to consider upgrading to Ryzen to reduce power usage and increase production. I retired a AMD Athlon II x4 640 (Phenom II x4 945 w/o L3 cache) box a couple years ago. AM3 socket is pretty much obsolete now.
[Edit 1 times, last edit by AgrFan at Sep 1, 2020 1:55:19 AM] |
||
|
|
BobbyB
Veteran Cruncher Canada Joined: Apr 25, 2020 Post Count: 638 Status: Offline Project Badges:
|
Not overclocked.
----------------------------------------An upgrade would require buying all new stuff except for the case and power unit. Not an option... and it's not the money. The reason I am running obsolete equipment is just that. It was junk just collecting dust for which I found a use when I discovered WCG. I have 3 going and 1 2011 era PC.... 9 years makes it close to obsolete. So I'll let them burn out. I was considering Raspberry Pi 4 but they do not seem too productive in terms of WUs processed. [Edit 1 times, last edit by BobbyB at Sep 1, 2020 4:50:27 PM] |
||
|
|
spRocket
Senior Cruncher Joined: Mar 25, 2020 Post Count: 280 Status: Offline Project Badges:
|
Lesson learnt from this: When you have an older machine which just did general home processing and convert it to crunch here at WCG then make sure the cooling is correct. BOINC is a good stress tester, indeed.  I sent a I have some old boxes crunching away, too. I discovered one had a partially-bad RAM stick, and wound up isolating that out to use what I could. Another is a Phenom II 550 Black Edition, which I lucked out on and found that I could unlock two more cores. It's happily chomping away on WCG since I started on it, running at about 57°C with the stock cooler. I can overclock it or I can have the extra cores; I can't have both, so I'll take the cores. |
||
|
|
|