climateprediction.net home page
Still got problems with my athlon xp-m

Still got problems with my athlon xp-m

Questions and Answers : Windows : Still got problems with my athlon xp-m
Message board moderation

To post messages, you must log in.

AuthorMessage
Cluster

Send message
Joined: 4 Feb 05
Posts: 11
Credit: 4,398,892
RAC: 0
Message 10912 - Posted: 15 Mar 2005, 13:23:18 UTC

I really don’t know what to do except stop running cp on my Athlon. its a Athlon xp-m with dual ddr 400, timings spd, abit nf-7 s v2 running windows XP sp2. 2200mhz 200x11.5 3x3 ram ratio.

I can not get this machine to run cp with out the model crashing - When this happens, the actual pc dose not crash, just cp and crashes so badly that it either resets the model or gives the -5 error (client error) and downloads a new one, either way losing all the work that has been done.

I have no problems with any other programs or with stabilty (except cp).

At first i thought it was a heat/overclocking issue but after turning my processor right down to 1800mhz and installing tread master which limits the max CPU to 30% (or what ever you set it to) load, keeping my temps at 45-47c and still got the same problem.

I then though that one of my memory modules might be faulty so i tested them with "pc check", they both tested fine. I then though it could be that i have different makes of ddr 400s dim in each memory slot so i removed one so that it was running in single channel mode of one dim. Still got the same problem tried it with the second dim on its own and still had the model crash after 20 or so hours. I then turn the ddr 400 memory right down to 333 (FSB 166) to under clock the memory to see if that was the problem, still had the model crash.

After this a detached from the project, deleted all folder related to boinc, download the new client and a new cp model, woke up this morning to yet another model crash.

Out of all my pc's running this project, this is the only one that im having problems with, all others are Intel and g4's.

I would say there is a problem with AMD processors on this project but I know that a lot of people don’t have any problems, Team AMD seems to be doing ok, but i cant get it stable on this athlon xp-m.

I can run any other project or software and not have any problems, this pc is usually left on all the time and i don’t really ever have to reboot it (even after a boinc cp crash) as windows stays stable.


I am at a loss, im not sure what to try next, i have just stopped my self spending out on new memory and heat sinks etc..... as I don’t really have any money to burn for this project and as the pc is fine for everything except for this climate software.

Any suggestions are welcome, it would be good to get to the bottom of this for my sanity, it’s been bugging me for weeks now.

For the time being, ill put this pc back on Seti.



ID: 10912 · Report as offensive     Reply Quote
Profile Andrew Hingston
Volunteer moderator

Send message
Joined: 17 Aug 04
Posts: 753
Credit: 9,804,700
RAC: 0
Message 10928 - Posted: 15 Mar 2005, 14:42:50 UTC

If it isn't hardware, and not something like disk access, then the obvious culprit will be a software clash. For example, I discovered for myself during a recent upgrade that there is a series of problems associated with the combinations of CD/DVD software that some of us tend to accumulate on our machine, installed almost by default. They ought to coexist happily, but sometimes do not and several people have reported faults that seem to relate to that. The problem in my case was fairly obvious once I'd tracked it down as the system would crash even without BOINC, but others seem to have only met the problem when running CPDN.

I use that as an example only. These things can be a real pain to track down, but it is worth persevering as it is more than likely that whatever is going wrong will plague you eventually when running something else. The standard advice is to work through the system startup fault finding routine - look in help and support in WinXP start (search on Startup options).

Best of luck. If you find yourself in need of help and solace, then somebody far more knowledgeable than me will probably be willing to hold your hand.
ID: 10928 · Report as offensive     Reply Quote
Cluster

Send message
Joined: 4 Feb 05
Posts: 11
Credit: 4,398,892
RAC: 0
Message 10938 - Posted: 15 Mar 2005, 15:06:18 UTC - in response to Message 10928.  


It could well be a software conflict, tracking it down will be a nightmare though. I have so many different thing running on this pc as its my main home computer. Ill try running boinc on a new profile with none of my apps or untils running. Might put a spare hard drive in and do a fresh install to see if it can complete a cp run with nothing else loaded with out unnecessarily trashing my hard drive.



ID: 10938 · Report as offensive     Reply Quote
Cluster

Send message
Joined: 4 Feb 05
Posts: 11
Credit: 4,398,892
RAC: 0
Message 11172 - Posted: 20 Mar 2005, 13:31:36 UTC

I have re-installed XP, left all other software off the system except for my virus scanner and had just CP running on the athlon and the model still crashed but windows did not.

I will try running the linux version on top of windows with colinux (need to run it on top as i have no linux drivers from my wireless network card) to see if that will run without crashing.

If this dont work, then i give up.


ID: 11172 · Report as offensive     Reply Quote
Helmer Bryd

Send message
Joined: 16 Aug 04
Posts: 147
Credit: 7,930,288
RAC: 9,915
Message 11196 - Posted: 21 Mar 2005, 0:04:01 UTC
Last modified: 21 Mar 2005, 0:29:39 UTC

I have a rather similar system: XP-M 11x200=2200MHz dual channel 200MHz memory and Abit NF 7.
You say 11,5x200 but that is 2300MHz.

If you get temps like 45-47c with 30% load I think your system is overheating at full load. Mine usually craps out at 52c.

If you run Prime95 stess test and watch the temp carefully?
http://www.mersenne.org/freesoft.htm

Edit: forgot to say: keep the voltage on the DIMM's at atleast 2.8v and vcore at 1.65v or more. (Mine needs 1.675v)


ID: 11196 · Report as offensive     Reply Quote
old_user23880
Volunteer tester

Send message
Joined: 10 Oct 04
Posts: 223
Credit: 4,664
RAC: 0
Message 11202 - Posted: 21 Mar 2005, 1:01:42 UTC

The -5 error code is a calculation error. Check that when you close the model, you suspend it first and give it plenty of time to shut down cleanly. If the stability tests don't point to a problem and you still want to run cpdn, you could try my solution (my Athlon and boinc are not good friends - I had lots of crashes and wasted far too much time). This is only for people who don't care about boinc stats and don't want to run other boinc programs.

You can run classic cpdn instead. It almost always runs stably, you get visualisations to observe your models as they progress, you see graphs of completed models, it runs at about the same speed as boinc/cpdn and you can still post on all the cpdn boards if you want.

Uninstall boinc/cpdn.
Go to the cpdn home page.
Click on Open University course (on the left of the page).
Click on Download model.



__________________________________________________

ID: 11202 · Report as offensive     Reply Quote
Cluster

Send message
Joined: 4 Feb 05
Posts: 11
Credit: 4,398,892
RAC: 0
Message 11237 - Posted: 21 Mar 2005, 13:40:04 UTC

Thanks mo.v for the suggestion/solution, if i can not get to the bottom of this i will give clasic ago - good to know that im not the only one having problems.

Thanks Aquila, i will test my system with prime95, i used pc check last time which cut windows out of the picture, it will be intresting to see if my memory is ok within windows.

I will allso give the voltages you reconmend ago, they are higher than mine as i currently run them at default. sorry for the typo, i do run this pc at 2300mhz.

I have just had a look at your computers and all your AMD processors are running Linux and your intel's are running windows - is this because you find linux more reliable with this project on amd processors? - im intrested as this is the route i might try next.

Thanks all.
ID: 11237 · Report as offensive     Reply Quote
Helmer Bryd

Send message
Joined: 16 Aug 04
Posts: 147
Credit: 7,930,288
RAC: 9,915
Message 11247 - Posted: 21 Mar 2005, 17:45:20 UTC

No, I don't know if Linux is more stable on AMDs, I use it since I don't have a Windows license and crunch times are about the same on AMDs. Mine P4/Win is a leased box.

If you can manage 2 hours in Prime95 stress test I think you will be fine, mine craps out after 1:30 so I'm living on the edge, hehe..

I guess the default vcore on XP-Ms are something like 1.45v and that is very low.
The downside of upping the voltages is more heat but with nice cool winter air coming through the went shaft here I'm doing fine (so far..)

A good memory test program is Memtest86: http://www.memtest86.com/
ID: 11247 · Report as offensive     Reply Quote
Cluster

Send message
Joined: 4 Feb 05
Posts: 11
Credit: 4,398,892
RAC: 0
Message 11467 - Posted: 29 Mar 2005, 13:10:46 UTC

I ran Memtest86 for 24 hours and got no errors or problems. I have upped the voltage on the memory to 2.8 and upped the processor voltage to 1.6 but CP still crashed.

Its realy fustrating that i can not get this pc stable with cp, there are a quite a few other people having the same problem with this project with the same processor, im finding it hard to beleive that these arnt related in any way.

Bit fed up with fault finding now, its making me want to buy a new computer so i have just stopped running cp through Boinc on this pc. for any other task, application or Boinc Project, it runs fine and dose not crash.

Just added a mini mac to replace this computer although its got no where near the cruching power, 6.8 sec's compared to 2.5 sec's but it has not crashed.

Might give classic ago.
ID: 11467 · Report as offensive     Reply Quote
Cluster

Send message
Joined: 4 Feb 05
Posts: 11
Credit: 4,398,892
RAC: 0
Message 16186 - Posted: 23 Sep 2005, 11:32:41 UTC

Well i almost gave up with CPN on my athlon leaving it to crunch seti and predictor data (pitty as its my fastest machine) for last couple of months. After seeing that cpn has moved on a couple of versions i thought that i give it another go. My athlon downloaded the 4.19 sulpur cycle and has been happily crunching away for the last week and a half (on 4.12 the model would crash after about 4 hours or so).

The new Sulpur Model seems to be much more Athlon xp friendly as i have not change anything on my system.

Well Done CPN for making the latest build a lot more stable on Athlons.
ID: 16186 · Report as offensive     Reply Quote
Cluster

Send message
Joined: 4 Feb 05
Posts: 11
Credit: 4,398,892
RAC: 0
Message 16829 - Posted: 28 Oct 2005, 10:00:28 UTC - in response to Message 16186.  

Well i almost gave up with CPN on my athlon leaving it to crunch seti and predictor data (pitty as its my fastest machine) for last couple of months. After seeing that cpn has moved on a couple of versions i thought that i give it another go. My athlon downloaded the 4.19 sulpur cycle and has been happily crunching away for the last week and a half (on 4.12 the model would crash after about 4 hours or so).

The new Sulpur Model seems to be much more Athlon xp friendly as i have not change anything on my system.

Well Done CPN for making the latest build a lot more stable on Athlons.


Blast - through my own tinkering trying to get my machine stable at 2.5ghz i have crashed my sulpur model 4.19 - I have put my machine back to the stable settings (2.4ghz) and Boinc downloaded a new model, this time hadsum 4.13 but as before the model crashes after a few hours so im again taking this computer off CPN to dedicate it to another project. Will try again if more sulpur models or a new hadsum version released. 4.12 and 4.13 wont run on my athlon although the sulpur model ran fine.

ID: 16829 · Report as offensive     Reply Quote
racinjimy

Send message
Joined: 19 Apr 05
Posts: 53
Credit: 6,325,436
RAC: 0
Message 16830 - Posted: 28 Oct 2005, 11:10:42 UTC - in response to Message 16829.  

are you at least 24hours Prime95 stable?

if not then you have a hardware issue

nowhere in your thread have I seen that you used Prime95

someone suggested 2hours of Prime stability, that is just not enough as you run CPDN 24hours/day
ID: 16830 · Report as offensive     Reply Quote
Helmer Bryd

Send message
Joined: 16 Aug 04
Posts: 147
Credit: 7,930,288
RAC: 9,915
Message 16847 - Posted: 29 Oct 2005, 2:30:14 UTC

I think vcore=1.60 is too low, the default for Bartons at 2.2 GHz is 1.65.

ID: 16847 · Report as offensive     Reply Quote
Cluster

Send message
Joined: 4 Feb 05
Posts: 11
Credit: 4,398,892
RAC: 0
Message 17497 - Posted: 28 Nov 2005, 15:57:58 UTC - in response to Message 16829.  

[/quote]
Blast - through my own tinkering trying to get my machine stable at 2.5ghz i have crashed my sulpur model 4.19 - I have put my machine back to the stable settings (2.4ghz) and Boinc downloaded a new model, this time hadsum 4.13 but as before the model crashes after a few hours so im again taking this computer off CPN to dedicate it to another project. Will try again if more sulpur models or a new hadsum version released. 4.12 and 4.13 wont run on my athlon although the sulpur model ran fine.
[/quote]

Well the problem seemed to be the different types of anti virus packages i have been using and all so my firewall.

it turns out that anti vir xp and AVG is incompatible with some hadsum version like 4.12 and 4.13.

I also found that zonealarm crashed the model when it asked if this program should have access which is a pain because you dont know what hadsum version you are going to get from the server so when the model changes you have to update your firewall with the right exceptions, hadsum 4.13 --> hadsum 4.14 etc so i have now removed this (it was also causing problems with the new build of firefox, im behind a router with a hw firewall so should be ok)

I have just changed to Avast and now my athlon crunches any model that the server gives it without complaint, I might even be able to return a fully processed result from this computer which has so far evaded me since joining CPN.
ID: 17497 · Report as offensive     Reply Quote
Profile geophi
Volunteer moderator

Send message
Joined: 7 Aug 04
Posts: 2167
Credit: 64,480,464
RAC: 4,170
Message 17520 - Posted: 29 Nov 2005, 2:52:00 UTC - in response to Message 17497.  

I have just changed to Avast and now my athlon crunches any model that the server gives it without complaint, I might even be able to return a fully processed result from this computer which has so far evaded me since joining CPN.

Good to hear Cluster. The AV problems are probably more widespread than we know. It could also probably be worked around by excluding the boinc folder and subfolders from access and scheduled AV scanning.
ID: 17520 · Report as offensive     Reply Quote

Questions and Answers : Windows : Still got problems with my athlon xp-m

©2024 climateprediction.net