One of my better crunchers is down ...

[petrusbroder]

One of my better computers is down. It is a A64 3200+, with 512 Mbyte RAM and a HDD and DVD-reader.
Worked OK late last night.
This morning just dead. When restarted, it shows the BOIS-messages OK, you can see all in the BIOS. The HDD is present, DVD-drive is present, all seems OK to me. You can go into the BOIS and change the boot-order and such stuff. The comp is not OC'ed, has had no problems at all.
But it won't boot: not from the HDD, not from the DVD-drive (with e.g. the Win-XP-installations CD). It won't boot from a memory-stick either - but I am not so shure if it ever has done that (I have a linux-distribution on a 512 MByte -memorystick).

I know the following:
The graphics-card works
The power supply works well enough.
The keyboard works.

If the HDD was broken then the comp should boot from the WinXP-CD-rom.
If the DVD-drive was broken it should boot from the HDD.
If memory was so bad it would not boot I should get beeps from the computer(?). Hmm is that correct? And the BIOS should report something ...

My suspition is that it is the Mobo is broken.

Any other ideas?

I'll report back when I know more - I just do not have the time to fix this before tonight (it is sunday morning just now - and the sun is bright, 28 degrees C (= 82.4 degrees F) so it will be a hot day today - I expect some 34 - 36 degrees C (= 90 - 93 degrees F) which is quite hot considering my geographical location (just 130 miles south of the polar circle ...) The kids want to the beach (that will be a killer - more than 5 people/50 yards wink.gif), picnic, water, a nice breeze ...

 

[Philippart]

i wrote a little guide for you at TAS

 

[petrusbroder]

Thanks!
Here is an update @ TAS!

In short: its not the CPU, not the the RAM (they both work OK in FreeDOS), the HDD is OK (Spinrite says so), the comp boots into FreeDos (from CD-rom) but does not boot into: Win95 (have a good bootdiskette and CD-rom), WinXP, Linux-Ubuntu or Linux-RedHat (have those on bootable CD-roms.

Clearing the CMOS: no change.

Observation: the checking of the HDD takes about 2 - 3 times longer than would be expected; when benchmarking the HDD on this comp the burst speed (7.5MB/s) is less than the sustained speed (7.6MB/s)

Looks like the mobo?


Edit: the d...d spelling

 

[RaySun2Be]

I'd definitely try another good PSU before tearing the system down. I had a similar problem with a MB/CPU combo. It would get so far, then freeze. Swapped the PSU, and it works like a charm.

Then again, it could be the MB.

That's one nice thing about having a crack-rack with similar components, you can troubleshoot by swapping them out.

 

[BlackMountainCow]

Did you try memtest86+ yet, if it actually boots?

Does the PC show anything like "the longer it runs the bigger the prob" or "the longer it runs the lesser the prob"? You know, thermal fracture problems.

Did you check all you capacitors on the mobob for bulged caps?

I wouldn't dismiss the PSU problem so easily Peter. The PC of a buddy of mine had something similar happen, all at once, and after a week worth of investigating, I found that a capacitor in the PSU had a bulged cap which caused the 12V+ rail to fluctuate heavily on cold start and under stress.

You never know ...

 

[petrusbroder]

It is down to a choice between mobo and PSU. I can get a very similar new mobo (which would mean that I would not have to re-install Windows) very inexpensive (approx. 60 US$) and a good PSU is about 50 US$. What should I try first? The PSU - since I probably will need it later even if the "old" PSU is good and the new one does not solve the problem and since it is much easier to change a PSU than a mobo ...
Or the mobo - since the comp has been running OK under FreeDOS for more than 6 hours with the HDD and graphics card going all the time (testing it ...) without any power-problems ...

Decisions, decisions ... I prefer just now only one of the items ...

 

[BlackMountainCow]

Well, the question is, if you don't have another PSU already at home in one of your other PCs you could use for a test.

I did that once to test a PSU without too much effort.

Put the two PC cases next to each other, unhook the cables of the good PSU and connect the cables with the HDD/mobo/etc. in the other case of the questionable PSU. If you push the power on button on case one, case two will start up. If the PSU was the prob, you'll know in a second.

:thumbsup:

 

[petrusbroder]

Good idea - why didn't I think of that ...
Thanks Christian!

 

[BlackMountainCow]

So if that doesn't help, I'd go for the mobo option.

 

[RobertE]

Originally posted by: petrusbroder
It is down to a choice between mobo and PSU. I can get a very similar new mobo (which would mean that I would not have to re-install Windows) very inexpensive (approx. 60 US$) and a good PSU is about 50 US$. What should I try first? The PSU - since I probably will need it later even if the "old" PSU is good and the new one does not solve the problem and since it is much easier to change a PSU than a mobo ...
Or the mobo - since the comp has been running OK under FreeDOS for more than 6 hours with the HDD and graphics card going all the time (testing it ...) without any power-problems ...

Decisions, decisions ... I prefer just now only one of the items ...

Get both, then your 1/3rd of the way to a new rig. :evil:

 

[petrusbroder]

LOL @ RobertE.
Yeah, you are right - but then I have to get one more KVM-switch because all my ports are taken

 

[Fardringle]

Does the system freeze up completely when trying to boot from CD or hard drive, does it "see" the drives but fail to locate any bootable OS on the drives, or outside of the BIOS setup does it act as if it cannot find a bootable drive at all?

If a known good working PSU from another machine doesn't fix the problem, and the computer will boot and run stable from a floppy, then I'd say you likely either have a problem with an IDE/SATA cable (I've seen cases where a single loose wire in a cable will disable a system) or the IDE/SATA controller on the motherboard is bad. I'd try a drive cable from another PC to see what happens before you spend the money and time to replace the motherboard.

 

[petrusbroder]

It is not the cables. I have tried the system with different cables - the problem remains. Also: the computer boots using FreeDOS ...
In BIOS all the drives are seen perfectly well.

I have the impression that the system gets somewhere while booting and then freezes ... and that this does not happen when booting an 8-bit operating system from the CD-rom.

I have decided to get a new PSU (the ones I have at home are not powerful enough - 250 W) and a new mobo. I'll try the PSU first - because if it works OK I can return the not opened mobo to the store ... for a full refund. If the "old PSU is OK I know that a spare 450 W PSU will sooner or later find a home in an other computer

Thanks for the suggestions, team-mates, I'll report back the diagnosis when the computer is in working order.

 

[petrusbroder]

Here is the update on the problem:

Got new PSU (450W) and Mobo (ASUS K8N) for 110 US$.
Changed PSU: same problem as before.
Put the old PSU to an other computer: no problem at all.
Conclusion: Old and new PSU work equally well and can run crunchers for at least two hours att full load.

Changed mobo: with old PSU: Boots now from harddisk, get to the screen where you can choose if you want to start winXP i normally or in safe mode (with/without network etc). Which ever you choose: computer crashes. This is better than with the "old mobo"...

Then:
changed HDD: no change
changed all cables: no change
ran mobo with ram and graphics card only: boots OK to BIOS, add HDD boots to above screen, add DVD-drive boots to above screen.
change RAM: get to the above screen.
I put in the RAM into different computer: memory check at boot OK, does not boot first time, boots OK second time.
I put a stick of other comps's RAM: boots not at all.
Change back to the original RAM and:
put WinXP installation CD-rom into the DVD-drive: boots OK, reads CD-rom, gets to the screen with the line: "checking hardware" - crashes.

So: the problem is either flakey RAM or the CPU. I do not have any other computer with a socket 754 processor.
Now I have a good (new) mobo, a bordeline RAM and probably a bad CPU.
Then I have the old (probably good) socket 754 mobo ... and a good PSU.

Do you agree to the following:
1. I need to get a new CPU, a new stick of RAM to get the old computer to work.
2. If I get a an other socket 754-CPU and a stick of RAM I can get a new cruncher built (I have cables, HDD, CD-rom-readers, but no case ...

Is it worth to get the extra CPU, RAM and case (without PSU)?

 

[BlackMountainCow]

Phewww, Peter, quite an odyssey you got going there.

@ 1: Don't you have a local hardware dealer where you can borrow a stick of RAm for a small fee for a day? I can do that here and it helped me a lot of times to solve probs like this. Was that CPU oced? To be honest, that would be the first time I hear of a CPU actually failing (unless some moron oced to make the CPU a barbecue).

@ 2: that's correct, but I think you didn't include the "wife" factor there

@ 3: Is it worth? That's something only you can answer, depending on your electricity bill, you wallet size and the famous "wife acceptance factor"

 

[petrusbroder]

LOL @ BMC ...

The wife factor is not a problem - since I am single - and the kids do not care
No, to "borrow" a stick of memoty would be nice, and i would have worked a year ago, but that shop closed ...
The CPU was not OC'ed. I have not had any CPU fail on me yet, and ususally - if a CPU fails - it does not start at all ...So it should be the RAM ...

OTOH: buying a stick of RAM is not too hard - you can never have too much RAM in a computer. If the old RAM is OK I can still add the new one to the comp which has the least - since I have soem comps with 384 MBytes only ...

We will see - I'll get some RAM, if that is the problem I'll have the pleasure to be able to build one more cruncher soon ...

 

[petrusbroder]

Next update:

It is not the RAM: having obtained a brand new stick of RAM: the problem still the same.
Summary: The problem persists although the PSU, mobo, RAM, graphics card, HDD, CD-rom-drive, all cables were exchanged and tested in different combinations. The only thing in common is the CPU. So now I got to find a new Socket 754 - CPU ... movig to the FS/FT-forum ...

 

[BlackMountainCow]

Now that's really odd ... you'd be the first guy I get to know with a failed CPU without OC

 

[petrusbroder]

Yeah, and I got a bonus: The CPU had a 2-year warrenty; I have found the receipt and i bought it only 22 months ago!
Now i want to see how the warrenty will be honored ... there are no A64 3200+ to be had in Sweden - trust me checked.
I also - through FS/FT obtained an other A64 3200+ at a very good price - will come in in about 6 - 8 days from a trader with some 31 positive heat-evaluations - and no negative.
Lets see if this works out: I got an extra mobo, etxra RAM, spare HDD, spare DVD-drive, spare PCI-graphics card, all cables needed --- this smells like a new cruncher and the old one repaired and working for about 300 USD or less ... when the A64 CPUs come in. Not too bad ... could be worse

 

[petrusbroder]

Thanks, all team mates for your support! :beer:

 

[RobertE]

Don't forget to pick up the extra KVM switch.

 

[petrusbroder]

Yeah, you are right - I put that onto my shopping list!

 

[petrusbroder]

Two comps assambled - just waiting for the CPU. One may come tomorrow or Friday, the other late next week ...

 

[petrusbroder]

The two comps are still assabled - but no CPU has arrived. No problem - I know why.
The - after quite a few weeks - at last my "old PowerMac" (earlier 700 MHz G4, now 867 MHz G4) is upp and running. All WUs had expired, but I am now starting from scratch - for RA and for other projects ...

It has been some very tough weeks in real life: more than my share of workload: in the passed 10 working days it has been + 59 hours more than usually. Thats why I am not so often here. It will be better - in about one week or so ... count down has started.

 

[petrusbroder]

The problem is solved.

One of the two CPUs has arrived and I dropped it into the "old mobo" with new RAM, known-to-work-graphics card, HDD, CD-Romdrive, cables etc ...
Did not boot - but BOIS did something odd: detected a RAM R/W-error.
Exchanged RAM: the problem remained.

Put the new CPU onto the new mobo, with known working RAM, graphics card, HDD, etc: booted no problem. Removed the known working graphics card, HDD, CD-Rom (they belong to an other working computer ...) put in the old RAM, HDD, graphics-card, CD-Rom-drive: it booted without problems, WinXP started, BOINC started all is well.

So: from previous experimenting I knew it was a problem with the "old" CPU (did not boot in old or new mobo). Iknew the RAM worked (in other working comps) and all the other peripheral components worked.

Now I have one assambled computer with a broken mobo, but everything else working OK, and a new CPU coming to me very soon (can't return it ...). Got to get a new mobo, happily I know where to get it for some 40 US$ ...