CPU/Mobo Is this a PSU problem?

Status
Not open for further replies.
K

KingKrool

Forerunner
OK... this is a big one, so please bear with the long description

History:
1. The earthing at my place was damaged, as a result of which my PSU has been damaged. However it does not give me much problems most of the time. As long as power keeps flowing in, there is no problem, only when it is switched off from the mains (or the rear PSU switch, or the UPS), then on restarting, it will reboot once (and only once). Since I have an inverter+UPS this does not normally bother me.

2. When I tried attaching a third harddrive once, it refused to boot. So I removed the drive. It still would not boot, but fixed itself after an hour.

3. I recently bought a DVD writer, attached it and there was no problem.
Current problem:
1. I bought 512 MB of Dynet RAM
2. When I added it in Slot2 (Slot 1 contained previous 512 MB stick), the computer would start loading windows and freeze. Or it would give a blue screen with errors like PAGE FAULT IN NON PAGEABLE AREA or BAD POOL CALLER. Booted into safe mode, but it BSODed after 3 minutes.
3. I removed the old stick, put the new one in slot1. Computer booted and appeared to be fine.
4.Put the old stick in Slot 3. Computer booted. But when I started any 3d application (3dMark 2005, HL2, NFSUG2) the computer would freeze (at first it just gave a d3d9.dll error in the event log, afterwards it just plain rebboted)
5. Removed the new RAM stick. Now only the old RAM was in slot 1.
6. Computer booted, but suffered from the same random freezes.
7. After leaving the system unplugged for a little while, I put Old RAM in slot1, new in slot 2, again it would not boot.
8. Removed new RAM, system booted and I downloaded windows memory diagnostic tools (sorta like Memtest, memtest wouldn't work at all for some reason, it justkept on printing AX...BX.....DX.... over and over again.). Also ran 3dmark, it ran fine.
9. Ran the test with only old in Slot1. All successful.
10. Ran the test with only new in Slot3. All successful.
11. Ran the test with new in Slot3 and old in Slot1. All Successful.
12. When I tried rebooting it gave a BOOT DISK Failure. On checking in the BIOS, the IDE drive was not being detected on Primary master.
13. Switched off, pressed the IDE cable in, and restarted. This time it hung at that Windows slider screen.
14. On rebooting, it gave the same disk error.
15. Tried again, this time it booted into safe mode, no BSODs. Then when I tried regular mode, it would go to the slider screen, and the slider would just keep on sliding.
16. Tried ubuntu live CD. It loaded successfully, came up to the full GUI, when I clicked on the Menu, everything froze.
17. Removed the new RAM stick, computer has booted up and I am using it now. It appears to be just fine.

I think it must be a PSU problem, seeing as how the computer freezes whenever there is some load on it, ie when the GUI is present, GFX card is fired up, boot process etc. Plus the anomalous HDD detection.

But confirmation would be nice. My pop will throw a fit if I buy a PSU without reason, I have spent enough on the comp of late as it is .
 
i posted a very similar problem today...if you have a dmm, use that to check the voltage....sunny's guide on voltage testing a psu will come in handy (from the guides & tuts section) else use gigabyte's easy tune utlity to check the psu rail voltages while stressing the pc...
 
Already read that one.

Well I have no DMM.

As for Easytune, it is no use, because the moment the PSU is stressed (invariably by the GFX card) the whole computer freezes. Also, most of those stress inducing moments are in a full screen mode. Furthermore, half of the time, the computer does not boot.

Plus there is the simple logic:

1. Lates problem arose when I installed the New RAM

2. Memory test on it is ok

3. Therefore must be hardware or s/w prob (other than RAM)

4. Even Linux faces the problem

5. Therefore it must be hardware, and probably PSU cos that is the only relevant component that is used in a different way in the memtesting.
 
KK, the two rams are not compatible. meaning they have different spd values stored on them.(if any at all)
Do one thing.
go to bios setup , and in the memory timings window. select manual (not SPD) and relax the memory timings a bit. for e.g. if the existing timings are "auto" or 3-3-3-7 increase it to 3-4-4-8 etc .....

This will force the m/c bios to accept this new value for both the ram modules.

kindly post back ur observations ....
 
follow deejays post & added to that run memtest(with both rams installed), see if your ram's are fine, this looks more of a case of incompatible rams, running@different timings.

try to post cpz screenies of your spd's slot 1 & 2 ...
 
Well, I will try the timings (I was running manual at 3-3-3-7 whereas the new RAM has spd 3-4-4-8), but then how come Windows Memory diagnostic runs fine, and only gfx etc seems to stress out the thing?

Plus, what about the missing boot disk?

Anyway, will post about it after 24 hours, cannot play with computer b4 that, need it right now.
 
^graphics tend to stress the memory as well, that's why. i can vouch for memtest & prime95.

why dont you set manual to 3-4-4-8, chose with higher latencies's ;)

missing boot disk, huge post got lost in there, make it detailed but concise.
 
well KK, you are the software expert. you must be giving us solutions for this anamoly in windows.
i think DMA (HDD etc..) operations are affected much more by memory timings than windows API based memory access. which do have some form of inherent error correction code. (just a hunch, correct me if i am wrong)
and Windows Memory diagnostic might be using those ECC API's. (another shot in the dark)

well anyways you try out the relaxed settings .....

the psu is not ruled out too but first lets omit this ram timing issue as the cause of ur problem.
 
No, windows diagnostics is exactly like memtest. Same bootdisk. Same tests.

And those tests were fine. But I will try the timings, so let us see.
 
No, windows diagnostics is exactly like memtest. Same bootdisk. Same tests.
Click to expand...
run it from inside windows, with applications running, this is from memtest's manual ;)

Unlike other memory checking software, MemTest is designed to find all types of memory errors including intermittent problems. Therefore, it needs to be run for several hours to truly evaluate your RAM. The best way to use the system is to set it up to run overnight, for a fully rigorous test. Letting MemTest run long enough to achieve 100% coverage will only identify RAM that frequently corrupts all data written to it. You can also run MemTest while you use your computer for other tasks, which will help identify memory errors which only show up while the computer is under load.
Click to expand...
 
Well, today morning, the computer gave me the same "disk boot failure..." and does not load. I switch it off, and then back on, and it works.
 
To eliminate the PSU prob, unplug all your optical drives and secondary HDD's and test. I also am of the opinion its more of a mismatch RAM problem.
 
Well, I currently have just the old RAM plugged in at 3-3-3-7 (which it has been doing for two years now), and it still gives a boot disk failure. (That is what I meant above). Plus, I did once run it at 3-4-4-7 (but not 3-4-4-8) and it gave errors. I will try 3-4-4-8 in the evening, I have a presentation today and cannot take the risk right now.
 
These are the current timings and SPD for the old RAM (transcend). It is workingfine (apart from that boot disk thing in the morning, but I have not yet stressed it with 3d gfx)





Just to give u some info on the earlier setup.
 
hey kk

two things i noticed

your problem is either in morning and when pc is stressed so may be the ram is at fault but do check you psu at some friend`s place or test some other psu too.

runing pc without earthing is like living on the edge, i have lost one tv tuner, one mother board and a cd writer to bad earthing.

second point : you mentioned you have one invertor and one ups, once i had this kind of setup for my operation theatre instruments and my service engineer advised me against it. he said some techie things which went above my head and told me to remove ups from my ot. he told for optical and electronics instruments you either have invertor or ups prior to your instrument. as only invertor will cause pc to reboot in case of power failure so you are better off with only ups protecting your pc.
my 2 cents

ps: do you get mild electric current shock while touching your cabinet??
 
I used to get shocks, but the earthing is fixed now.

I too know of the inverter+UPS problem, but that is not the issue here either, as the problem there is only when the UPS runs on the inverter power (which is currently not the case). Actually the problem is in the afternoon and evening.

OK, so now I installed the new RAM in slot 1, and removed the old one. I set the timings manually in the BIOS, and after booting into windows I ran Memtest for a few minutes (264% coverage), then switched to 3dMark05 (which would previously immediately hang/crash/freeze). But this time it ran perfectly, indeed there was only a loss of 4 points (probably due to the higher latencies).

Anyway check the following CPU-z screenshots:



 
Unmatched chips is really a non issue with A64 in single channel. I'm running a 1G stick of Micron (which suddenly resurrected itself from death :P) and a 1G stick of Twinmos UCCC, set to 2.5-3-3-5 stable upto 230MHz. I have a feeling that the new RAM stick is faulty. Its best if you could memtest it in another rig.
 
But the new stick by itself does not give any problems in memtest etc.

Anyway, things have now gone from bad to worse..
I put in both sticks, it booted and I ran memtest, it immediately gave errors (either stick alone did not). So I stopped it, and tried running 3dmark05 which completed the first test (higher fps than before), and then hung just after the loading screen for the second test (blank screen). I rebooted, and the system booted and gave some stop error (7E, but no other label). Then I tried again, it booted, the mouse cursor moved but it wouldn't select anything. So I switched off, pulled out the new RAM and rebooted. It crossed the windows slider screen and then froze with some strange coloredbocks across a generaly black screen. Next time when I booted, it refused to start even in safe mode, with the following error:
Windows XP could not start because the following file is missing or corrupt: \WINDOWS\SYSTEM32\CONFIG\SYSTEM

I am in the process of repairing the error via the method mentioned in
http://support.microsoft.com/?kbid=307545
This message was typed on the laptop.
 
OK, now I am back on the desktop

Here is the error I got in memtest (surprisingly, windows memory diagnostic, which is just like memtest86 did not give any errors yesterday, and I am not about to try it again today!)

 
Status
Not open for further replies.
Top Bottom