Whoey's IT weekend blowout...

[Whoey]

I'm in the middle of doing some troubleshooting and maintenance for a coworker of my wife's home PC. It was frequently crashing, and overheating apparently.

First thing I did was pop off the side of the case, to see that there was so much dust accumulated in the stock Intel cooler that the fan was having no effect on the heatsink. Pulled it out and apart, cleaned it out (really wish I had an air compressor to blast some air through all the little nooks and crannies) and put it back in with new thermal compound. I then popped in my Ubuntu Live CD and checked the smart status etc on the Hard disk (which seems fine, although it gets thrashed when you boot the actual installed WinXP) And also ran MEMTEST. That spit out errors like no tomorrow. So I pulled all the memory, and installed one at a time, and ran memtest until I rooted out the bad stick. 1/3 1gb sticks were totally bad... memtest reported over 32000 errors in 5min on that dimm, the others were clean. I popped back the ones that were ok, and 2x512mb that I had laying around... no point putting 2x1gb, as it's 32bit OS, and the video card has 512mb... (4gb max including video ram) Booted the installed system, and did a bit of driver and software updating (since I have 100MB connection, and believe she's on mobile internet) including putting a new antivirus in place as the installed one seemed corrupt or half installed or something. Ran a full scan, and found a few suspect files that got cleaned.

I've been told not to delete any files as the owner didn't have a chance to update her backups. I have installed CCleaner and was tempted to give it a run as it usually won't affect any personal files, but as I'm unsure I'll suggest she runs it once she gets her backups done. I have also planned to do some sort of defrag before I return it to her.

Can you think of anything else that may help? I plan to tell her she should invest in a new 5-10eur CPU cooler as I'm not too sure that stock intel is capable anymore... although the CPU didn't go over 48C when the AV was doing the full scan (it hit 68 and crashed before cleaning previously doing less intensive tasks). There's an open IDE connection and I'm tempted to look through my old IDE drives to see if I can throw in an extra 160/200GB or something as she's got about 230/250gb filled on her SATA drive that's installed.

She's a self employed teacher, so she's really limited in budget, and I'm doing whatever I can to help without incurring too much cost. She does want to get a new machine, but can't really afford it atm so whatever I can do to help prolong the life/usefulness of this machine is welcome.

[mvyrmnd]

Unless you're planning on overclocking a system , the stock cooler is just fine. I've never had an overheat issue with a properly installed, clean, stock heatsink.

If the RAM is maxed, the drive less than 80% full and has been defragged, the startup items cleared up (I use HijackThis to tidy it all up) then the system will be going about as well as it can withougt either a full format or major upgrade.

The single best upgrade these days is a SSD. My 4yo MacBook is giving me no reason to replace it since a SSD upgrade.

Rather than add a second, slower, drive - upgrade her to a 500GB drive. it'll be far better in the long run.

Also, Google for and run an app called Crystal DiskInfo. It'll give you a readout of the drive's SMART status. if it reports ANY issues at all, replace the drive. It may only report a few reallocated sectors, but they have a habit of spreading. I treat any SMART warning as a red flag of impending doom.

[Whoey]

In general I agree, but once you pull a stock Intel cooler off they never attach just right again, I hate the snap down system and prefer the screw or even latch system the older P4s had. I had to pull it to clean it properly as it was literally caked with dust between the HS and fan and the only way to get the fan off is to pull the entire thing out (there's never enough space around it in the case).

SSD is a good option for system disk, I recently replaced the failing drive in my brother in law's laptop (160gb sata) with a 120gb SSD SATA and also added 1gb of ram, and it vastly improved the system. We also moved his main system disk in his desktop from the struggling 1TB to a 120gb SSD, and that also vastly increased the system benchmarks.

I don't think it's an option for her, so I'm trying to go with mostly software fixes or whatever spares I have at hand (which admittedly, is rather a lot). Her existing disk is a 250gb 7200RPM 8mb cache model, which is pretty decent. The only SATA disks I have on hand are either dead/malfunctioning or in use in our systems. I have a lot more IDE disks as none of our setups (or my NAS units) accept IDE, so they are leftover from older setups.

[mvyrmnd]

In general I agree, but once you pull a stock Intel cooler off they never attach just right again

I can't say I've ever experienced that. They are tricky little smurfs, though.

[Whoey]

Rather than add a second, slower, drive - upgrade her to a 500GB drive. it'll be far better in the long run.

Also, Google for and run an app called Crystal DiskInfo. It'll give you a readout of the drive's SMART status. if it reports ANY issues at all, replace the drive. It may only report a few reallocated sectors, but they have a habit of spreading. I treat any SMART warning as a red flag of impending doom.

I ran SMART off the linux live CD... it's rather handy. It was green. What I don't understand is why although some disks report a Healthy status, the system thrashes the hell out of it when loading programs or the OS itself. This is the case with the 1TB in my brother in law's machine, it's a solid healthy disk according to SMART, but once I migrated the OS to the new SSD, and symlinked the USER dir to the large drive (boy was that a real PITA! MS does not seem to factor in migrating to smaller SSDs), the boot times and program loading changed completely.

I'll have to dig around, but most of the spare drives I have won't really be worse than what she has, I believe I have similar spec 7200RPM 8mb cache IDE models and in any case, for extra storage the speed is not as crucial as it is on the main disk.

[Whoey]

Apparently spoke too soon, CPU is overheating again (was rising towards 60 with no actual tasks running apart from background stuff), I suspect it was fine when laying flat on the desk, and even though I gently placed it on the floor vertically, it must have let go. 

[mvyrmnd]

60° is hot, but not what I'd consider dangerously so.

Modern CPU's have built in thermal throttling that cuts in at 95°C. I've had chips on stress tests that have maintained 95° for more than 24 hours with no harmful or lasting effects.

For an idle temp it's not ideal - but I don't think it's a critical issue.

[Whoey]

according to the Intel page, the IHS should not go over 60.1C, and it's always over according to the Intel Desktop Tool

[mvyrmnd]

Fair enough - but I've never had one fail at even significantly higher temps. I suspect that's a highly conservative number.

[mvyrmnd]

... once I migrated the OS to the new SSD, and symlinked the USER dir to the large drive (boy was that a real PITA! MS does not seem to factor in migrating to smaller SSDs),

Symlinks are required to pull this off under MacOS (this is how mine is set up), but in windows you can simply move the folders and redirect them in the registry - no file system
shenanigans required.

[Whoey]

I found a rather helpful guide somewhere... but I had to faff about with it for quite some time to get his actual user account working again. I'm glad I keep all my media (music/video/etc) filed neatly in separate directories and partitions rather than use that library system in win7, but he seems to like it. Although these days 95% of my time is spent in Ubuntu...

[derekmac]

Want to quickly find storage bloat?

This link has a tool for Windows, Mac, and Linux

http://www.itworld.com/it-management/378130/how-find-out-whats-taking-space-your-hard-drive

THANK YOU!!!  I've been looking for something like this, and this does the trick!! 

[detron]

... once I migrated the OS to the new SSD, and symlinked the USER dir to the large drive (boy was that a real PITA! MS does not seem to factor in migrating to smaller SSDs),

Symlinks are required to pull this off under MacOS (this is how mine is set up), but in windows you can simply move the folders and redirect them in the registry - no file system
shenanigans required.

You can also avoid the manual registry editing by going to the properties of the folder (pictures for example)  at the location there,  and select move files to new location

Sent from my SPH-L720 using Tapatalk now Free

[detron]

Apparently spoke too soon, CPU is overheating again (was rising towards 60 with no actual tasks running apart from background stuff), I suspect it was fine when laying flat on the desk, and even though I gently placed it on the floor vertically, it must have let go. 

If you replaced the thermal paste, and dusted behind the mobo,  then I would put Decent odds on the processor itself being the thermal issue.   

Sent from my SPH-L720 using Tapatalk now Free

[Whoey]

it's in a definite state... I'm trying an XP repair install...

Could not clean under the motherboard due to some smurf using the wrong thread screws... I don't seem to be in any shape to do much except try and get it running long enough for her to do a backup of whatever's important. I'm not sure if a clean install would be a solution to some of the issues it's having, but it seems to run just fine off live CDs

[AlephZero]

it's in a definite state... I'm trying an XP repair install...

Could not clean under the motherboard due to some smurf using the wrong thread screws... I don't seem to be in any shape to do much except try and get it running long enough for her to do a backup of whatever's important. I'm not sure if a clean install would be a solution to some of the issues it's having, but it seems to run just fine off live CDs

Cleaning under the MB will not do any good thermal wise... If you did replace the thermal paste then I'd say the issue is with the MB

[Whoey]

I've noticed that EVERYTHING seems to get hot in this machine... the Hard drive is quite warm to touch (I've had drives run at similar temps) as is the passive chipset heatsink (actually rather hot) and the Nvidia geforce 7300 (also rather hot, passive heatsink, no fan) The CPU heatsink seems cooler than it was before, but since I can't really get into the WinXP install anymore I can't see what temp it's idling at.

I redid the thermal grease on the chipset heatsink as well, the old stuff flaked off like chocolate shavings...

I've had it powered off for the time it took me to whip up a batch of my homemade Lasagna, and now trying another winxp repair install and it's failing to copy files... It has done this before, I think my CD is on the way out, or perhaps the laser head is dirty.

previous repairs copied the files to 100% then said configuring windows (not sure on the exact phrasing of this stage, this is all installing in Spanish) and then I've had it BSOD, with Page Fault in Nonpaged Area.

I ran Memtest for 10 min, no errors.

[detron]

A while back Memtest on my machine ran for 2 hours with no problem, but loads of errors aspirated after that.   Turns out I had one bad sick of ram.  1 bad 8 gig stick out of 4 sticks caused havok,  had I stopped Memtest before it finally got to ask mem areas,  I would have thought the ram was ok.
Once I got replacement ram,   I did a full memtest.  Took forever

Sent from my SPH-L720 using Tapatalk now Free

[detron]

A shot in the dark here,  but if everything is running hot,   check for proper output voltages from the power supply

Sent from my SPH-L720 using Tapatalk now Free

[Whoey]

each attempt seems to yield different results... this time it said the C drive was damaged and could not be accessed... 

I ran chkdsk from Hirens bootcd before, and it reported no errors.

Tempted to throw another disk in it and do a fresh install, mounting her old disk as storage.

[detron]

Good luck.   This one sounds frustrating.
Could be either several problems,  or one problem that affects several components.

I think the worse headache I ever had was where someone had damaged the AGP port years ago.   The damage was hard to see, but only caused an issue once or twice a week.   Drove me mental.

Basically one or twice a week they had an error.   Sometimes the screen froze,  someone they got a BSOD, and sometimes the computer would just power off.

After working one this pc for days,  I happened to notice the base of the AGP port was scored.  Put in a new  mobo.  Problem solved.

Nightmare

Sent from my SPH-L720 using Tapatalk now Free

[Whoey]

Good luck.   This one sounds frustrating.
Could be either several problems,  or one problem that affects several components.

I think the worse headache I ever had was where someone had damaged the AGP port years ago.   The damage was hard to see, but only caused an issue once or twice a week.   Drove me mental.

Basically one or twice a week they had an error.   Sometimes the screen froze,  someone they got a BSOD, and sometimes the computer would just power off.

After working one this pc for days,  I happened to notice the base of the AGP port was scored.  Put in a new  mobo.  Problem solved.

Nightmare

Sent from my SPH-L720 using Tapatalk http://tapatalk.com/m?id=10']now Free[/url]

That would probably be ok for her... gives her plenty of time to recuperate her stuff to her backups.

Doing my best, but rather than getting better, it seems worse, now the last two attempts at a repair install have said no install found to repair. I may have to get back in there with the hirens disk and clean out temp garbage. ATM re-running at least 1 full pass of memtest, 20min so far no errors yet. In the past I've found that usually if there's a memory error that affects windows, it happens pretty quick.

Could it perhaps be Memory SPD related? the old 2x 1024mb are Hexon Tech DDR2-232, and the new 2x 512mb Samsung DDR2-400 I assumed that the faster ram would just clock down to the slowest speed... that's always been my understanding of how it works.

I have it set as follows for the dual channel so that the dimms are matched:
slot 0 - 1024mb
slot 1 - 512mb
slot 2 - 1024mb
slot 3 - 512mb

I may also try swapping out the PSU as well for a newer one I have. And perhaps unplug her disk, and try setting up a clean OS on another HD just to see if the machine is stable or it's the disk. Ideally I want to finish this sometime today to return to her tomorrow.

[Whoey]

1 pass, 30min, no errors 

[mvyrmnd]

This all sounds kinda familiar.

What vintage is the motherboard? There were a batch of board from nearly EVERY manufacturer that had faulty capacitors that would fail and smurf up the clean power on the board.

Check each capacitor to see if it's bulging or leaky. If the caps near the CPU or RAM are bulgy, then the board is a writeoff.

[Whoey]

It's an Intel DP965LT mainboard, specs/manual online seems to hint at a 2006 release, same for the Core™2 Duo Processor E6600 I didn't see any obvious bulging of caps last time I had it open, but I wasn't exactly looking for any. IIRC, last time it actually got into windows, the intel desktop tool was reporting acceptable voltages across the board.

Unfortunately I don't think any of the other mainboards I have on hand can handle that CPU, as they are all older, and even with latest firmware aren't compatible according to the manuals. (and some of those boards are probably suspect as well!)

[mvyrmnd]

It's an Intel DP965LT mainboard, specs/manual online seems to hint at a 2006 release, same for the Core™2 Duo Processor E6600 I didn't see any obvious bulging of caps last time I had it open, but I wasn't exactly looking for any. IIRC, last time it actually got into windows, the intel desktop tool was reporting acceptable voltages across the board.

Unfortunately I don't think any of the other mainboards I have on hand can handle that CPU, as they are all older, and even with latest firmware aren't compatible according to the manuals. (and some of those boards are probably suspect as well!)

I've had board with bulgy caps report good voltages. It'd be worth another eyeball of the board. 2006 is at the later end of the era of dodgy caps, so you might get away with it.

If you're having trouble diagnosing an issue, you need to go back to basics. Known good PSU, one stick of known good RAM, see if you can finish the install. If it fails, the board or CPU are suspect. if it succeeds, than add one piece of hardware at a time until it dies.

[Whoey]

That's the thing, there's not a whole lot to strip off this machine, the only accessory card is the video card, everything else is built in.

I may pull the ram in favour of some other good (faster) 1gb dimms I have on hand, and see if that works better than how it is now... the original ram seems slow for the board specs imho. I also have a similar video card I can try (geforce 7600 passive)

fingers crossed... atm waiting on 2nd memtest pass to complete, then I'll try some more swapping...

[mvyrmnd]

That's the thing, there's not a whole lot to strip off this machine, the only accessory card is the video card, everything else is built in.

I may pull the ram in favour of some other good (faster) 1gb dimms I have on hand, and see if that works better than how it is now... the original ram seems slow for the board specs imho. I also have a similar video card I can try (geforce 7600 passive)

fingers crossed... atm waiting on 2nd memtest pass to complete, then I'll try some more swapping...

Then pull it! and put in the other RAM. Something has to be causing the problems

[Whoey]

yes, windows XP

[mvyrmnd]

yes, windows XP

Tell her to buy a Mac.