Wirehead Studios

I dunno, it seems that every time I post on this forum these days it's to do with asking for help fixing my crap PC. Sigh. Well, here we go again:

Fresh back from resolving my RAM issues (thanks everyone BTW) just in time to receive my brand new AMD Sempron 2400+ CPU through the post I spent in total 3 hours getting the bloody heatsink on and got everything properly installed and marvelled at the inproved loading times on pretty much everything I ran.

Then, while I was playing Juiced the PC restarted after 10mins. Initially I though nothing of it (happens all the time with said game) and tested out a few other games. Each time the PC restarted itself within minutes of loading without so much as a BSoD to give me half an inkling of what was wrong. I tried unchecking "Automatic Restart" in the advanced settings in My Computer but this only had a temporary effect. I've also tried disabling the video shadowing in the BIOS and this gave a false impression of working up until about an hour ago. I've checked the temperature of my CPU numerous times and so far it hasn't broken 40C.

I could be wrong, but I was thinking that perhaps my old Mainboard is putting some sort of cap on the CPU temp which is perhaps too low for my new processor. Could this be the case, and, if so, is there any way to raise the bar? Any other suggestions would be much appreciated also. I really don't want to spend another 3+ hours refitting my old proc again.

If there is, it'll be in the bios.

You could try a better, faster fan, mind.


My sons did this a lot, and when I took of the fan from the CPU, all the thermal paste had 'gone' and the CPU had over heated the heatsink so it was tarneshed, like the bottom of an old pan!

new paste did the trick, and getting the heatsink on exactly right.

it was an AMD2000+ BTW. ran well for about 4 months, then he built a new PC, and it is sat under his desk, but I'm told its still OK.
Try the BIOS, and check your paste and fan.

If this started when you added the new CPU, then I'd suspect the possibility of either a defective CPU, or else a voltage problem.  Check your voltage levels, either in the BIOS or if your mainboard included a monitoring utility, check them with that too.  Make sure they're within specification for the CPU.  If your power supply is weak it can cause random reboots, BSOD's, or any number of odd behaviors.  Make sure you don't have the heatsink touching any capacitors either, and make sure it's socketed correctly.  Sempron/Athlojn 64 systems shouldn't have this problem, but it's good to double check anyway.  Sometimes those thermal diodes on the mainboard aren't entirely accurate or reliable as well.  Make sure the CPU is socketed all the way in, too, and that no pins are bent or weak and that you've used enough thermal compound between the CPU die and the heatsink.

Another thing, sometimes overlooked, is that you might have accidentally loosened a cable or interface card.  Make sure everything's securely seated.  A friend of mine has a problem with losing the modem, and it's because vibration likes to work the the card loose in the PCI slot and the port vanishes, and the computer gets unstable.  Reseat the card and voila, problem solved.

Sometimes a BIOS flash is needed, and sometimes clearing the CMOS data can help with any bad settings that might be interfering with the system's stability.  I'd also nose around google and see if this is some kind of known issue before tinkering in earnest.  If it's only happening when running specific games, it may be a driver-related compatibility problem that's showing up with the new CPU, but wasn't obvious with the old one.

Well, while the reboots have seemed to be indiscriminate for so far, I have noticed that they are happening more often with certain programs than with others (it happens more often with GTAIII than with San Andreas, for example, yet SA is by far the more demanding game).

The problem with checking the voltage is that I was all over the manual and there is nothing to state what it actually requires. Nor is there anything about recommended running temperatures, minimum power supplies ar anything else. Basically a diagram showing you how to put the thing in (which I knew already).

As for the thermal paste, is it possible to buy that stuff on its own? I ask in the event that I may need to remove everything and start over. Also, is it possible to use too much (please excuse me, this is the first time I've ever upgraded a CPU)?

One thing I am absolutely shit-scared of doing is updating my BIOS. The reason being that this old crate was second-hand when I originally got it and such an upgrade would be permanent. I'm afraid of messing the whole thing up completely.

Given everyone's responses so far, I'm most inclined to go down the power supply route, however without any guidance as to what voltage is needed I'm kind of screwed for the interim.

Power Supply have enough watts? ]LoRE[ had this problem when he got 1gb RAM (upping from 512), his computer would randomly restart. He had a 330 (I think) powersupply, then switched to a 480 watt and he no longer gets the random restarts.

Hmm... so it reboots more when it's running a resource intense application.  That suggests it could be power supply, or temperature related, but any time it's a game you have to look at drivers as well.  A lot of times a simple driver update can answer the question.  If you update and it goes away, problem solved.  If not, well, it's cheaper to update a video driver first before swapping power supplies around.

Yes, you can buy thermal paste separately.  There's a great product called Arctic Silver 5, it's the best heat conductor paste around.  Otherwise, white thermal paste is sold at just about any computer supply shop.  It is possible to use too much of it, but more often than not it's too little that's the problem if you're overheating.  You should at least be able to see some kind of temperature in the BIOS.  Just record what it's running.  We have your CPU spec, so I can let you know if it's too high once I have the temperature data.  The best time to get a temperature is after the PC has been on for a while, so if you can check it from within Windows that's best, otherwise run the PC for a while, reboot and go into the BIOS and get the temperature then.  If you do buy something like AS5, you must remove all the old paste off the CPU before applying a different kind of paste.  I've found lens cleaning cloths work fantastic for this.  Just get the kind like they use on camera lenses or rifle scopes.  They're lint free and pull the putty off like nothing else.

It's also pretty easy to check the power supply's rating.  Just open the case and jot down how many watts it is.  Let me know your hardware specs beyond your CPU (graphics card, sound card, how many drives and what type, etc) and it'll at least give me an idea if you're underpowered.

Well, I ran a virus scan last night and my comp came up completely clean (odd in itself).

My CPU's temp tends to hang around the 30-35C / 85-90F mark. This is after manually rebooting whenever the comp restarts as I haven't been quick enough on the draw with the delete key yet. Power supply and PC stats to follow.

Also, this might be  :offtopic: but I've noticed that since installing the CPU the little red LED on my analogue joypad keeps flickering on and off. The analogue mode remains enabled, but on occasion the light goes off altogether. Could this be power-related?

Your temperatures are excellent, so I don't see an overheat being the problem.  Difficult to say on that LED... those connect through the gameport, which is also a MIDI connector and is run through the sound hardware.  Check the joystick cable to make sure it's not loose or damaged.  Also, if you use a sound card instead of onboard sound, make sure it didn't get knocked loose in the slot.  Otherwise, double-check you didn't knock any jumpers loose on the mainboard when installing the heatsink if you use onboard sound.

Thanks, Pho. I'll check that out later. My joypad is a USB, so I'm pretty much guessing here anyway.

So far everything's pointing towards a weak power unit. Here's the output specs verbatim from the sticker:

MAX POWER OUTPUT: 300W (+3,3V & +5V = 200W MAX)

My PC specs as they stand are as follows:

1.3GHz AMD Sempron 2400+ CPU
512MB RAM
Inno3D GeforceFX 5500 256MB 3D card
Matshita 4x (I think) DVD-ROM drive
Memorex 32x MAXX CD-RW drive
Maxtor 6L040J2 40GB master HDD
Samsung SP0802N 80GB slave HDD
FIC Mainboard AZ11EA motherboard

From what's been said, I'm thinking that a better PSU is the order of the day, but I'm still unsure of how much power the mainboard can take. This model is discontinued and details available on it are sketchy. That said this comp is only 3 years old.

300 watts is underpowered.  Even Athlon specs require a 350 Watt PSU minimum, and I wouldn't want to go below that with a Sempron, especially with two hard drives.  You don't have to worry about the mainboard taking too much power.  The mainboard only draws what it needs, and you really can't have "too much" power for a PC.  The only thing really power-related that can harm the mainboard, besides a defect causing bad voltage from the power supply, is if you have a fan that's over wattage plugged directly into the mainboard.  It's not a published spec, but most boards can't power a fan that draws more than 4 watts (0.3 amps) from the 3 pin connectors built into the main board.  Some newer boards can run up to an 8 watt fan, but you have to check the manual to be sure.  If in doubt, don't go over 4 watts.  This applies only to fans powered directly from the main board.  If you are running a fan from the power supply's 4 pin cables it's not a problem, even if there's a single yellow wire that connects to the 3 pin connector on the mainboard.  That's only the tachometer line and is of no concern for power demands.

I'd recommend at least a 460 watt PSU for what you've got.  The brands I favor are Enermax and Antec, but I don't know what's available where you are.

That would explain why my floppy drive disappeared (not that I use it anyway)!

I've had a quick scope around the net and a decent PSU appears to be cheaper than I first expected. I'm off work for a week over my bathday (sic) next month and the money I save in not going to work should cover the cost alone. ;) In the meantime, for gaming purposes I'll see if disabling my slave HD has any effect.  On the topic of fan watttage, it does plug into the motherboard, but I'll need to keep digging on that score. I found the specs for the CPU itself and the temp turns out to be WELL within acceptable levels, but nothing on the fan yet.

Thanks once again. I'll let you know in a few weeks if I get the problem sorted (it's worth noting that NOT ONCE did my computer reboot while playing Gen the other night, so I can be thankful for small mercies  :rules: ).

I'm also starting to think its your PSU , it would explain some things like why in some games the PC restarts and with others not .
because some games use more power from your CPU AND VGA card

its good possible that it doesnt crash that fast in San andreas sinds its more optimized then vice city .
making you use more power in Vice city .

Aye, and Gen is Q3 tech, which isn't that CPU hungry.  I'm certainly happy that's running well!

*touches wood*

Well, I got a new 500W PSU yesterday and, guess what? IT DIDN'T WORK!! I am now at a complete loss as to what is causing these bloddy random reboots. I've tried drivers, virus scans, AGP settings and other BIOS fiddling, reinstalling WinXP, a new PSU, IDE cables, the heap. I can't think of anything else. I'm very close to taking a blunt object to my tower. I'm past the point of getting pissed off and am almost at the stage of admitting defeat. Knowing my luck, I'll put my old proc back in and it will still reboot. I can't take much more of this. :wall:  :wall:  :wall:  :wall:  :wall:  :wall:  :wall:  :wall:  :wall:

In that list I'm not seeing ram-tests.. you probably already thoroughly checked the dimms, but just in case: are you sure it's not the RAM?

on the motherboard , check your condensators , if there have no longer the from of cilinders ( they can stand thicker then usuall ) that means they are dried out or are just broke .
I know I had this problem since gigabyt used a bad type of condensators .

I do have Memtest. I wanted to test the RAM, but without a floppy drive that's pretty much impossible.

This is when the "fun" begins.  Something has to be causing the problem.  Upping the available power was still a good idea as you were definitely under spec, but if that didn't fix the problem, something else must be the culprit.  Let's run down the list.

You may want to clear the CMOS data and take the BIOS settings back to factory.  If there's one thing set wrong, like a voltage level, or a bus timing, that can cause instability.  I know you checked stuff, but sometimes it's easy to miss.  The CMOS data can get garbled from time to time as well, which clearing it will fix.  I've had this happen on older machines, especially when the CMOS battery starts getting weak.  If you do go to clear the CMOS data, make sure you jot down all the old settings first.  There's also the possibility of a BIOS flash needing to be done to fully support the CPU.  Not having a floppy disk makes that process a lot harder, because usually BIOS flash software wants to be loaded from a DOS boot disk.

If you can't test the memory with software, there's another way to do it.  If you have more than one RAM module, try running it on one module by itself, then the other.  See if anything changes.  If it behaves good with one module, but not the other, try putting the one it misbehaves with in a different slot, booting it with just that module, and see if it still happens.  If it does, the chip is bad.  If it doesn't, test the "good" RAM module in the slot the "bad" one is.  If it malfunctions, you know the slot is bad.  If not, sometimes pairing them in different slots fixes instability.  Some of my old computers only ran if the RAM modules were paired in specific slots.  Reverse them, and it wouldn't even boot.  If you only have one RAM module, try moving it to a different slot and seeing if it helps at all.

If that doesn't work, try removing the sound card.  If you're using onboard sound, disable it in the BIOS.  Run the machine.  See if the problem goes away.

If that doesn't work, try disconnecting anything else that is "non-essential".  CD-ROM drives, modems, ethernet cards, anything that is not required to test this problem.  See what happens.

I'm hoping you didn't get a defective CPU, or suffer an electrostatic discharge when installing it (you DID use a grounding strap, didn't you?).  If power is good, and removing components does not help, that leaves only the mainboard or CPU as the culprit.  I know you'll hate doing this, but try swapping the CPU back out.  If the problem goes away, you either have a defective CPU or the mainboard is not fully supporting it for some reason, in which case I would suspect the BIOS revision.  I've seen situations where components "just don't get alone", or a mainboard that runs good with one part won't run at all with another.

Do yourself a favor and invest in a floppy drive as well.  I know they seem obsolete, and people have this urge to "do away with them for good", but they're cheap (only $6.00 US for a Mitsumi 3.5", $8.00 for a black one) and they can be an ass saver for certain things.

Well, after a bit of BIOS fiddling I got my floppy drive up and running again. I haven't ran Memtest fully yet (I'll do that at bedtime). I did earth myself before putting the new CPU in, but because of the position of the socket getting the old fan off and the new fan on took over an hour each way, so I'm not ruling that out.

My BIOS has 2 options: Load Optimised Defaults and Load Fail-Safe Defaults. I'm not sure which is the factory setting. I haven't tried disabling the onboard sound yet, so that's worth a go.

I did actually try removing the CD-ROM drives last week, but Windows Product Activation had other ideas, so I couldn't test it out (any time I make a hardware change WPA will ALWAYS generate an invalid code for some reason, so a reinstallation is needed which requires the CD drive. Rock and a hard place, eh?).

I'll give the RAM swapping a go, as I'm willing to try anything that has any possibility of sorting this out.

Ugh, WPA strikes again.  Whoever dreamed that up should suffer a slow and painful death, followed by eternal torment in the lowest depth of hell.

It was woodsman .

I had a similar experience with this yesterday. I spent all most all day fixing my box. I booted my computer and it said "failed memory test". I thought that was weird, and it did not boot, no screen no nothing. So I got inside my computer, and tried to re-feed my RAM, I switched my two gig sticks around the available slots. Then the computer finally booted, so I took the opportunity, to reset my CMOS, and then it booted from the OS.

However, then it under clocked my CPU, down to 1.15 GHz! From a 2800+ processor! So that was a bitch to fix. I went back in my BIOS, and reset my defined settings that were originally there. Then it did the "failed memory test" thing again. So I did the same thing as last time, went into the BIOS and farted around in there for a bit. I eventually narrowed down the problem to the CPU over clock, I thought, that was weird because how could it fail a memory test, if it just doesn't like the over clock? So I originally had the CPU frequency set to 200 MHz at a multiplier of 8.5. So I tried a frequency of 166 MHz at a multiplier of 11.5. Then everything was happy hunky dory. I don?t know if this will help you any but it's worth it to post. It may very well be your RAM.

I can explain that one.  If you overclock the front-side bus, and the RAM is not rated to the higher bus speed, it will cause errors like nothing else.  For example, PC2700 RAM is rated at a 166 MHz FSB.  If you set the FSB speed to 200 MHz, the RAM test will tank.

Right, progress report:

Below is a table of combinations of RAM that I've tried so far and their various effects. My motherboard has 3 slots, and I have 3 RAM chips in total: 2x256MB (A & B) and 1x128MB (the faulty one, C). All were tested using the GTA Liberty City mod for Vice City as it is the most consistent and always reboots within 10 minutes.

COMB    Slot 1   Slot 2   Slot 3  
1              A           B           C
2              A           B            -
3              B           A            -
4               -            A           -
5               -            B           -
6               -            B           A
7              B            -            A
8              B           C           A
9              B            -            -

RESULTS:
1. My original setup, before the CRC errors started.
2. The fail-safe setup, fine up until I installed the new CPU.
3. The PC still reboots with this setup.
4. Ran OK with no restarts, even after 3 hours of solid play.
5. Ran OK too, but I only tested it for about an hour.
6. Started rebooting again.
7. Same with this setup.
8. Held firm during gaming (usual memory glitches aside), but when I started up my PC again later on it rebooted the first time I ran Internet Explorer. This is the first non-game reboot I've had.
9. Ran fine for over 2 hours.

I choose my words very carefully here as my PC has a habit of making an arse out of me as soon as I form anything even remotely resembling a conclusion. I haven't tried every possible permutation yet, but from what I've gathered so far the problem seems to involve having more than one RAM chip installed. Realistically, though, I'm none the wiser as to what the actuall cause is.

I sincerely hope it isn't anything to do with my BIOS. Right now I don't have the ?25-or-so to reflash it. Nor do I have a credit card to buy it even if I did have the cash.

It seems that "slot 3" is this main thing that is giving you a hard time. Motherboards are nortorious for having bad RAM slots. If you still have a warranty for that motherboard, I'd take the opportunity to get it replaced.

Well, it's true that RAM chips can be temperamental buggers, and I THINK I've sussed out the problem. It isn't slot 3 per se, it would seem that chips A and B (my two 256MB ones) are conflicting. When I use either one, the system seems to run fine, and if I use either one in conjunction with my dodgy 128MB chip, there are still no restarts (apart from the odd BSoD but that's irrelevant), but whenever I run both together the system automatically throws up the proverbial rope.

I can't really be arsed using the dodgy chip as it has an adverse effect on eveything I do (downloads, gaming, not to mention those bloody CRC errors), so as it stands, I'm down to 256MB RAM in slot 3 for the interim. The only game I can't run properly at the moment is Doom 3, which I can live with for a few weeks until I get new RAM.

God, I hate computers ...

Nice work on the testing, we're getting somewhere now.  I've seen this kind of RAM conflict before, it's very common with older SDRAM DIMM modules.  DDR doesn't seem to suffer from this problem as much, but it does happen.  Some suggestions:

1)  Check your BIOS information for your "bank interleave" settings.  Make sure you don't have anything unusual or aggressive set.  Usually leaving it to "auto" is best.

2)  Check your BIOS information for your CAS Latency timings for the RAM chips.  You could try relaxing the CAS timings on the modules.  Typically higher numbers are slower.  Example:  2-5-5-6, 2T could be changed to 2.5 6 6 8, 1T.  These are examples, I don't know actual CAS timings since they're specific to your RAM chips, so go by what's available in your BIOS.  The "By SPD" option reads the timing off the chip, and usually that's default setting, but if the chips are timed differently that could be causing your problems and the BIOS doesn't always display the SPD-dictated timings.  Check the timings for each individual bank, and set them as slow as possible, setting both banks the same, and test that out.  If that makes it behave, you'll run a little slower overall, but at least you'll be stable for the time being, and you can try incrementally increasing them a little bit until you find your fastest stable state.  If it doesn't help, then it's possible you've got modules that flat out don't like each other.

Now if it's a matched set of RAM sticks, then you could have a problem with the system addressing a specific amount of RAM.  That's either mainboard or CPU.  This is where it would be best to see if you could test the board with the same amount of memory using a different pair of matched memory modules and seeing if the problem persists.  I've also seen it where specific brands or specific models of RAM were incompatible with specific CPU's.  That's rare, but it does happen, especially with cheaper "off-brand" or generic modules.  

Since this is a problem with both modules in at once, it sounds like the 256 MB RAM modules you have are either from different manufacturers, and/or have different CAS latency timings, so give that a shot first.  I hope it helps.

All 3 chips are different makes, so the timings seem like a good place to start. I'll give those a go tomorrow. Right now I'm off to the Euro server for some uninterrupted fragging ;)

EDIT: After discovering that there was absolutely NO MISSION of getting a decent game of Gen on 256MB of RAM I decided to go ahead and check the BIOS again with both chips installed, hence my jumping in and out of the server tonight.

The bank interleave was completely disabled. Also I couldn't find any options relating to CAS latency, so I couldn't check that out, but I tried all of the available options with no joy. Looks like new RAM is on the cards ...