I'm so sad. My Athlon-XP cannot handle Gentoo anymore.

[NeddySeagoon]

MrVahn,

It sounds very much like failing capacitors in the Vcore regulator, so the voltage the CPU works from is not held within the very tight required tolerances. Its a sign of a poor quality motherboard.

If you replace them - be sure to get capacitors with a low ESR, a high ripple current and a high temperature rating.
They get a very hard life in a Vcore regulator.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

[MrVahn]

Yeah. I think it is of low quality because my dad bought it for me, and he prefers something cheap than something of quality. I'll try the soldering thingy. Sounds exciting.

[NeddySeagoon]

MrVahn,

Break the old parts off, then remove the wires that are still in the motherboard.
This operation needs intermediate soldering skills. Both wires from the capacitors are soldered into what is called a 'power plane' this is literally sheets of very thin copper, not the normal PCB 'trace'.
This is good electrically but bad for soldering, since the copper conducts the soldering heat away too.

NB: The capacitors are 'polarised' they must be fitted the right way round.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

[MrVahn]

Oh. This is really tricky indeed. Right now, running my system seems to be like being in a borrowed time. Once the heat rises up, my applications will randomly crash. I think I'll have to find one that is experienced with circuits to do the soldering job for me. This is something I can't fiddle with. I do not even have beginner soldering skills. This time I'll have to watch and learn.

[padoor]

i would suggest leaving the top open capacitors in its place and solder new capacitors on the other side of the board for fear of track burn off.
many layers of copper tracks are connected to the soldered points.
if the old capacitors are shorted only we have to remove it.
apparently no short circuit is there. soldering on the otherside should do the job. only connect correct polarity as neddy
said earlier. there is a 70 percent chance you succeed in this task than trying to break away old ones and then clearing holes. you really need hot air jet to that job. than a soldering iron.
take all prcautions of CMOS handling. hold the board only by its sides.
all the best luck
_________________
reach out a little bit more to catch it (DON'T BELIEVE the advocate part under my user name)

[NeddySeagoon]

padoor,

A few things ...
There is little or no space for new parts between the case and the bottom of the motherboard. Plug in cards may no longer fit.
You cannot afford the additional inductance and resistance that the longer componet wires on the back of the board will need.

These parts are all mounted on the 0v power plane on one side and either the 5v power plane or the Vcore power plane on the other. There is little or no chance of track damage but it is easy to damage the "plated through" holes that bring the connections to the surface both sides.
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

[MrVahn]

It's tough to find nice capacitors in my local electronics shop. I'll have to go to another city to buy the capacitors I need. I'll do this this weekend. Thanks for the tips guys.

As I have noticed, Neddy is very knowledgeable whether it be software or hardware. How did you manage to know such things? Just wondering.

[SnEptUne]

[MrVahn]

Just to inform you guys, I bought a brand new-second hand mobo. I find it hard to find the capacitors I need, and my trusty electrician friend went out of town so I opted to buy another one. Sad to say, no more brand new mobos for athlon xps are available. It is already being phased out here in our place. It works fine now. At leased for Debian.

[jorgepeixoto]

I imagine that replacing a mobo shouldn't be too expensive; this won't be a top-of-the-line mobo anyway. Perhaps a computer technician can even sell you a new mobo and accept your old one as part of the price of the new. A computer technician can use an old mobo to make tests, I imagine.

In any event, try to replace the mobo and see if the problem gets solved.

Also, the computer of a friend of mine freezed randomly. The problem went away when I underclocked the computer for him. You may try it.

[MrVahn]

I still encounter segfaults while compiling although normal usage is fine now. Maybe it's just the heat. I'll try to compile this weekend at nighttime when the temperature is lower.

[jorgepeixoto]

I think that temperature is not the only parameter. Even if the temperature is "OK", you can sometimes improve stability with underclocking.

The computer of a friend of mine would only be stable while underclocked. And in the BIOS Setup screen you could see that one of the monitored voltages had a wrong value if the computer had a normal clock, but an acceptable value if the computer was underclocked. So it seems that a problem in the PSU or the mobo caused a wrong voltage when the computer had a normal clock.

So, even if you think the temperature is acceptable, try underclocking your computer. I belive that in my friend's case the underclock was from 133 MHz to 112 MHz in the FSB, but I am telling this from memory and I have the memory of a grandma.

Also, while I run the risk of saying something obvious:

1-) Give your computer "breathing space". Leave it away from walls. If the air is trapped around your computer, it will be quite hotter

2-) In my computer, I removed the cover of the case to ease the ventilation. You can try it.

3-) put a ventilator right next to the computer, specially if you followed tip #2 and left it open. Even if you don't want to do this permanently (for the cost in eletric power, for example), you can see if the computer becomes stable and see if the problem is related to temperature.

4-) Yes, compiling at night seems to be a good idea, at least to test and see if the problem is temperature-related. You can combine this with tips #2 and #3. You can even make this permanent and create a habit of compiling when it is cold.

5-) Hey, at some point, maybe it is simply not worth trying to salvage bad hardware. Maybe you should take your pc to a technician, have him find the problem and replace the offending part, and save your time.

[MrVahn]

I found another problem in my memory modules. They are both DDR PC2700 but one of them is at 332 and the other 333. I removed the one at 332 because it offers lower RAM. Now the total RAM in my system is only 256MB. I have noticed this while I'm using Ubuntu (I have already stripped Debian out of my box) and my system hanged while using firefox in multiple tabs. The temp was pretty cool back then so I concluded it was the ram module with the lower number (what do you call that number again?).

I have already done your advice. I changed the layout of my room to allow more cooling for my CPU, removed the cover, and put an electric fan near it to help dissipate heat.

I'm going to try to install Gentoo via Ubuntu and see if it works.

[jorgepeixoto]

From Wikipedia, PC-2700 DDR-SDRAM operates at 166 MHz, and people call them 333 MHz because, since they transfer data twice per clock, they transfer data at 333 MHz.

So I suppose that when you say 333 you are referring to the frequency.

But AFAIK, there is no such a thing as a 332 MHz memory. Probably someone truncated 166.66... MHz to 166 MHz and, when multiplied by 2, this gives 332 MHz. Of course the correct way to do the multiplication is to multiply 166.66... MHz by two, giving 333.33... MHz and then rounding to 333 MHz.

So I believe that your two memory things can operate at 166 MHz (referred to as 333 MHz). However, there are other parameters, such as CAS. I believe you should set the parameters to conservative values, something that the two memory things can support. In any event, it would be a good idea to remove one of them and test. It seems you have already done this, so, keep your computer with one memory thing and test.

If you use lightweight apps such as Enlightenment (instead of GNOME) and claws-mail (instead of Evolution) then I believe you could even live permanently with 256 MB.

Good luck.

[NeddySeagoon]

MrVahn,

I don't believe your RAM timing is an issue, as jorgepeixoto says.
However, your different RAM sticks may need different amounts of time to become ready to read.

Some brain dead BIOSs, only read the set up information from the first RAM stick and use it for them all.
If thats the slow one, its OK. If its the fast one, the slow RAM is overclocked and can produce the symptoms you describe.

Try fitting your RAM sticks in different orders in the sockets.
NB: If you have a motherbord that supports, "Dual Bank" the RAM sticks in both halves of the dual bank must have identical timings.

If you can post the numbers from the tops of the RAM chips, we can probably look them up on the Web. Preserve the formatting and include the manufactueres name. (From the chips, not the stick)
_________________
Regards,

NeddySeagoon

Computer users fall into two groups:-
those that do backups
those that have never had a hard drive fail.

[MrVahn]

Ok. I have put back the RAM stick that I removed, but into a different slot. There are only 2 slots for RAM in my mobo. I just interchanged them.

Now for the numbers:

My 256MB ram:
pqi PQ4D328Y5
0525
27DA0757EG

pqi PQ4D328Y5
0525
M7DA0942EQ
Sticker:
DDR-333 256MB
0102-05B1
MDAC-328HA

Manufactured by pqi. Two of the chips has 27DA0757EG on it. The rest has M7DA0942EQ. There are about 8 chips on the memory stick.

My 128MB ram:
spectek AF 0031
S160016VMDTW
-64 *2BH2
*is written vertically
Sticker:
P16M44H6D-6A
128MB PC2700 DDR
0332 N 3 63 44 KTMM

There are only 4 chips on the stick. * only appears on the first chip.

[NeddySeagoon]

MrVahn,

Googling on

[MrVahn]

Thanks for that. I'll google it up too and see what I can find. As of now, I am testing whether interchanging the ram sticks will still make my system freeze. AFAIK I made my system freeze by opening up a lot of tabs in firefox, most of which are youtube tabs, with 256mb on the left slot and 128mb on the right slot ( slots are to the right of the cpu ). Now 128mb is on the left and 256mb is on the right. I hope it won't freeze this time.

[jorgepeixoto]

Hum, Firefox is not too much CPU intensive AFAIK, but it is memory hungry. So, if you freezed your PC by opening a lot of tabs, it seems to me that your problem is memory-related.

I would test with one only memory stick.

[MrVahn]

I did that already even before testing both sticks interchanged but it still froze. The stick that I used is the 256mb one. If my 256mb stick is bad, then I would be forced to buy a new stick. 128mb ram just won't fit my needs and I think the 128mb stick is the bad one because it is older.

As of now, I have a lot of programs open. Anjuta, rhythmbox, torrent, terminal, and firefox with a lot of tabs. No freezing as of now.

[MrVahn]

UPDATE: I have managed to install Gentoo via my Ubuntu install. I did it by opening up a terminal and following BobP's Stage 1/3 install. And guess what, it worked like a charm! I even had firefox open with several tabs of video streams. I haven't used an installation CD this time. I have had a single segmentation fault, but that was because of noontime. I'm nearly finished with my second emerge -e system. Before I changed motherboard it would not even finish compiling the toolchain (binutils gcc-config glibc gcc.)

[reidbold]

If problems come back, check the hard drive... I know everyone's thinking ram,cpu,mobo, but those things usually last 10 years as long as they don't get damaged (like overclocking). Hard drives are much more vulnerable, and can cause all sorts of weird errors all over the place.
_________________
-Reid