Troubleshooting a P4 Prescott with 533Mhz ram

JCB75

Reputable
Dec 30, 2014
27
0
4,540
Wasn't sure where to put this topic. I feel the Hardware section needs a troubleshooting forum since issues could comprise any hardware.

I have a machine that's been running for almost ten years without many interruptions. Mostly it's been used for Office and web apps by day and as a nightly backup target for about 7 other machines besides itself.

Asus P5LD2-Deluxe
Intel Pentium 4 650 Prescott Socket 775 LGA 3.4 Ghz
2 x 1GB 533Mhz Simpletech PC2-4200


For all this time the BIOS' CPU, memory, and FSB settings were always left on Auto. Recently, the machine started showing the "Overclock voltage has failed", Press F1 to configure, or F2 to load defaults (can't remember exact wording), during POST. The power in the building is terrible. It's a warehouse in the middle of a cattle ranch in the middle of nowhere. All we can get for internet is via dish. The lights flicker several times a day. lol We have every machine on a consumer grade 750 surge/battery backup but they don't always catch the ridiculous hiccups of the power down there. The power situation is a techies nightmare. Sometimes I wake up in the middle of the night screaming over it. lol Recently, I have been getting random shutdowns, and blue screens. Most of the blue screens are page faults, or IRQ is less or equal. The errors are somewhat random without a pattern to it but they do seem to be memory related. Keep in mind this machine has been solid for almost ten years besides PSU replacements. I believe the power supply is good. I have tried it in another machine. I have everything unplugged from the mobo except the CPU, and RAM but it IS still in the case. I've cleared the bios numerous times. I have to get a lucky pass to get in the bios before it hangs but if I set the Overclock setting to Manual, the CPU frequency to 133Mhz, and the RAM frequency to 533Mhz it will run pretty solid. I've only had one lockup booted into a mini XP boot cd this way. RAM could also be Auto and it will run but anything over 133MHz on CPU hangs it eventually. I've stressed it with S&M successfully. I've tried about every other combo of setting to keep the 266Mhz CPU speed so the CPU will run at 3.4Ghz but it hangs very quickly unless I set it at 133Mhz. It also hangs less with just one stick of RAM, than two. If I remember correctly I can put the second one in after the CPU frequency has been set to 133Mhz without issues. It still hangs with two but not as quickly or as much. If the bios is set to Auto for CPU frequency, and RAM with two sticks it will hang before I can get in to change anything. I feel like I may be reducing heat by lowering the CPU freq and that is why it's running better. Because it used to run for so long and no longer does I feel like it's component failure. Possibly a mobo short but I am skeptical and don't want to remove everything just yet. My question is to those of you who know your overclocking stuff better than I. What does this behavior seem to indicate? Due to the age of this machine I wouldn't mind buying a used proc to try but if it's the mobo I think I'm going to start new. Before it failed it was a reliable backup target but now it's destiny, if any, is a lowly thick client. lol Are there things I can do to troubleshot the CPU, or mobo? I also wonder about memory module compatibility. These aren't on the list. They are just some I had laying around. I don't have any spares of any of this which is making it harder. Anything I want to do I'm going to have to convince someone it stands a chance of fixing it. Thought I might get a stronger impression from those with more troubleshooting experience.

Thanks!
 
Solution
Is there something I need to do to finalize this thread? I'm not sure any solution fits unless someone wants to post died of old age and I'll pick that. lol

JCB75

Reputable
Dec 30, 2014
27
0
4,540


Well, I thought it was until it just locked up with one stick while writing this. Up until then it was tolerating S&M for about two hours on the single stick.

 

JCB75

Reputable
Dec 30, 2014
27
0
4,540


No. lol It's fairly new and tested well in another machine. I wish. I seem to be replacing PSU's more than anything else. Probably has to do with our location and quality of power.
 

JCB75

Reputable
Dec 30, 2014
27
0
4,540
I feel that it's either the CPU, mobo, or possibly memory compatibility. But, when the machine started acting up it had 4 other 1GB memory modules in it which I had to use to get a machine going to replace this one. Changing the memory didn't help the issues so I doubt that it's the memory since the machine has the same issues on both sets.
 

popatim

Titan
Moderator
Have you changed the cmos battery, reset the bios to defaults, and cleaned & changed the cpu coolers thermal paste? If your southbridge has a spring loaded cooler on it you might want to replace its thermal paste also.

Also the 650 is 200mhz with a multiplier of 17 processor. Not 266.
 

JCB75

Reputable
Dec 30, 2014
27
0
4,540


CMOS batt: Replaced

Reset the bios by unplugging PSU and CMOS battery, and set bios clear jumper for one hour. That was the most extreme case. I have also cleared it in short order but at one point nothing would respond and letting it set this way brought it back.

CPU thermal paste changed, heatsink cleaned. I'm not very impressed by this CPU clip design. I pulled it off for a second look and noticed my new paste was barely touching the heatsink. I think I have it seated properly now. The trick is to push the metal frame down before pushing the plastic twist locks down. CPU temps begin from cold at 58C. They quickly go up to around 70C which to me seems hot but I've read this may be normal for this processor.

I haven't tried the southbridge. Will do.

I'm not completely versed in overclocking but I don't see a place to change the multiplier. I can set it to 200MHz for the CPU. I have and it locks. It also locks at 150MHz, but not at 133Mhz.
 

InvalidError

Titan
Moderator

Prematurely? Most of your components are the better part of a decade out of warranty. A decade is about the time where I stop feeling upset when stuff starts failing.

You can try bumping voltages a few notches if you want but it might not help you for long. Almost any cheap system you can put together today will be almost an order of magnitude better than your old P4, so I would not recommend spending any remotely significant money attempting to fix it - better off saving that money for a new CPU+MoBo+RAM.
 

JCB75

Reputable
Dec 30, 2014
27
0
4,540


lol True. I meant premature as in could one be pushed to fail before the other. Maybe they are too interrelated. Having to take care of around ten machines I run into this problem more than I like where I'm never sure what parts could be salvaged from a problematic system. I think diagnostic testing is a shortcoming in our field. Are there not reliable tools in the sub $300 range for CPU and mobo?
 

JCB75

Reputable
Dec 30, 2014
27
0
4,540


200MHz noted and tested. Also hangs.

Southbridge recoated with Artic Silver 5 thermal paste. Still hangs.

 

JCB75

Reputable
Dec 30, 2014
27
0
4,540
I don't think I will rely on this system much at this point for any critical tasks but part of this was to attempt to understand better which component had failed. Memory is easier to just try in a future system but I hate to hang on to a bad part or throw away a good one. When it comes to the CPU and mobo I often find myself scratching my head which was to blame.
 

InvalidError

Titan
Moderator

One simple diagnostic tool to start with is memtest86: if you get memory errors regardless of which DIMM you use and which slot you put it in, then you know you have a major issue with either the motherboard or CPU since it is highly unlikely that all four of your DIMMs and DIMM slots have similar issues.

As for trying your RAM in a future system, if you have DDR2, you can bury that idea since all CPUs launched in the last five-plus years require DDR3.
 

JCB75

Reputable
Dec 30, 2014
27
0
4,540
I wish I had other modules to try. I will keep the DDR2 because I have other dinosaurs grazing on the lan. I have tried these same modules in almost every possible slot configuration on this board. The more devices I plug in the more often I see it locking up. As for this one I think I'm ready to call it. I pulled it from the case and no better. It exhibits the same random unreliability as when it started all this. Let this beast pass into extinction! You've served me well P5LD2-Deluxe, and P4 Prescott. It's been a good long run. But your evolutionary cousin has made you obsolete and it's time to go... Think I might consider a thin client in this throw away and replace way of the future. Thanks all. :)
 

JCB75

Reputable
Dec 30, 2014
27
0
4,540


When I got it out of the case I did notice bulging caps. Not extreme, but all over the place.
 

JCB75

Reputable
Dec 30, 2014
27
0
4,540
Is there something I need to do to finalize this thread? I'm not sure any solution fits unless someone wants to post died of old age and I'll pick that. lol
 
Solution

InvalidError

Titan
Moderator

The "worst example" of failing cap would have bulged top with crusty "stuff" (dried electrolyte, usually brownish-orange, might look a bit like rust) leaking out of it.

In very extreme cases, the pressure relief vent (the stamped metal) may show clear signs of having been sundered by pressure when it did its job. The alternate extreme case is pressure relief failing to actually relieve pressure and the capacitor's shell shooting right off its rubbery/plastic-y base.
 

popatim

Titan
Moderator



I took care of that for ya! And I would consider bad caps as part of the aging process...