System Crashing with minimal use

Hi Rockster, thanks for advice. Okay I think I'll give it one more go tomorrow and re-apply the paste using your method. You mentioned twisting the HS atop the CPU was the worst thing I could do??? Arctic Silvers website instructions for the P4 chip advises this to be done??? I am confused. I take it you mean (the worst thing I can do) that it makes air gaps more probable???
As to the heat sink and fan, they are the original spec for the PC (HP pavilion t3060uk) - they are mated to the system so I see no reason to change them.
As for Startup, I have said earlier I am running on selective startup. The only items checked in the startup area of msconfig are ps2, KBD, hpsysdvr, jusched, Inicio, APVXDWIN, ISUSPM and lxbkmgr. So there is no great load on start up to the best of my knowledge.
Will let you know tomorrow how I get on after re-application of the paste. Thanks again.

 

Rockster, I wondered why NOrmsby mentioned twisting, so I looked at the instructions. They have added "large die" (Athlon 64 and P4) instructions at the bottom of the list.
They are pretty much "put a blob on and squash it ". Section 9 & 10.
http://www.arcticsilver.com/arctic_silver_instructions.htm

What might be happening NOrmsby is that the force of the heatsink clip/clamp is not enough to spread the compound far enough over the surfaces. If that is correct, then Rockster's suggestion of a new heatsink and fan combination would be my preference.

The suggestion of spreading the compound with a card may be possible if you were careful. They describe it for the smaller dies (the razor blade section). The compound I am using at the moment has a template of stiff plastic, the area cut out is slightly smaller than the surface area of the die/heat spreader. The plastic is probably a little thicker than a sheet of paper. They even supply a blank "credit card ". You put the compound on the die within the template area and keep passing the C/C across it until it is even.

If you try pushing down on the heatsink too hard you could crack the motherboard, so rule out shear force...and "twisting" does not seem feasible.

Think again about another heatsink/fan. I might imagine that to reduce costs, HP use special components that might only work with their manufacturing procedures.

Matt

 

Well said Matt. I was totally unaware of the "twist and push" addendum and consider it totally unnecessary if thermal paste is applied properly in the first place. Perhaps it is now felt that this may help spread out the paste but historically this has been considered as a potential contributor to "air pockets" between the die and the heatsink. Artic Silver 5 is quite "sticky" compared to Artic Silver 3 and its predecessors making proper application a bit more of art.

Ditto again on an aftermarket HSF - your stock HSF and its "identical replacement" is an inferior product compared to whats available at a relatively minor cost. In my book, a quality HSF is one of the most important upgrades that can be made to any system and the fact that your temps are showing such wide disparity indicates more than a simple software problem.

Back to the software issues - I would also strongly recommend that you find, download, install and use StartupCPL as opposed to a selective startup with msconfig. You will find this will give you considerably more control and goes well beyond the limited menu available in msconfig. Thats why it was recommended it in the first place.

 

Pete,

I 'twitched' a little bit after seeing that Core Voltage also. After making a quick trip to the Intel website... this ain't good. (see screenshot). This would definitely cause some extra heat.

B

 

Hi Guys and thanks again for your time and valuable info. Mattman, you are quite correct with the force issue. It is not something I would attempt given that there is a predetermined gap between the cpu and the hs and this is governed by the assembly which contains the hs and fan. The assembly screws on to the board and the predetermined spacing of the screws/jig does not let you go any further. What jumps out at me here is that if the original thermal conductor was a pad then it must have filled the pre-determined gap whereas the paste would have to be quite thickly applied to acheive this. Bear in mind also that the machine worked very well - impressive even - prior to this problem so I cannot relate the fault to the hs/fan, although I appreciate and accept what you guys are saying in relation to it's quality. It's something I may consider at a later date, especially if I intend to run high end graphics games for the kids, etc., but presently there's very little on this machine and is not near close to being overworked.
Rockster, thanks again for Srartup CPL. I now have this installed and did the Googles at Bleeping Computer. As a result my start up includes only the following (which I deem necessary for start up) : APVXDWIN, hpsysdrv, KBD, PS2, and SCANINICIO. I note that Pete has expressed some alarm with the core voltages, so I await your feedback on this. Thanks again, Neil.

 

Thanks Pete, been into the BIOS but cant find how to read the voltages in there. Dont want to upset anything until I know how, please advise how to do this thanks, Neil.

 

Thanks again Pete, had another look. Assuming you mean the BIOS SETUP UTILITY (pressing F1 during boot up) I dont have a Power monitor. In Hardware monitor it only gives me CPU temp and CPU/System fan speeds. I've looked at everything in there but no voltages. You must be bored sick of this/me by now!

 

NOrmsby,

Here's a free utility that will monitor your temperatures and voltages... SpeedFan. Please post the voltages/temps again to see if there are any differences. I starting to suspect the power supply. I had a HP computer w/ a P4 chip, the power supply blew and took the motherboard and the CPU with it. It was only about 18 months old. HP uses a rather small, cheap power supply.

B

 

Thanks Mr. B, ok got SpeedFan and here are the voltage readings :
+5v = 4.95v
Vcpu = 2.21v
Vcc = 3.30v
+5v = 5.00v
+12v = 11.86v
+3.3v = 3.38v
Vbat = 0.00v

This is a copy of the readout which shows "failures" ??? :

Win9x:NO 64Bit:NO GiveIO:YES SpeedFan:YES
I/O properly initialized
Linked ISA BUS at $0290
Linked Intel 82801FB ICH6 SMBUS at $0400
Scanning ISA BUS at $0290...
Scanning Intel SMBus at $0400...
SMBus msg : Failed reset at end $20 ($40)
Address $20 appears to be WRITE ONLY...
SMBus msg : Failed reset at end $2E ($40)
SMBus msg : Failed reset at end $2E ($40)
SMBus msg : Failed reset at end $2E ($40)
SMBus msg : Failed reset at end $2E ($40)
DME1737 (ID=$89) found on SMBus at $2E
Address $2F appears to be WRITE ONLY...
Address $30 appears to be WRITE ONLY...
Address $32 appears to be WRITE ONLY...
SMBus msg : Failed reset at end $3C ($40)
Address $3C appears to be WRITE ONLY...
SMBus msg : Failed reset at end $50 ($40)
SMART Enabled for drive 0
Found ST3200822AS (200.0GB)
End of detection
SMBus msg : Failed reset at end $2E ($40)
SMBus msg : Failed reset at end $2E ($40)
SMBus msg : 22:58:00.109 SELB $2C to $2E (4 retries)
SMBus msg : Failed reset at end $2E ($40)

Looks all double dutch to me but perhaps you may find something here which may help.

Also, interestingly, when I had SpeedFan opened I did a search and noticed that the cpu usage climbed up to around full tilt and then stabilised between 50-53% during the search. Again the fan came on full tilt.
Thanks again, Neil.

 

I did the roundabout of links at Intel. They have a "Processor Identification Utility" that has some tests:
http://support.intel.com/support/processors/tools/piu/
http://downloadfinder.intel.com/scr...roductID=483&OSFullName=All Operating Systems
http://support.intel.com/support/processors/tools/piu/sb/CS-015479.htm

A CPU utility:
http://www.cpuid.com/cpuz.php

The 2.2 volts could be an incorrect reading. Third party software could not be guaranteed to read accurately. HP machines do not have a "full" BIOS.

They talk about a thermal monitor, but I can't see what it does:
http://support.intel.com/support/processors/pentium4/sb/CS-007999.htm
It may increase the core voltage to try and reduce the power output and cool the processor if it gets overheated. Rockster2U might like to comment. I only know the basics about core voltage.

Running at 70*C plus is not going to work and it will be reducing the life of the processor (I don't like to say, but it may be damaged already and 2.2 V Core voltage may be a result). If the pressure method of applying the thermal compound does not work, you will need to use the credit card method and this will mean applying a layer that is really thin and really even.
I got my application technique down to a layer thinner than a sheet of paper. How much of a sculptor are you?

If it does not work in the end, I always know my learning experience has increased, it sounds like you are gaining some whatever the outcome

Matt
Edit: also check the air flow through the case and if it could be improved with addition fans. You may find it will stabilize the system in the 50-60*C area

 

Matt,

Asked Neil to try SpeedFan so the readings could be compared with his earlier Everest voltage check just in case there was a software error. The readings are almost identical (with the vCore voltage being too high). What do you think about the power supply possibly causing this? From a past experience... I think that is what caused my HP's P4 premature death.

B

 

Thanks Mattman. I tried all the Intel tests and all passed without any problems. Have tried several progammes to check the voltages and all come up with the same results. Now I know why I cant check it in the BIOS! thanks for that.
I think I'll give it one more go with the Arctic Silver tomorrow but I'm 99% convinced that by applying the thinnest of layers then the two surfaces will not be in total contact. If the previously installed PAD was roughly 1-2mm thick this would have taken up the space not being occupied by a thin layer of paste. Remember you cannot tighten the screws any further than allowed by the predetermined spacing on the jig/assembly and this was confirmed on my second try. When I removed the hs after applying just a small amount the spread pattern was minimal and covered approx 10% of the two faces. So obviously if I apply a finer layer the faces will not contact anywhere near 100%.
This is seeming more and more likely to be the root of the problem. And, yes the damage may already have been done but I have an inkling that this is not the only fault. If you have a look at my "crazy mad cooling fan" thread back in February, something must have caused this prior to any removal of the hs/fan.
Just another wee thing at the back of my head, in all the changes and re-applications of whatever thermal pastes I have used - the best results (although not correct) have been when a thicker amount has been used. Up until this most recent crashing problem I used Servisol as the agent between the faces (in a large quantity after a couple of tries with less and then a little more) and it seems to have worked a lot better than the Arctic Silver.

In fact, I'll wait for a bit until you guys maybe read this before doing the re-application - just in case I might have shed new light and it suddenly jumps out!
Thanks again, Neil.

 

Thats a good catch Pete and Mr B. Unfortunately, I too am an AMD guy. Regardless, if the defualt core voltage is 1.65 and the board is running 2.21, this is a big problem. There must be some kind of BIOS setting which is currently on auto that can be changed to manual giving you access to lower the core voltage. I'm not sure I would initially throttle all the way down to 1.65, but you might want to try something around 1.75 or 1.8 for starters. If you can run at 1.65, then thats where I'd try to settle down permanently. I'd also think that it would be sound advice to verify your current BIOS and check for revisions and see if there is a more recent BIOS that corrects this core voltage issue. It is possible that the CPU historically was a problem and someone pushed this voltage to 2.21 just to get the machine to run - I've seen similar but this is extreme if 1.65 is the correct default. Again, I'm an AMD guy and know most all cores well enough to recite in my sleep, not so with Intels.

Would also say that if your retention brackets don't permit good contact between the CPU and the HSF with a thin layer of paste, this would be further evidence that someone has tampered with it. Can you "shave" the retention bracket so the screws tighten the HSF down better?

For clarification, while I don't think this is the sole source of any problems you are having, its certainly something that needs to be corrected.

Good thing there are several people reviewing this thread Neil.



Edit: Ouch - just re-read Mr B's screenshot - 1.4V core - so 2.21 would be a 50% boost. Something is very wrong here - even worse than referenced above. Neil, lets reconfirm the CPU. Thanks.

 

Rockster,

The BIOS in a HP computer are pretty well 'crippled', I don't believe there is anyway to get at the voltage settings. I checked the HP site for a BIOS update for Neil's computer... there is one. There is not a wealth of information about what is updated (although it does mention "processor microcode to support newer Intel CPUs ")

Find the HP link HERE.

B

Also found the specs for the computer HERE.

 

If NOrmsby has added additional hardware or upgraded to more power hungry hardware I think it could. As you said HP machines have a power supply only really large enough to power the system when it is purchased.
Here is a power calculator:
http://journeysystems.com/power_supply_calculator.php
The core voltage is set through the BIOS. The main voltages actually look good, but that does not mean the PSU is not powerful enough to run all the hardware. Going back to basic hardware and checking the core voltage again would probably be a reasonable test.

That report from Speedfan seems to point to a problem reported by the SMBus ($2E). I'll see if I can find anything about it. Is there any way of finding out from the Speedfan website? Maybe I will have to go back to Intel BTW, I am an AMD person too.
Uh ha, I just Googled "DME1713 ", and it mentions the thermal monitor I linked to in my last post. I have not checked it, but seems a coincidence:
http://www.almico.com/forumvotes.php?id=4810737
I'll reread that information at Intel and see if I can make some connections. Edit: I may have quickly read thermal (temperature) and monitor from different parts of the Google summary. I will still check anyway.

I will check for other system diagnostics. Does anyone have any other suggestions for a system diagnostics? Mr B, does HP list any?

Matt