Hi there everybody.
:)

The system I am working on now, consists of an ASrock P4VM800 m/board, 512MB 3200 DDR RAM(1 stick), 80GB Seagate ATA100 7200RPM drive, CD-drive, Intel Celeron D CPU(socket 478) @ 2.4GHz, XP Pro SP1a.

Q1) Can you use 3200 and 2100 RAM on the same board? ASrock manual says you can, but does not specifically say you can have both at the same time, which is what I was trying to do. With a 256MB stick of 2100 DDR in DIMM2 socket, BIOS sees all the RAM, but when you try to do CPU intensive things, the system reboots without command. I have unticked the "Re-boot automatically" option in the settings. After about three auto(un-commanded) reboots in the middle of something, I removed the 2100 256MB RAM stick, and since then, things seem to work at speed without issue. MY THEROY is that the slower RAM stick will not be fast enough to work with the data stream from the system, and causes data corruption, forcing the system to reboot/hang... Yes? No?

Q2) What kind of PSU should you use with a Celeron D 2.4GHz? I'm using a 350W one at the moment... I assuem this is enough... I ask, because I generally only work with AMD, and don't know the new Intel's that well...

Q3) PSU voltages: On this setup, BIOS reports healthy POSITIVE voltages, but does not report on negative voltages. Once in XP, and with Everest Home Edition running, it reports that -12v is 2.03v POSITIVE...
:eek:
-5.00v is reported as 1.02v POSITIVE. CPUcore and 3.3v rails are just fine.
The system seems to run OK, but these figures don't seem right to me...

The PSU was brand new, as was the m/board, CPU and RAM.

Can anyone comment on any of my questions?


MM.

answer to question 1.---yes, you can have both ram at the same time, but the better ram is stuck at going the speed of the slower ram. the reason ur computer is restarting when you are doing cpu intensive things is because it is overheating and your computer is shutting itself down so it won't hurt itself.

question 2)---look on newegg.com, they have plenty of options. psu's work for intel and amd, generally.

I am thinking your problem lies in the slower than FSB ram mixed with ram that can run full the speed of the FSB. What I mean is that the Celeron D 2.4 runs on a FSB of 533MHz, which would = pc2700/333MHz. Now having pc2100/266MHz running is not a problem, or to run faster ram. But I am thinking that possibly the pc3200 is trying to run at the fastest speed it can with the Celeron's FSB of 533MHz =pc2700/333MHz, while the pc2100 cannot keep up. Giving you errors.

I am not real sure about this because usually the two sticks of ram would be running at the same speed, the lowest of the two sticks by default = pc2100. I would look in the bios to see what speed both sticks are running. If they are running at pc2700/333MHz then that would be the problem. But if they are running at pc2100/266MHz then I am way off base.

If they are running at pc2700, then see if there is an option to manually set the speed of the ram down to pc2100/266MHz, and see if that takes care of the problem.

Also there is the possibility that new ram is bad. Have you tried the new ram by itself? Run memtest yet? What is the error code when the machine reboots?

Again I may be way off base, but it is worth a look in the BIOS to double check the ram speed. Also sometimes certain ram don't like each other, or the mobo doesn't like certain ram.

I don't know which version of Everest you are running, but mine ver 1.51.195 does not give the -5V or -12V voltage values.
http://img58.imageshack.us/img58/5444/everest5dv.jpg

I agree with jlreich that it is most likely a RAM issue.

Q1) - Yes you can use both types. But the RAM speed of both sticks will be reduced to the slowest of the two. In this case PC2100. This is one of the features of improved SPD handling with DDR (and DDR2). This ensures that no RAM is running above its rated speed.

Your theory is incorrect. DDR RAM will never be set to operate faster than it is rated. Also remember that the FSB and the RAM bus speeds are separate from eachother. You can run them out of sync without any problems (other than optimization concerns). This is what the memory controller does - it keeps the FSB of the Celeron D at 533MHz, while keeping the RAM speed at whatever the RAM can handle (again, in this case it's PC2100 with both sticks). In the logical sense, the FSB and RAM bus never touch eachother. They are completely separated by the memory controller.

In this case, your errors were most likely caused by a compatibility problem or bad RAM. Do like jlreich suggested and run MemTest86 on each stick of RAM (separately) in each slot to see if that stick of PC2100 is having problems. (Don't forget to disable the CPU's L2 cache in the BIOS before running any MemTest86 test.)

Q2) - PSU requirements are not based solely on the CPU. It's a whole system kind of thing. A Celeron D system with 512MB of RAM, one optical drive, one hard drive, and onboard video needs only about 300W. A Celeron D system with 1GB of RAM, two optical drives, three SATA hard drives, and a Radeon X800 GTO needs about 400-450W. It depends on all your hardware in a single system.

Q3) - Programs running within Windows should only be used to give you ballpark figures of your voltages and temps. For the real numbers, go into the BIOS. The BIOS is always the most accurate. And in this case, Everest is giving you weird numbers anyway. Just ignore Everest for the negative voltage rails.

Thank you all, for your help.
I have run MemTest86.

It reported no problems, and completed all tests WITH THE ONE STICK OF 3200 RAM in place only (the 256MB taken out)

I will run the test with the 256MB in DIMM2(and 3200 ram out of DIMM1) later tonight.

SAPHALLINE: I could not disable the L2 cache. I looked right through all the BIOS settings, but there is no ability to disable it. It is there, as the summary says it has a Celeron D @ 2400MHz, and 256KB L2 cache. MemTest also sees the L2 cache.

BIOS is AMI - perhaps they don't let you turn this on and off???
There is a section on CPU configuration, but these are all to do with the ASrock overclocking feature(which is not being used - CPU is running at Intel spec of 2.4GHz).

Also, running temp of Celeron D is 54'C, with an ambient temp of around 25'C.
Too hot for an Intel?
I made VERY sure when I was building, that heatsink/fan assembly was properly clicked into position on the positioning/locking tabs, before rotating the clamping levers across, so the heatsink is sitting nice and level, and fully down in the base...

CPU and heatsink/fan came new in the box, so the phase-change material on the bottom of the heatsink was new, clean, and not touched before installation.

I am going to fit extra case-fans tonight, and there is a place for an exhaust fan right beside the CPU, so am going to fit one in there, but just curious.


MM.

running temp of Celeron D is 54'C, with an ambient temp of around 25'C.
Too hot for an Intel?Woah! :eek: Is that 54C at idle!? Holy $#^&!! No Celeron D should be running that hot. Is the ambient temp the temp outside the case or inside? Also, when you put that HSF on, did you check the thermal pad for any plastic film? Check the instructions that came with the CPU to see if there was one.

Do you currently have any case fans whatsoever? If not, that would be a problem. Even with a relatively cold CPU like a Celeron D, no modern system should be turned on without at least one case fan (besides the PSU fans).

I don't know if it was just me, but my Celeron 2.4 ran around that at first. Then after about ......maybe 2 months I went back into BIOS and it was running high 30's - low 40's idle.

I was actually getting ready to go out and get some Arctic Silver and reset it, but it seemed to be running right where it should be now, and has been since.

If that was just the break in period for the stock thermal pad it seems like along time to me. :confused:

Perhaps I should remove the heatsink, and double check?
I have more silver-greese that I could use, after I clean the old pad off.

Did not notice any film on the phase-change pad - it looked matt grey.
If there was a film on it, I would assume the slight gloss from the film, should have caught my eye...
The CPU heatsink feels quite warm/hot to the touch, so I assume there must be some kind of physical contact between the phase-change pad, and the heatsink, or the CPU would probably have cooked itself by now...

When I first powered up this new box, I went straight to the BIOS, and into HW monitor - I do this with all my new builds, to watch the CPU temp for a few minutes on idle. If it keeps going up and up and up, something is wrong, and you can catch it before it cooks itself part-way through the XP install!!!

There are no case fans at all.

I have added one at the front(80mm), and am going to add the extra exhaust one later tonight.(80mm also)

Also, the PSU is a twin-fan type, and the fan that is on the inside of the PC case, points towards the CPU.

Now, the thing is: This fan is sucking air into the PSU case, and the CPU fan is trying to suck air onto the CPU heatsink.

Could I have a little bit of a vacuum effect here in that two fans trying to suck air at the same place, cause a reduction in the ideal airflow???

This is my theroy, because the exhaust from the PSU is HOT, and the PSU case is hot to the touch.

I have turned the PSU around 180'(my case allows this, thankfully!), and I will see if that, along with the extra exhaust fan help cool things down.

CPU fan RPM is 2500RPM or so, so the BIOS says.
My other AMD XP-3000+ machine runs it's CPU fan at almost twice this - around 4700RPM.

It is an authentic Intel heatsink and fan, and the fan-blades on this CPU fan are at quite an acute angle. I assume this is to increase the airflow, but doing this should increase the air-resistance, so slowing the speed of the fan itself...

Yes? No?


MM.

I would say that a large part of the heat problem here is due to the fact that you had no case fans. You can't do that again, not with a modern system. Even a Celeron D (which is the coldest modern desktop CPU) needs at least one case fan, preferably two or more. The HSF unit sounds like it was properly installed, so I wouldn't worry about that just yet. But the case cooling scheme needs to be improved!

The overall airflow inside the case isn't disturbed to any noticeable degree by the PSU fan and the CPU fan. If they were facing toward eachother with opposite airflows, then maybe. But they are facing in different directions, so no worries there. Also think of it in terms of the rest of us - we all have the same thing going on in our systems with no ill effects.

Don't count the PSU fans as case fans! They serve entirely different roles! Any fan(s) in a PSU are there to cool the PSU. That's it! While the air coming out of your PSU (in the back) is warmer than any exhaust from a case fan, it shouldn't be hot! This means that your PSU fans are trying to cool the PSU using air that has already been heated by the rest of the system, which reduces the efficiency of the heatsinks inside your PSU, which is causing your PSU's efficiency and peak wattage to drop like a rock! This is bad! Case fans, on the other hand, were specifically designed to do nothing but cool the whole case, and thus the system. They are much more suited to the role of cooling what your PSU shouldn't have to. Having an intake fan is a good first step, but in order to keep your CPU and PSU healthy, add at least one more fan (exhaust first).

The stock HSF units are specifically designed to keep their respective CPU's from overheating. There are many factors that need to be taken into consideration, so you can't compare fan rotational speed alone. First of all, the AthlonXP's had slightly smaller fans, so immediately they need to turn faster to get the same CFM rating. Secondly, the AthlonXP 3000+ is a heat monster compared to a Celeron D @ 2.4GHz! Thirdly, heat dissipation efficiency and area need to be measured for the individual CPU's at idle, full load, and average load - which is then used to design the proper thermal pad and heatsink design/dimensions as well as the materials used. This is then formed into a reasonable shape, size, and weight for standard desktop systems to allow the proper cooling and surrounding airflow in both average and cramped environments. The fourth step is rigorous testing of the HSF units in various conditions and systems, even with multiple speeds of the CPU in question. These test systems & CPU's are measured for longevity and extreme cut-out for CPU survival rates. Once these designs are validated, the fifth step is to manufacture them and provide the appropriate thermal design whitepapers to all parties involved (OEM partners, mobo manufacturers, system designers, and consumers through the website). As you can see, there's no simple way to compare CPU HSF units.

Ok - thanks for all that info - great stuff!
:)

I have now fitted the exhaust fan, and took the oppurtunity to check that the CPU heatsink is sitting correctly - it is, and the thermal-pad is sandwiched nicely between the CPU heatsink, and the CPU itself, so there can be no problems there, I think(as you said, Saphalline...)

Once I fitted the exhaust fan(now I have an 80mm fan sucking air into the case at the front, and the exhaust fan beside the CPU, sucking hot air away from the CPU heatsink assembly, and exhausting it out the back of the case), things changed quite a bit!
:eek:

CPU temp as reported by BIOS at idle is now around 43'C with an ambient temp of approximately 28'C.(quite warm weather by NZ standards!!! :p )

This rises to around 53'C when the CPU is doing things(according to Everest, now that we are out of the BIOS), but when the CPU is finished it's tasks, the temp drops slowly back down to aroun 43/42'C.

The air from the PSU is cool/warm, not hot anymore.

The air from the CPU exhaust-fan vent is warm, getting a little warmer when the CPU is doing something, which I would expect.

Both the intake(front) and exhaust(rear) fans are running on 12v.

The air-flow from the exhaust is quite strong and steady.

If you lick your finger, then put it inside the case, you can feel the air flowing over it, so I think this has got to have improved things in a great many ways.

FROM THIS POINT ON, I will build all my systems with case-fans(both intake(cold) and exhaust(hot). I never bothered before, as I assumed that the CPU heatsink and fan could cope on their own.

One point to you guys, one penilty to me!
;)


MM.

Glad to hear things are running good now. :) You just can't say too much about proper cooling. ;) :)


One point to you guys, one penilty to me!
No penalty! Your problem is solved and we all get to learn from it. ;) And Saph gets all the points. :p

Oh good! Things look much better now!

As jlreich said, no penalties here! I just get worked up when I know there are hot CPU's out there. ;) Now that the temps are under control, I feel much better about your computer's situation. :D

So do I!!!
:p

Just for confirmation: I assume that around 50'C or so is OK, provided the CPU is working at 100%???

As I wrote - once the CPU is finished it's work, the temp drops back into the mid to low 40's...


MM.

That's right about where you want it. If your not going past 50C at full load, then your in good shape. ;)

Thanks - that's great!
:)

I'm happy enough(in my view, mind!) with a CPU running at 50'C or so when it's working. I've seen some systems running at 65'C and higher, and one machine running at 72'C(CPU temperature). Not surprisingly at this heat, the system was extremely unstable!!! The system was shoved under a desk, with no air-flow around the case, things piled on top of it, with a panel-heater(US read: radiator) behind the case, to keep their feet warm!!!


MM.

with a panel-heater(US read: radiator) behind the case, to keep their feet warm!!!
LOL! That's funny. :D That's like putting a plastic bag over your head and wondering why you can't breath. :p

Hehe. Heheheh. Funny, yet sad. :p

...we fixed that system, by replacing the CPU, and moving the location of the box to a cooler area. The box was moved first, and monitored for a few days, but although it had cooled to around 50'C odd when running, the machine was doing very odd things, even though it had cooled. My theory was that if the idle(or up to 45% CPU use) temp was 72'C, then when fully loaded, it must have almost been glowing!!! I thought it was very likely, that the CPU itself had been damaged via heat-stress, so we pulled it out, and plonked another(slightly faster) one in, and ever since then it has run with no problems - I only thought of it, when talking about my Celeron D in this thread...


MM.

Yep, most likely that CPU melted. Most CPU's start to suffer a melt-down at 70-75C. 50C or higher at idle automatically means a shortened CPU lifespan. As for the rest of your parts, the general rule of thumb is that for every 10C rise in temp, lifespan is reduced by 10%! :eek:

Gotta think about cooling. Heat is a killer. That's probably why I see so many Pentium and Pentium II machines still plugging away, but poorly cooled PIII and P4 systems die quickly. Modern CPU's can't take the heat, because they produce too much of it themselves.