On 7/28/15, D. Hugh Redelmeier <hugh@mimosa.com> wrote:

| From: Russell Reiter <rreiter91@gmail.com>

Be sure of your diagnosis before applying drastic remedies. You claim overheating, and I previously responded on the assumption that this was a correct deduction. Now I'll question it: what observation convince you that overheating is a problem?

There were hard crashes with no hanging and no warning, just the power turned off in order to save the cpu. These crashes were mitigated by directing cold air from my AC directly into the unit till I did the bios update.

Also: I think that overclocking is really just a game. If it isn't fun, don't do it. The cost (in uncertainty etc.) is usually more than the benefit (subjectively unobservable speed-up). Don't do it and then complain about things going wrong! If I don't think it's worth your time to do overclocking, I'm sure it's not worth my time to help you do it.

The board default is to set overclocking as auto. I didn't choose it. I haven't found a table of recommended settings like I once got with service manuals when I bought a board. Now all I can find is quick start or user manuals.

I think that generally a computer needs to be close to twice as fast to feel significantly faster. Of course there are cases where crossing a threshold matters (eg. playing videos).

Well the combination of being able to choose a lower display resolution and whatever was changed after the bios update did show about a 10-15% reduction in CPU access when playing DVD's according to Plasma's graphical system monitor. I've bumped the resolution back up to 1024X768 for the time being. I'd don't watch a lot of DVD's, but I do use video to aid me in ad-hoc tweaking. I do note that desktop sprite animations are slower.

| <snip previous> | > existing solution. (AMD CPUs came with a disk of stuff instead of | > paste and it is supposed to be single-use: detaching and re-attaching | > a heat sink is not intended to work.) | | Could you describe this disk stuff in a little more detail? did you | get the package locally?

The disk came on the bundled heat-sink supplied by AMD. It was perhaps an inch in diameter and a little thicker than a sheet of paper. The idea is that it melts and spreads when it is sandwiched between the HS and the CPU (clamped) and the CPU heats up. So it spreads and becomes very thing. But it isn't supposed to do that twice.

It's not something that you'd buy by itself. It may or may not come with a Heat-Sink / Fan unit (HSF).

I haven't assembled a system in years. Enough decent off-the-shelf systems come along. So the AMD HSF units may come with different arrangements now. Here's a random thread about an AMD thermal pad: <http://hardforum.com/showthread.php?t=863613>

| I think I am going to change the heat sinc on this unit. The one that | came with it has this copper pipe threaded through it, Makes it look | sort of like the headers on the 67 Mustang I once owned. UNlike those | headers these ones look like all show and no go.

If it once was good enough, why would it not be now? Dust, deteriorated contact with CPU, ...?

I think perhaps I got it because it wasn't good enough, or not set up well enough or maybe both. I've seen a couple of threads about this board where significant overheating was an issue for some MS users. This machine formally had windows on it and a video card which had been removed. I remember having quite a time getting linux on to one particular laptop. It would overheat right off. I was able to finally install Debian after booting from a dos disk, formatting a fat partition and then installing the MS DOS power management utilities for the OS that was originally bundled with the unit. That's the one I had to sit on a bag of frozen peas for the unit to stay cool long enough to get anywhere. Now I have a couple of those reusable freezer blocks for the picnic cooler in case I have to do that again. Sorry can't help myself ...Gotta give peas a chance.

Does it feel hot when you think that the CPU is overheating? If not, there is a problem, I would think. After all, it is supposed to be carrying the heat to the air.

It seemed to be a little too hot around the base of the heat-sinc, while the fins of it were quite cool on the top and sides, which made me think perhaps there was too much thermal paste or perhaps none at all, or even poorly mated surfaces from the beginning.

| Things got better after I flashed the bios.

My guess is that the BIOS update would not change thermal management, but we will never know. (I rarely read much into the manufacturer's notes about BIOS updates because they disclose little and hide lots.)

| After I switched back to | GDM Plasma did stop hanging and I got to see that desktop for the | first time on this unit.

That's good.

| I did read that north bridge switching is not enabled in later | releases of this boards bios and I have since found there is an older | bios with this FSB feature available. Although I usually hesitate to | trim the north bridge clock. If you've got to trim a clock, trim the | big clock first and see what acts up first.

Don't pick at it! Use the BIOS defaults or conservative defaults (my machine doesn't crash so I rarely see the BIOS screen).

The defaults are overclocking enabled. I wish I could find a table of optimal settings from the manufacturer. My first conservative guess got me the BSOD.

| I wish I was better at math :-)

Me too. But I think you mean arithmetic. For which I use a calculator.

| And not so lazy. I reset the cmos without jumpering it off. When I | reset it I could set the date but not the time.

How can you be prevented from setting the time? Isn't date and time one setting?

It's the oddest thing. In the menu screen there is one line to set the date in the usual way. You know tab to the next setting and use the plus and minus keys to go up or down. However when you move to the next line to set hours, minutes and seconds, you can change them but enter won't commit them. F10 will write changes to the cmos but on restart the default time is always 00:00:00.

| My working theory is that the ready state of the cross fire video bus | consumes too many cycles. This board can drive four monitors. Since | it's only driving one monitor, my first thought on stabilising was to | knock the north bridge back a notch.

You need to apply Occum's Razor. Your explanation is way to unlikely and complicated for an early theory.

You don't actually have a video card, right? You are just using the on-motherboard video, right? That should not be stressful.

Right now I"m using the onboard card. I did try with a PCIeX card I picked up. It's a slightly newer vintage Radeon but that made no difference to the VGA output so I went back to the onboard card.

| There are Linux utilities from 2008 on the site; audio drivers and | stuff and there is a check script:

The fact that it talks about XFree86 means that it is way out of date. Everyone uses Xorg's X.

| The question for me right now is, can I just add XFree86 libs to get | quick time?

Don't do that! Or, if you do, stop asking for help. You are way off book.

| Oh yea and whether or not it's worth it to modify the CPU | heat-sinc.

First guess: not yet.

After the bios update there doesn't seem to be as much heat coming off it, but I can't tell for sure. The heat-sinc always feels cool to the touch on the top and sides but around the base is another matter.

| I know I haven't unseated it yet. That does't mean it hasn't been | removed and reseated before this.

True enough.

| I remember I use to be able to find | little tubes of acetone to clean parts. I guess a trip to Active | Surplus might find something similar, but I sure don't want to store | even a litre of acetone at my place, just to use a couple of | tablespoons of it now.

Acetone used to be available as nail-polish remover. Small amounts. In drug stores.

Thanks didn't think of that.

My very few run-ins with cleaning gunk off CPUs and heat sinks involved paper towels, finger nails, and elbow grease.

Apparently their are kits available with both solvent and thermal grease at various price points. I just recall Acetone, being so volatile, leaves the least amount of residue. On the link you posted on thermal pads these guy's were talking about lapping matching surfaces on ground glass, that's how I removed bits of rust from the bottom of a jack plane, although I did that on the ripping bed of a bandsaw instead of a glass plate. They also did not have too much nice to say about thermal pads, just to use razor thin applications of thermal grease. Thanks Russell

--- Talk Mailing List talk@gtalug.org http://gtalug.org/mailman/listinfo/talk