On Tuesday 05 February 2008 10:10, Per Jessen wrote:

Randall R Schulz wrote:

...

...

...

Maybe. The power supply could be defective. Can you measure the voltages while the test runs?

I did in fact have lm-sensors running to monitor the voltages, but I didn't keep a close eye on them. ...

I meant actually measuring the voltage with a multimeter.

Regardless, after Mike and David pointed me to the page in the Gigabyte manual that recommended at least a 400W powersupply, I've ordered a new one, which should be arriving tomorrow. I'll let you know how it works out.

Well, as I said originally, the first thing that struck me about your system description was the puny power supply. I'd say the likelihood that this will correct the problem is pretty good. I'd go for something with some head-room, though. 450W or 500W would be good.

thanks Per

Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

On Tuesday 05 February 2008 11:24, Per Jessen wrote:

Randall R Schulz wrote:

...

... I meant actually measuring the voltage with a multimeter.

I thought you might have meant just that, but I hadn't checked the motherboard for any specific test/measurement points. Are they normally marked/indicated? (for mere mortals to access with a probe I mean). Regardless, I'm not sure if a regular multimeter would have really helped catch subtle variations - maybe an oscilloscope. Have you tried it with a plain millivoltmeter?

The only thing an oscilloscope would disclose is excessive hum. I think any decent digital voltmeter with two and a half digits of precision would be sufficient. As to taking the measurements, the back side of the molex connectors usually allow a probe to contact the edge of the connector pin, and that's sufficient to get a reading.

...

Yeah, I did just that. Initially I was thinking 500-550W, but when I considered that the box would go unstable (under load) without major disk-activity and without the high-consumption graphics card, I decided to go a bit further and picked a 850W supply. Pricewise there wasn't much of a difference, and I didn't want to go for e.g. 500W only to see it too give up under full load.

That's a good idea if you expect to expand the system or re-purpose the cabinet for something heftier later. Offhand, I don't know how the efficiency of today's power supplies varies with the percentage of their rated output they're producing. In other words, does a higher-capacity PSU consume more mains power for a given level of consumption on its output side? If so, you wouldn't want to have such a large PSU, since that would be wasteful.

/Per Jessen, Zürich

Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Randall R Schulz wrote:

...

Offhand, I don't know how the efficiency of today's power supplies varies with the percentage of their rated output they're producing. In other words, does a higher-capacity PSU consume more mains power for a given level of consumption on its output side? If so, you wouldn't want to have such a large PSU, since that would be wasteful.

I haven't looked at switchmode technology for quite some time, but I would expect a fairly fixed efficiency in the 80-90% range, quite probably more. There will be a minimum consumption required in order to achieve optimal operation, but otherwise I think the output power will be e.g. 80% of the input power, and the rest is turned into heat. or to say it another way...(at 80% efficiency) a 150 watt supply or a 500 watt supply should either one pull the same current from the mains, if they alternately supply the same 80 watt load (ie MB)...approximately 100 watts from the mains in each example. It would NOT be wasteful to

On Thu, 2008-02-07 at 10:13 +0100, Per Jessen wrote: the environment...(but maybe to your wallet). Tom in NM -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

...

...

...

Offhand, I don't know how the efficiency of today's power supplies varies with the percentage of their rated output they're producing. In other words, does a higher-capacity PSU consume more mains power for a given level of consumption on its output side? If so, you wouldn't want to have such a large PSU, since that would be wasteful. I haven't looked at switchmode technology for quite some time, but I would expect a fairly fixed efficiency in the 80-90% range, quite

Randall R Schulz wrote: probably more. There will be a minimum consumption required in order to achieve optimal operation, but otherwise I think the output power will be e.g. 80% of the input power, and the rest is turned into heat. or to say it another way...(at 80% efficiency) a 150 watt supply or a 500 watt supply should either one pull the same current from the mains, if they alternately supply the same 80 watt load (ie MB)...approximately 100 watts from the mains in each example. It would NOT be wasteful to

On Thu, 2008-02-07 at 10:13 +0100, Per Jessen wrote: the environment...(but maybe to your wallet). You're forgetting there may be some additional power consumed by the

Tom Patton wrote: power supply itself. This might be a bit more for a larger supply. Even with no load, all power supplies draw some power. But many power supplies are only efficient when

James Knott wrote: they don't operate close to full load. kind regards Philippe -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Randall R Schulz wrote:

On Saturday 09 February 2008 10:08, Philippe Landau wrote:

...

But many power supplies are only efficient when they don't operate close to full load. This seemed counterintuitive to me, so I decided to do a little looking around. According to the graph on page 14 of "Power Supplies: A Hidden Opportunity for Energy Savings" http://standby.lbl.gov/CEC_Workshop/Docs/Ecos_FinalReport.pdf this is not the case. It very much depends on the power supply in question. Today i would recommend buying one with over 80% efficiency. These often reach up to 85% efficiency, but only when they operate around 50% load: Power Supplies 101: A comprehensive guide Motherboards.org: http://www.motherboards.org/articles/guides/1487_7.html

The study you quote may be a bit outdated (2002) or be valid for cheap PSUs (Compaq). It is certainly good to check the characteristics of the PSU before you buy one: http://www.80plus.org/manu/psu/manu_psu.htm But as seen earlier in this thread, calculating the power consumption needs of the components used is not always easy. Kind regards Philippe --

Switching power-supplies have an efficiency peak when producing about 50% of their maximum rated output. However the curve on the low side of the efficiency peak is much steeper than that on the high side.

So my suspicion was mostly correct. A PSU that has more than twice the capacity that it will be called on to supply will waste significantly more power than one being used in the 50% to 100% range.

...

kind regards Philippe

Randall Schulz

-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

On Saturday 09 February 2008 09:39:13 am Randall R Schulz wrote:

On Saturday 09 February 2008 11:22, Philippe Landau wrote:

...

Randall R Schulz wrote:

...

On Saturday 09 February 2008 10:08, Philippe Landau wrote:

...

But many power supplies are only efficient when they don't operate close to full load.

This seemed counterintuitive to me, so I decided to do a little looking around. According to the graph on page 14 of "Power Supplies: A Hidden Opportunity for Energy Savings" http://standby.lbl.gov/CEC_Workshop/Docs/Ecos_FinalReport.pdf this is not the case.

It very much depends on the power supply in question.

Of course.

But to go back to the original question (mine), the issue is whether highly over-rated supply is wasteful of power. It appears they are.

Randall Schulz

I disagree: I looked at a bunch of efficiency plots from 80plus.org. They generally peak at 50% of the rated load, at around 84-85%. their low load (20%) efficiency is around 82%, the high load (100%) efficiency is around 80%. so, for a 400 watt load, typical i would say for today's computers, an 800 watt power supply would draw around 470 watts, as it would operate at 85% efficiency. for the same 400 watt load a 400 watt power supply would draw 500 watts, efficiency at 80% for the same load, a 550 w power supply would draw around 488 watts, efficiencyaround 82% for the typical efficiency curve shown on this site, the same 400 watt load, supplied by a 2000 watt power supply , would draw 488 watts, as at 20% load the efficiency is at 82%. so, it is better to get a power supply at least twice as big as the load. Even when the rating is way greater, say 4 times the requirement, the efficiency is better than using a power supply at over 75% of it's capacity. the SUV syndrome fails when applied to power supplies:) d. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

On Saturday 09 February 2008 04:31:06 pm Randall R Schulz wrote:

On Saturday 09 February 2008 16:46, kanenas@hawaii.rr.com wrote:

...

On Saturday 09 February 2008 09:39:13 am Randall R Schulz wrote:

...

...

But to go back to the original question (mine), the issue is whether highly over-rated supply is wasteful of power. It appears they are.

Randall Schulz

I disagree:

I looked at a bunch of efficiency plots from 80plus.org.

Can you help me find such plots?

Thanks.

Randall Schulz

go to http://www.80plus.org/manu/psu/manu_psu.htm, as posted in this thread. then click on a particular brand and select a unit. d. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

kanenas@hawaii.rr.com wrote:

On Saturday 09 February 2008 04:31:06 pm Randall R Schulz wrote:

...

On Saturday 09 February 2008 16:46, kanenas@hawaii.rr.com wrote:

...

On Saturday 09 February 2008 09:39:13 am Randall R Schulz wrote:

...

...

But to go back to the original question (mine), the issue is whether highly over-rated supply is wasteful of power. It appears they are.

Randall Schulz I disagree:

I looked at a bunch of efficiency plots from 80plus.org.

That only applies to an 80Plus certified PSU. If you look at the PDF document which Randall linked to, you'll see that most PSUs have bad to increasingly horrible efficiency as load drops to 40% and lower. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

kanenas@hawaii.rr.com wrote:

On Saturday 09 February 2008 09:39:13 am Randall R Schulz wrote:

...

On Saturday 09 February 2008 11:22, Philippe Landau wrote:

...

Randall R Schulz wrote:

...

...

But many power supplies are only efficient when they don't operate close to full load. This seemed counterintuitive to me, so I decided to do a little looking around. According to the graph on page 14 of "Power Supplies: A Hidden Opportunity for Energy Savings" http://standby.lbl.gov/CEC_Workshop/Docs/Ecos_FinalReport.pdf

On Saturday 09 February 2008 10:08, Philippe Landau wrote: this is not the case. It very much depends on the power supply in question. Of course.

But to go back to the original question (mine), the issue is whether highly over-rated supply is wasteful of power. It appears they are.

Randall Schulz

I disagree:

I looked at a bunch of efficiency plots from 80plus.org. They generally peak at 50% of the rated load, at around 84-85%. their low load (20%) efficiency is around 82%, the high load (100%) efficiency is around 80%.

Only for power supples that are endorsed by 80plus.org. If you had bothered to look at the PDF that Randall presented, you would see that NON-80plus.org power supplies appear to have bad to dismal efficiency at loads below 40% rating. And as suggested, those curves are in fact typical.

so, for a 400 watt load, typical i would say for today's computers, an 800 watt power supply would draw around 470 watts, as it would operate at 85% efficiency. for the same 400 watt load a 400 watt power supply would draw 500 watts, efficiency at 80%

Until it burned up. Power supplies are generally rated on how much power they consume, not how much they supply. You're demonstrating that you really don't know enough about this subject to give qualified comments on it. Perhaps if you took some electrical engineering courses, specifically a course on electric power. Impedance matching is a very big issue, as well as Q-factors and other things which all have HUGE impacts on efficiency. To get the 80+% efficiency across the board is NOT a trivial thing. It takes more components, and higher-quality ones at that -- both of which cost money -- which is why manufacturers typically haven't made them UNTIL their was an organization promoting it so that the retail buyer would understand the added value of these more expensive supplies.

for the same load, a 550 w power supply would draw around 488 watts, efficiencyaround 82% for the typical efficiency curve shown on this site, the same 400 watt load, supplied by a 2000 watt power supply , would draw 488 watts, as at 20% load the efficiency is at 82%.

so, it is better to get a power supply at least twice as big as the load. Even when the rating is way greater, say 4 times the requirement, the efficiency is better than using a power supply at over 75% of it's capacity.

the SUV syndrome fails when applied to power supplies:) d.

You're really not qualified to advise others on this subject. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

* Per Jessen [02-12-08 15:13]: ...

In mid 1908s when I worked in electronics for s short while, a switchmode supply could achieve 80% efficiency without any major design effort.

Geeee. And to think, all this time I thought you were younger than me :^) -- Patrick Shanahan Plainfield, Indiana, USA HOG # US1244711 http://wahoo.no-ip.org Photo Album: http://wahoo.no-ip.org/gallery2 Registered Linux User #207535 @ http://counter.li.org -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Per Jessen pecked at the keyboard and wrote:

Patrick Shanahan wrote:

...

* Per Jessen [02-12-08 15:13]: ...

...

In mid 1908s

They had electronics in 1908? :-))

...

...

when I worked in electronics for s short while, a switchmode supply could achieve 80% efficiency without any major design effort. Geeee. And to think, all this time I thought you were younger than me :^)

-- Ken Schneider SuSe since Version 5.2, June 1998 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Aaron Kulkis wrote:

Per Jessen wrote:

...

Aaron Kulkis wrote:

...

Power supplies are generally rated on how much power they consume, not how much they supply.

But when there is a relatively direct connection between the two, what does it matter?

Yes. An 80% efficient 400W power supply is NOT rated to supply 400W of power, consuming 500W. It's rated to CONSUME 400W of power, supplying only 320 W.

So, yes, it does matter, because 400W of devices needs a 500W+ power supply.

I think I meant "does it matter" wrt the subject. /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Tom Patton wrote:

or to say it another way...(at 80% efficiency) a 150 watt supply or a 500 watt supply should either one pull the same current from the mains, if they alternately supply the same 80 watt load (ie MB)...approximately 100 watts from the mains in each example. It would NOT be wasteful to the environment...(but maybe to your wallet).

Thanks, that's exactly what I was trying to say. /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

On Tuesday 05 February 2008 17:46, Basil Chupin wrote:

...

Next time you get a new box to upgrade your system you will have a bit of spare capacity.

Won't that box have its own power supply?

Ciao

Randall Schulz -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Hello, On opensuse 10.3 I have a nfs share where I've copied the DVD into. Then I do a network installation from this nfs source. Everything goes well until the installation of the openoffice package. I get the message that openoffice package is broken, integrity check has failed. When I click on 'retry' it is installing the package without any complains. Also it happens from time to time, that it works fine. So I'd say 95 % ends up in an error and 5 % works fine. I've already resigned this package and downloaded a new (same) package from the internet and have put it into the repository but still same error message during the installation. Any ideas? Thanks Wolfgang ______________________________________________________________________ This email has been scanned by the MessageLabs Email Security System. For more information please visit http://www.messagelabs.com/email ______________________________________________________________________ -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Hello. Is there a built in command in linux to trace/dump internet browser interaction with servers. For example I am downloading a file by clicking a button on a web page. I do not see the link itself, though I want to. How can I dump this information? I am using firefox in suse 10.3. Thanks. SDA. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

On Sunday 03 February 2008 21:00, Per Jessen wrote:

David Bolt wrote:

http://europe.giga-byte.com/FileList/Manual/motherboard_manual_g a-ma790fx-ds5_e.pdf>

...

[1] That is the minimum the manual recommends, and it does say that using less will result in an "unstable or unbootable system."

Mike, David - thanks. I really had not seen that one. Note to self: RTFM.

When I first read the equipment list, and then saw that 350w ps, it was my first thought. But I went and checked the manual first. Some of the new equipment is very different from what I'm running. I've had power supplies go bad and replaced other parts not thinking that is what it was. Good Luck. Mike -- Powered by SuSE 10.0 Kernel 2.6.13 X86_64 KDE 3.4 Kmail 1.8 10:54pm up 172 days 3:27, 5 users, load average: 2.09, 2.49, 2.72 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Mike McMullin wrote:

Per, I'm curious as to why they sell such a low wattage power supply. The "lamest" that I could find locally was 465 Watts for a system that I just built, and it had a plug for the SATA drive that I put in.

The powersupply was in a cabinet I'd bought a while ago, but judging from what's on offer today, it's in the same range. 350W, maybe 400W in a typical office cabinet. Anything more than that comes with the usual "gamers" tag, which I've always found to be only about marketing. The range for individual power-supplies seems to be from 500W to 1000W. /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Basil Chupin wrote:

On 26 Oct last year I posted in a reply to Fred Miller a table of how to calculate how much wattage your PSU ought to be able to handle for the components in your system. The table I provided is:

QUOTE

As a guide, I have been using the following table to work out what sort of wattage "my" PSU should be able to handle:

Hi Basil, thanks, that was an interesting table. I have my doubts about the 7W/128Mb for the memory - to get rid of 8x7W=56W, a 1Gb DIMM would need a sizeable heatsink. Also, according to the datasheet for the Kingston DDR2-800 1Gb DIMM, its power- consumption is only 1.872W. /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

On Monday 04 February 2008 12:18, Basil Chupin wrote:

Looks like improvements have been made to RAM to make them more energy efficient, which is a very good thing.

And also; it is probably "all" retgarding thememory. Control circutry etc included. Speaking from experience as a field tech i'd sat anything more a 2 disk system requires above 400W. Especially if you have a powerful graphics card. Also, i have found (maybe not so scientifically) that AMD tends to be more "hungry"... The new Core2 Duo is actually lower in consumption than its predecessor Dual Core. As previously stated, try a bigger PSU. Id go for 500 or 650 W --          /Rikard ----------------------------------------------------------------------------- email   : rikard.j@rikjoh.com web     : http://www.rikjoh.com mob: : +46 (0)763 19 76 25 ------------------------ Public PGP fingerprint ---------------------------- < 15 28 DF 78 67 98 B2 16 1F D3 FD C5 59 D4 B6 78  46 1C EE 56 >

Per Jessen pecked at the keyboard and wrote:

Per Jessen wrote:

...

Last week I put together a new system:

Gigabyte GA-MA790FX-DS5 (BIOS F3 which includes the AMD patch). AMD Phenom 2.2GHz (quad-core) 4 x 1Gb Kingston KVR800D2N5/1G (on Gigabytes approved list). ATI Radeon X1650 graphics. 2 x Hitachi SATA disks. The usual stuff - DVD, CDRW, floppy.

[snip]

...

Well, this system is highly unstable under high load. If I start 4 x mprime torture tests, it will automatically reboot in 30-40minutes.

To you who've been offering advice on this matter, here's the interim status:

As suggested, I upgraded the powersupply from the meagre 350W to 850W. Almost certainly overkill, but I didn't really want to play around with this for a long time, so I decided go for safe. Very nice powersupply too - Toughpower from Thermaltake. Well, unfortunately it didn't do much good - the system remains unstable at high load. In the meantime I have also raised the issue with Gigabyte, who have in fact been quite responsive. They sent me a beta-BIOS version to try out, which really got my hopes up. Didn't change anything though - system remains unstable at high load.

I feel pretty certain that this is a Gigabyte issue, hopefully they'll come up with an answer. After all, the problem is 100% reproducable.

Watch this space ...

/Per Jessen, Zürich PS: as a side-effect the new powersupply is _very_ quiet, which is _very_ pleasant.

Have you tried switching/swapping the memory modules? -- Ken Schneider SuSe since Version 5.2, June 1998 -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Per Jessen wrote:

Per Jessen wrote:

...

Last week I put together a new system:

Gigabyte GA-MA790FX-DS5 (BIOS F3 which includes the AMD patch). AMD Phenom 2.2GHz (quad-core) 4 x 1Gb Kingston KVR800D2N5/1G (on Gigabytes approved list). ATI Radeon X1650 graphics. 2 x Hitachi SATA disks. The usual stuff - DVD, CDRW, floppy.

[snip]

...

Well, this system is highly unstable under high load. If I start 4 x mprime torture tests, it will automatically reboot in 30-40minutes.

To you who've been offering advice on this matter, here's the interim status:

As suggested, I upgraded the powersupply from the meagre 350W to 850W. Almost certainly overkill, but I didn't really want to play around with this for a long time, so I decided go for safe. Very nice powersupply too - Toughpower from Thermaltake. Well, unfortunately it didn't do much good - the system remains unstable at high load. In the meantime I have also raised the issue with Gigabyte, who have in fact been quite responsive. They sent me a beta-BIOS version to try out, which really got my hopes up. Didn't change anything though - system remains unstable at high load.

I feel pretty certain that this is a Gigabyte issue, hopefully they'll come up with an answer. After all, the problem is 100% reproducable.

The next place I would look is thermal instability. What sort of temperatures are your CPU and North-bridge running at? Try better heat sinks, or fluid-cooling. Improving air circulation in the case never hurts either. (and a couple large diameter fans running at low speed are much quieter than small diameter fans running at high speed) -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Per Jessen wrote:

Still, lm-sensors report the following:

temp1 temp2 temp3

idle: 43-44 40-42 82

1 x mprime 44 46 84 2 x mprime 45 52 84 3 x mprime 47 58 85 4 x mprime 47 60 85

I did a 2nd try and managed to get the system running a little longer: temp1 temp2 temp3 4 x mprime 47 63 86 /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Aaron Kulkis wrote:

...

I did a 2nd try and managed to get the system running a little longer:

temp1 temp2 temp3 4 x mprime 47 63 86

Temp3 seems *very* high.

86 C is right on the edge of temperature tolerance for most consumer-grade semiconductor devices.

Yep, I agree.

Find out what temp3 corresponds to, and start figuring out how to cool that device better.

I have no idea how to work out what temp3 corresponds to. The lm-sensors docus aren't very inspiring on that subject. Interestingly, the BIOS "Health Status" only reports two temperatures - a system temperature and a CPU temperature. About 5min after power-on, these are hovering around 39 and 42 respectively. (room temp is 21C). /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Per Jessen wrote:

Per Jessen wrote:

...

Last week I put together a new system:

Gigabyte GA-MA790FX-DS5 (BIOS F3 which includes the AMD patch). AMD Phenom 2.2GHz (quad-core) 4 x 1Gb Kingston KVR800D2N5/1G (on Gigabytes approved list). ATI Radeon X1650 graphics. 2 x Hitachi SATA disks. The usual stuff - DVD, CDRW, floppy.

[snip]

...

Well, this system is highly unstable under high load. If I start 4 x mprime torture tests, it will automatically reboot in 30-40minutes.

To you who've been offering advice on this matter, here's the interim status:

As suggested, I upgraded the powersupply from the meagre 350W to 850W. Almost certainly overkill, but I didn't really want to play around with this for a long time, so I decided go for safe. Very nice powersupply too - Toughpower from Thermaltake. Well, unfortunately it didn't do much good - the system remains unstable at high load. In the meantime I have also raised the issue with Gigabyte, who have in fact been quite responsive. They sent me a beta-BIOS version to try out, which really got my hopes up. Didn't change anything though - system remains unstable at high load.

I feel pretty certain that this is a Gigabyte issue, hopefully they'll come up with an answer. After all, the problem is 100% reproducable.

Watch this space ...

/Per Jessen, Zürich PS: as a side-effect the new powersupply is _very_ quiet, which is _very_ pleasant.

Per, I will be watching because as I already posted I see the "exact" same instability problems with my: Gigabyte GA-N700SL 2 Gig of Ram Raid1: 2 Seagate (ST3250410AS_6RY28WW1) SATA 250G drives in Raid 1 ATA: Maxtor_5T020H2_T2H34J2 as swap I mean the *exact* same problems. Please do not forget to post any response you get from them. It also may be a kernel issue. Take a look at another bug I have open regarding this instability. The fact that it has spamassassin involved may be irrelevant. A Gigabyte,kernel issue may be the root cause: https://bugzilla.novell.com/show_bug.cgi?id=353252 Keep the list apprised, I'm interested -- David C. Rankin, J.D., P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Per Jessen wrote:

David C. Rankin wrote:

...

I will be watching because as I already posted I see the "exact" same instability problems with my:

Gigabyte GA-N700SL 2 Gig of Ram Raid1: 2 Seagate (ST3250410AS_6RY28WW1) SATA 250G drives in Raid 1 ATA: Maxtor_5T020H2_T2H34J2 as swap

I mean the *exact* same problems. Please do not forget to post any response you get from them. It also may be a kernel issue.

The last response from Gigabyte suggested I try Windows and verify if the stability also existed there. They also suggested Linux may not have the right support for the new chipset.

I replied that the only Windows license I have available is for Win2000, which is also unlikely to have support for a 2007 chipset :-) I think it's highly likely that I'll be returning this board in the next few days.

...

Take a look at another bug I have open regarding this instability. The fact that it has spamassassin involved may be irrelevant. A Gigabyte,kernel issue may be the root cause:

https://bugzilla.novell.com/show_bug.cgi?id=353252

Hmm, there's not much to go on to actually suggest a kernel problem?

/Per Jessen, Zürich

Per, I don't think it is a "new" chipset problem. The GA-N700SL I have is a 2005 board. Take a look at the log snippet below and let me know if you are seeing anything similar. I also experienced another meltdown Sunday morning and I have updated bugzilla (https://bugzilla.novell.com/show_bug.cgi?id=353252) with the following: <quote> Guys, This may be a pure kernel/Gigabyte issue as opposed to a sa-learn only issue. The sa-learn problem looks like it may the the result of a larger issue concerning kernel stability/instability on the gigabyte platform. I have experienced several additional meltdowns. Some the logs are blank. Some the logs contain useful information. The latest, yesterday involved the simple routine daily mandb process. The following was collected: Feb 10 04:30:09 bonza kernel: BUG: unable to handle kernel NULL pointer dereference at virtual address 00000044 Feb 10 04:30:09 bonza kernel: printing eip: Feb 10 04:30:09 bonza kernel: c02c6157 Feb 10 04:30:09 bonza kernel: *pde = 00000000 Feb 10 04:30:09 bonza kernel: Oops: 0002 [#2] Feb 10 04:30:09 bonza kernel: SMP Feb 10 04:30:09 bonza kernel: last sysfs file: /devices/platform/i2c-adapter/i2c-9191/9191-0290/name Feb 10 04:30:09 bonza kernel: Modules linked in: af_packet it87 hwmon_vid hwmon i2c_isa iptable_filter ip_tables ip6_tables x_tables apparmor loop 8139cp 8139too mii button sr_mod cdrom shpchp i2c_nforce2 rtc_cmos pci_hotplug rtc_core nvidia_agp rtc_lib i2c_core agpgart sg dm_mirror dm_log sd_mod ehci_hcd ohci_hcd usbcore dm_snapshot dm_mod edd ext3 mbcache jbd fan sata_nv pata_amd libata scsi_mod thermal processor Feb 10 04:30:09 bonza kernel: CPU: 0 Feb 10 04:30:09 bonza kernel: EIP: 0060:[<c02c6157>] Tainted: G N VLI Feb 10 04:30:09 bonza kernel: EFLAGS: 00010286 (2.6.22.16-0.2-default #1) Feb 10 04:30:09 bonza kernel: EIP is at _spin_lock+0x1/0xf Feb 10 04:30:09 bonza kernel: eax: 00000044 ebx: d4cdf978 ecx: d4cdf888 edx: 00000001 Feb 10 04:30:09 bonza kernel: esi: 00000000 edi: 00000039 ebp: d4cdf930 esp: d56dfd84 Feb 10 04:30:09 bonza kernel: ds: 007b es: 007b fs: 00d8 gs: 0033 ss: 0068 Feb 10 04:30:09 bonza kernel: Process mandb (pid: 3975, ti=d56de000 task=f7f79570 task.ti=d56de000) Feb 10 04:30:09 bonza kernel: Stack: c018eee3 00000282 00000246 00000000 dfafba40 dfb77e00 d4cdf978 d4cdf888 Feb 10 04:30:09 bonza kernel: 00000000 00000039 00000039 c0181a80 00000080 d4cdfc68 d4ccbe94 00027fd8 Feb 10 04:30:09 bonza kernel: dfffeaa0 0000008e 000280d2 c015bdfc 00003c90 00001e48 00000000 013fec00 Feb 10 04:30:09 bonza kernel: Call Trace: Feb 10 04:30:09 bonza kernel: [<c018eee3>] remove_inode_buffers+0x3b/0xc8 Feb 10 04:30:09 bonza kernel: [<c0181a80>] shrink_icache_memory+0xc5/0x1b6 Feb 10 04:30:09 bonza kernel: [<c015bdfc>] shrink_slab+0xde/0x13b Feb 10 04:30:09 bonza kernel: [<c015c6e4>] try_to_free_pages+0x173/0x247 Feb 10 04:30:09 bonza kernel: [<c015833e>] __alloc_pages+0x1a1/0x2d6 Feb 10 04:30:09 bonza kernel: [<c0163392>] find_mergeable_anon_vma+0x5e/0xb2 Feb 10 04:30:09 bonza kernel: [<c0160646>] __handle_mm_fault+0x482/0xa83 Feb 10 04:30:09 bonza kernel: [<c0170c32>] do_sync_read+0x0/0x10a Feb 10 04:30:09 bonza kernel: [<c01632a8>] arch_get_unmapped_area_topdown+0x9b/0x127 Feb 10 04:30:09 bonza kernel: [<c0162c52>] vma_merge+0x168/0x178 Feb 10 04:30:09 bonza kernel: [<c02c78fa>] do_page_fault+0x26c/0x5a6 Feb 10 04:30:09 bonza kernel: [<c02c768e>] do_page_fault+0x0/0x5a6 Feb 10 04:30:09 bonza kernel: [<c02c6492>] error_code+0x72/0x80 Feb 10 04:30:09 bonza kernel: ======================= Feb 10 04:30:09 bonza kernel: Code: 05 90 ff 00 30 d2 89 d0 c3 89 c2 90 81 28 00 00 00 01 0f 94 c0 84 c0 b9 01 00 00 00 75 09 90 81 02 00 00 00 01 30 c9 89 c8 c3 90 <fe> 08 79 09 f3 90 80 38 00 7e f9 eb f2 c3 90 81 28 00 00 00 01 Feb 10 04:30:09 bonza kernel: EIP: [<c02c6157>] _spin_lock+0x1/0xf SS:ESP 0068:d56dfd84 Feb 10 04:30:09 bonza su: (to nobody) root on none Feb 10 04:30:35 bonza syslog-ng[2748]: last message repeated 3 times Let me know what else I can provide. I am not the only user affected by instability on Gigabyte boards. Per Jessen on the openSuSE list is likewise seeing the exact same issues. I am more than happy to provide whatever you need to see if we can solve this. This has happened regardless of the kernel version installed. Currently we have: 11:35 bonza~> uname -r 2.6.22.16-0.2-default </quote> This is just strange and a huge pain when the server just sporadically just goes whacko. I'll post any additional information I can. Since this has occurred with both mandriva and opensuse, on different power supplies and with different ram and hardrives, it is pretty much limited to being a gigabyte motherboard problem. -- David C. Rankin, J.D., P.E. Rankin Law Firm, PLLC 510 Ochiltree Street Nacogdoches, Texas 75961 Telephone: (936) 715-9333 Facsimile: (936) 715-9339 www.rankinlawfirm.com -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

David C. Rankin wrote:

Per Jessen wrote:

...

David C. Rankin wrote:

...

I will be watching because as I already posted I see the "exact" same instability problems with my:

Gigabyte GA-N700SL 2 Gig of Ram Raid1: 2 Seagate (ST3250410AS_6RY28WW1) SATA 250G drives in Raid 1 ATA: Maxtor_5T020H2_T2H34J2 as swap

I mean the *exact* same problems. Please do not forget to post any response you get from them. It also may be a kernel issue.

[pruned]

Feb 10 04:30:09 bonza kernel: EIP: [<c02c6157>] _spin_lock+0x1/0xf SS:ESP 0068:d56dfd84 Feb 10 04:30:09 bonza su: (to nobody) root on none Feb 10 04:30:35 bonza syslog-ng[2748]: last message repeated 3 times

Let me know what else I can provide. I am not the only user affected by instability on Gigabyte boards. Per Jessen on the openSuSE list is likewise seeing the exact same issues. I am more than happy to provide whatever you need to see if we can solve this. This has happened regardless of the kernel version installed. Currently we have:

11:35 bonza~> uname -r 2.6.22.16-0.2-default

</quote>

This is just strange and a huge pain when the server just sporadically just goes whacko. I'll post any additional information I can. Since this has occurred with both mandriva and opensuse, on different power supplies and with different ram and hardrives, it is pretty much limited to being a gigabyte motherboard problem.

Have you installed the latest (if any, ie) BIOS upgrade for this mobo? Ciao. -- If you want to know what a man is like, take a look at how he treats his inferiors not his equals. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

David C. Rankin wrote:

Basil Chupin wrote:

...

David C. Rankin wrote:

...

Per Jessen wrote:

...

David C. Rankin wrote:

...

I will be watching because as I already posted I see the "exact" same instability problems with my:

Gigabyte GA-N700SL 2 Gig of Ram Raid1: 2 Seagate (ST3250410AS_6RY28WW1) SATA 250G drives in Raid 1 ATA: Maxtor_5T020H2_T2H34J2 as swap

I mean the *exact* same problems. Please do not forget to post any response you get from them. It also may be a kernel issue.

[pruned]

...

Feb 10 04:30:09 bonza kernel: EIP: [<c02c6157>] _spin_lock+0x1/0xf SS:ESP 0068:d56dfd84 Feb 10 04:30:09 bonza su: (to nobody) root on none Feb 10 04:30:35 bonza syslog-ng[2748]: last message repeated 3 times

Let me know what else I can provide. I am not the only user affected by instability on Gigabyte boards. Per Jessen on the openSuSE list is likewise seeing the exact same issues. I am more than happy to provide whatever you need to see if we can solve this. This has happened regardless of the kernel version installed. Currently we have:

11:35 bonza~> uname -r 2.6.22.16-0.2-default

</quote>

This is just strange and a huge pain when the server just sporadically just goes whacko. I'll post any additional information I can. Since this has occurred with both mandriva and opensuse, on different power supplies and with different ram and hardrives, it is pretty much limited to being a gigabyte motherboard problem.

Have you installed the latest (if any, ie) BIOS upgrade for this mobo?

Ciao.

Basil,

There are several bios updates out there, but the descriptions associated with them don't seem to relate to any issue I have. I guess it couldn't hurt to try. If it all goes south, I'm only out a flaky gigabyte motherboard anyway.

It is more than possible that the description of the fixes in the description of the BIOS upgrade(s) only contain/s what Gigabyte consider to be what most people want to see. The description of the fix given is, as you know, quite brief - to fit into the allocated space. You won't have the mobo going 'south' by installing the latest BIOS upgrade, as you know :-) , but you may very well get rid of the problem you have been experiencing. Ciao. -- If you want to know what a man is like, take a look at how he treats his inferiors not his equals. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Per Jessen wrote:

I'm quite tempted to move the system outside tomorrow - in the early morning we're a couple of degrees below - with a huge external fan, I should be able to keep the whole thing _very_ cold. Mind you, I will be disappointed if it then starts to work. The amount of cooling I've got should be enough.

OK, yesterday morning I moved the box outside, hooked it up with network and power. Outdoor temperature was -1C . Temperature rose to about 6 in the afternoon, but the machine actually ran pretty stable throughout the day. I had one lockup which needed a hard reset, but I haven't been able to reproduce that. Whilst running 4 x mprime, the temperatures as seen by sensors stabilised at this: work System CPU temp3 Outdoor 4 mprime 39 51 81 5 4 mprime 35 50 80 6 So it's all about cooling. The interesting thing is - the machine uses the AMD supplied CPU heatsink and fan. It's got additional fans and heatsinks to improve the air/heat flow. How do I improve on the cooling without looking at water-cooling and such? Today I've removed the left cover and set up a large fan (similar to the one shown in the picture on the right hand side here: http://en.wikipedia.org/wiki/Fan_%28mechanical%29) work System CPU temp3 Room 1 mprime 35 46 80 21 2 mprime 35 51 81 21 3 mprime 36 56 81 21 4 mprime 38 60 81 21 Well - the system remains unstable and automatically reboots after a while, maybe 10-15mins, of running 4 x mprime. I've already got an RMA# so I think this board is going back after all. /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Mike wrote:

The first thing I always do is get rid of the AMD supplied fans. I've used Thermaltake type heatsinks and fans. I know that Coolermaster and Thermaltake make pretty good equipment for CPU cooling.

How big is your case? Midtower, full tower?

Midtower, but fairly uncluttered. I think the airflow in the box is ok. /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Mike wrote:

On Thursday 14 February 2008 15:54, Per Jessen wrote:

...

OK, yesterday morning I moved the box outside, hooked it up with network and power. Outdoor temperature was -1C .

Temperature rose to about 6 in the afternoon, but the machine actually ran pretty stable throughout the day. I had one lockup which needed a hard reset, but I haven't been able to reproduce that.

Whilst running 4 x mprime, the temperatures as seen by sensors stabilised at this:

work System CPU temp3 Outdoor 4 mprime 39 51 81 5 4 mprime 35 50 80 6

So it's all about cooling. The interesting thing is - the machine uses the AMD supplied CPU heatsink and fan. It's got additional fans and heatsinks to improve the air/heat flow. How do I improve on the cooling without looking at water-cooling and such?

The first thing I always do is get rid of the AMD supplied fans. I've used Thermaltake type heatsinks and fans. I know that Coolermaster and Thermaltake make pretty good equipment for CPU cooling.

How big is your case? Midtower, full tower?

...

Well - the system remains unstable and automatically reboots after a while, maybe 10-15mins, of running 4 x mprime. I've already got an RMA# so I think this board is going back after all.

Could be the problem.

Also remember.. Fans at the bottom of the case blow inwards. Fans at the top of the case blow outwards. In a quiescent state, CFM in = CFM out [CFM = Cubic Feet/Minute = (ft^3)/minute, essentially the same as m^3/minute except for a dimensionless conversion factor of about 30. or, to make it more clear CFM in = CFM through = CFM out. This is an application of Thevenin's Current Rule and is practically a mathematical identity. This is an identity. if CFM in > CFM out, the case would eventually explode; conversely, if CFM in < CFM out, then eventually the case would be completely evacuated). What does this mean? If you have 1 fan blowing in rated at 300 CFM, and 2 case fans and a PSU fan blowing out, each rated at 300 CFM, then initially, the out fans will blow out 600 CFM more than the in fans, until the air pressure inside the case has dropped enough that... the in-fan is bring in a little more than 300 CFM...and the out-fans are blowing out a bit more than 300 CFM total (a little over 100 CFM each, 1/3 of their rating!). Additional in-flow might be provided by leaks or ventilation holes in the case. In any event in the above scenario, the air pressure in the case is low, which means that any fans and heat sinks used inside the case for CPU cooling, etc are operating in an environment in which there is less air density than the thing is designed for, and will therefore not cool as well as if there was more air inside the case. If the situation is reversed (3 fans blowing in, 1 PSU fan blowing out), we still get to the same quiescent state of the flow-through being slightly higher than the flow rate provided by one fan. Initially, the in-fans blow 900 CFM in, while the out-fan blows out only 300 CFM...but as internal air pressure RISES, the input fans flow rapidly drops to about 100 CFM each (again), and the out-fan now runs slightly above rated capacity (a bit over 300 CFM). But now...the air WITHIN the case is under increased pressure. The CPU fan is blowing MORE air molecules over the heat sink...therefore IMPROVING the cooling ability of the heat-sink/fan apparatus. In this second example, using case perforations for more air flow is more acceptable than in the first example... In both examples, the air flow through performations only happens due to a difference in air pressure inside the case vs. outside the case, but in the second example, the perforation air-flow is due to higher air pressure (and thus, higher air density inside the case), while in the first example, the perforation air flow is due to a vacuum condition inside the case, which is detrimental to cooling the CPU heat sink -- vacuum conditions within the case should always be resolved by increasing the amount of air forced into the case. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Per Jessen wrote:

Per Jessen wrote:

...

I'm quite tempted to move the system outside tomorrow - in the early morning we're a couple of degrees below - with a huge external fan, I should be able to keep the whole thing _very_ cold. Mind you, I will be disappointed if it then starts to work. The amount of cooling I've got should be enough.

OK, yesterday morning I moved the box outside, hooked it up with network and power. Outdoor temperature was -1C .

Temperature rose to about 6 in the afternoon, but the machine actually ran pretty stable throughout the day. I had one lockup which needed a hard reset, but I haven't been able to reproduce that.

Whilst running 4 x mprime, the temperatures as seen by sensors stabilised at this:

work System CPU temp3 Outdoor 4 mprime 39 51 81 5 4 mprime 35 50 80 6

So it's all about cooling. The interesting thing is - the machine uses the AMD supplied CPU heatsink and fan. It's got additional fans and heatsinks to improve the air/heat flow. How do I improve on the cooling without looking at water-cooling and such?

Today I've removed the left cover and set up a large fan (similar to the one shown in the picture on the right hand side here: http://en.wikipedia.org/wiki/Fan_%28mechanical%29)

work System CPU temp3 Room 1 mprime 35 46 80 21 2 mprime 35 51 81 21 3 mprime 36 56 81 21 4 mprime 38 60 81 21

Well - the system remains unstable and automatically reboots after a while, maybe 10-15mins, of running 4 x mprime. I've already got an RMA# so I think this board is going back after all.

/Per Jessen, Zürich

Can you still get peltier devices, they used to use them with early P1s and 486s. Their main use nowadays is in 12v cool boxes. They only have a 50% cooling efficiency so 100W of cooling takes 200W of power but they don't need all the pipes of a water cooled system. Also a possibility that your vcore is too high, that causes excessive heat. Dave -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Per Jessen wrote:

Dave Plater wrote:

...

...

Today I've removed the left cover and set up a large fan (similar to the one shown in the picture on the right hand side here: http://en.wikipedia.org/wiki/Fan_%28mechanical%29)

work System CPU temp3 Room 1 mprime 35 46 80 21 2 mprime 35 51 81 21 3 mprime 36 56 81 21 4 mprime 38 60 81 21

Well - the system remains unstable and automatically reboots after a while, maybe 10-15mins, of running 4 x mprime. I've already got an RMA# so I think this board is going back after all.

Can you still get peltier devices, they used to use them with early P1s and 486s. Their main use nowadays is in 12v cool boxes.

Yeah, I remember those. Interesting idea, but I think they're impractical with high power dissipation (the AMD Phenom needs to get rid of about 100W).

/Per Jessen, Zürich

It seems the cpus are outpacing heatsinks 60c is a very unstable temperature, think about it, you almost have 4 cpus in one chip carrier, except for the interface with the outside world. All this in a package smaller than a P4. The heatsink supplied is most probably made for a single or dual core anyway. Heat is the enemy of electronic reliability. Dave -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Per Jessen wrote:

Dave Plater wrote:

...

It seems the cpus are outpacing heatsinks 60c is a very unstable temperature, think about it, you almost have 4 cpus in one chip carrier, except for the interface with the outside world. All this in a package smaller than a P4. The heatsink supplied is most probably made for a single or dual core anyway. Heat is the enemy of electronic reliability. Dave

I've tried to find some recommendations on the CPU-temperature, but I'm not having much luck.

If you go to the AMD site you will find the specs for the various CPUs they make and the (max) temperature at which they are designed to operate.

The warning levels in the BIOS are 60/70/80.

These warning levels are ones that *YOU* set, manually, at which you either get a warning signal or the systems shuts down.

Also, I would have thought the heatsink supplied by AMD for the Phenom would be sufficient for 4 cores - but maybe not.

Nope. They provide the minimum required but it is really up to you to ensure that the correct heatsink is installed to suit your local conditions. Cooling/ heatsinks/thermal pastes is a science of its own. Go to Overclockers.com (and similar) to find out all about heatsinks/temperatures, lapping the heatsink, using the right compound, and so on :-) . It's a new world! :-) . Ciao. -- If you want to know what a man is like, take a look at how he treats his inferiors not his equals. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Philippe Landau wrote:

Basil Chupin wrote:

...

Per Jessen wrote:

...

Also, I would have thought the heatsink supplied by AMD for the Phenom would be sufficient for 4 cores - but maybe not.

Nope. They provide the minimum required but it is really up to you to ensure that the correct heatsink is installed to suit your local conditions.

Cooling/ heatsinks/thermal pastes is a science of its own. Go to Overclockers.com (and similar) to find out all about heatsinks/temperatures, lapping the heatsink, using the right compound, and so on :-) . It's a new world! :-) .

Are you saying standard thermal paste can cause problems ?

Depends on what you mean by "standard paste". There is that white gunk which was used many moons ago when the cpus generated very little heat but now you have some pretty nifty stuff which contains silver particles (Artic something, which is what I last used) so that the heat is transferred better from the cpu to the heatsink. But there is more to it than simply the 'paste'. The basic purposed of a heatsink is to dissipate the heat generated by the cpu (for each watt there is 1 degree C of heat). To be able to be dissipate that heat as efficiently as possible there needs to be perfect contact between the cpu and the heatsink - which is why there is an art involved in lapping the bottom of the heatsink to a mirror finish and to also make sure that the surface is FLAT so as to provide complete contact with all of the (tiny) surface of the cpu. Now, it is not always possible to obtain a completely smooth, flat surface on the heatsink so a THIN, VERY thin, layer of thermal paste is applied to fill in the microscopic imperfections of the surfaces. Put on a thick layer of the paste and you defeat the purpose because too much paste only hinders the transfer of heat. Someone in an earlier posting stated that even a speck of dust may interfere with the dissipation of heat thru the heatsink - and this is correct because there won't be the optimum contact between cpu and heatsink.

Are the matching CPU heatsinks by say AMD a risk ?

Cannot comment on this but I can say that all my cpus have been/are AMD and I have replaced the heatsinks which came with them with my own (as in ones that I bought). There are at least 2 sites (one of them I already mentioned, the Overclockers) which give the efficiency of various heatsinks as well as give you instructions on how to go about lapping a heatsink's surface. To add to the 'science' of heatsinks, there are explanations about the various materials used in making heatsinks and how efficiency is affected by these materials. For example, you can get a full copper heatsink and also an aluminium one but with a copper base attached at the bottom - and such an attachment adds another dimension to the efficiency of heat transfer because the 2 surfaces (heatsink and attachment) may each need lapping and an application of a dab of paste. If you are really interested in this subject go to some site like Overclockers and read the material there; the articles explain much better than I can about what could be done to get the optimum heat transfer from cpu to heatsink. Ciao. -- If you want to know what a man is like, take a look at how he treats his inferiors not his equals. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Basil Chupin wrote:

...

I've tried to find some recommendations on the CPU-temperature, but I'm not having much luck.

If you go to the AMD site you will find the specs for the various CPUs they make and the (max) temperature at which they are designed to operate.

Yep, you're right, I found it. The Phenom is listed at max 70C operating temperature. So assuming the measurements are correct, the CPU has been well within the range when it's been running at 50-60C.

...

Also, I would have thought the heatsink supplied by AMD for the Phenom would be sufficient for 4 cores - but maybe not.

Nope. They provide the minimum required but it is really up to you to ensure that the correct heatsink is installed to suit your local conditions.

Well, unless my "local conditions" are very, very different from "normal conditions", I'd expect the supplied heatsink+fan to provide adequate cooling. The machine is a desktop, operating in a desktop/miditower cabinet and at room temperature. Besides, the CPU isn't overheating. Anyway, the board is going back now - I'm wondering if Gigabyte knows that this board might be a little flaky. Otherwise I don't think they would have agreed to a return so quickly. Watch this space. /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Per Jessen wrote:

Basil Chupin wrote:

<snip>

...

...

Also, I would have thought the heatsink supplied by AMD for the Phenom would be sufficient for 4 cores - but maybe not.

Nope. They provide the minimum required but it is really up to you to ensure that the correct heatsink is installed to suit your local conditions.

Well, unless my "local conditions" are very, very different from "normal conditions", I'd expect the supplied heatsink+fan to provide adequate cooling. The machine is a desktop, operating in a desktop/miditower cabinet and at room temperature. Besides, the CPU isn't overheating.

I'm thinking that right now, AMD considers their primary market for the Phenom to be server rooms, with nice big air-conditioning units which create lots of air flow throughout the entire room.

Anyway, the board is going back now - I'm wondering if Gigabyte knows that this board might be a little flaky. Otherwise I don't think they would have agreed to a return so quickly.

That's possible. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Per Jessen wrote:

Basil Chupin wrote:

...

...

I've tried to find some recommendations on the CPU-temperature, but I'm not having much luck.

If you go to the AMD site you will find the specs for the various CPUs they make and the (max) temperature at which they are designed to operate.

Yep, you're right, I found it. The Phenom is listed at max 70C operating temperature. So assuming the measurements are correct, the CPU has been well within the range when it's been running at 50-60C.

...

...

Also, I would have thought the heatsink supplied by AMD for the Phenom would be sufficient for 4 cores - but maybe not.

Nope. They provide the minimum required but it is really up to you to ensure that the correct heatsink is installed to suit your local conditions.

Well, unless my "local conditions" are very, very different from "normal conditions", I'd expect the supplied heatsink+fan to provide adequate cooling. The machine is a desktop, operating in a desktop/miditower cabinet and at room temperature. Besides, the CPU isn't overheating.

My AMD cpu is shown as 'accepting' 85C. However, I have the BIOS set to issue an alarm and switch the system off if the temperature gets to 60C. It *had* reached ~55C so I replaced the AMD provided heatsink and fan. So now, on a very hot day, I get under this 55C reading. I also have an incoming and outgoing 12cm fans (controlled by the psu re RPM) in the Antec box as well as another incoming fan on the side blowing directly on to the cpu. The PSU is an Antec Truepower 550W which comes with both incoming and outgoing fans controlled by the temperature of the cpu.

Anyway, the board is going back now - I'm wondering if Gigabyte knows that this board might be a little flaky. Otherwise I don't think they would have agreed to a return so quickly.

Watch this space.

Watching with interest. In the meantime have a look at this URL which Carlos provided (in another thread)- http://www.heatsink-guide.com/ Ciao. -- If you want to know what a man is like, take a look at how he treats his inferiors not his equals. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Per Jessen wrote:

Dave Plater wrote:

...

It seems the cpus are outpacing heatsinks 60c is a very unstable temperature, think about it, you almost have 4 cpus in one chip carrier, except for the interface with the outside world. All this in a package smaller than a P4. The heatsink supplied is most probably made for a single or dual core anyway. Heat is the enemy of electronic reliability. Dave

I've tried to find some recommendations on the CPU-temperature, but I'm not having much luck. The warning levels in the BIOS are 60/70/80.

Also, I would have thought the heatsink supplied by AMD for the Phenom would be sufficient for 4 cores - but maybe not.

You would think, but I'm not wild about the fan-speeds neeed on the manufacturer-supplied heatsink with its dinky little fan. I started using after-market fans several years ago for both better cooling and lower (quieter) fan-speeds that result from using larger-diameter fans.

/Per Jessen, Zürich

-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Aaron Kulkis wrote:

Per Jessen wrote:

...

Dave Plater wrote:

...

It seems the cpus are outpacing heatsinks 60c is a very unstable temperature, think about it, you almost have 4 cpus in one chip carrier, except for the interface with the outside world. All this in a package smaller than a P4. The heatsink supplied is most probably made for a single or dual core anyway. Heat is the enemy of electronic reliability. Dave

I've tried to find some recommendations on the CPU-temperature, but I'm not having much luck. The warning levels in the BIOS are 60/70/80. Also, I would have thought the heatsink supplied by AMD for the Phenom would be sufficient for 4 cores - but maybe not.

You would think, but I'm not wild about the fan-speeds neeed on the manufacturer-supplied heatsink with its dinky little fan.

I started using after-market fans several years ago for both better cooling and lower (quieter) fan-speeds that result from using larger-diameter fans.

The bigger diameter of the fan the bigger diameter is the spindle, the bigger is the 'dead spot' where no air is being blown -- and this is also the where the heat-dissipating part of the cpu under the heatsink is located. Ciao. -- I was very heavily into pornography. Then my pornograph broke. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Rajko M. wrote:

instead of blowing user ear drums. Second, I haven't seen CPU cooling solution with bigger fan and no funnel to keep that air flowing to CPU, not somewhere in that direction.

There are some novel solutions about: http://www.pcp.ch/ASUS-Silent-Square-EVO-1a12328822.htm http://www.pcp.ch/Cooler-Master-Hyper-TX2-1a12299842.htm http://www.pcp.ch/Scythe-Andy-Samuray-Master-1a12296300.htm http://www.pcp.ch/Thermaltake-BigTyp-120-1a12174381.htm And plenty more of the same kind. /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Rajko M. wrote:

On Sunday 17 February 2008 07:53:35 am Per Jessen wrote:

...

Rajko M. wrote:

...

instead of blowing user ear drums. Second, I haven't seen CPU cooling solution with bigger fan and no funnel to keep that air flowing to CPU, not somewhere in that direction.

There are some novel solutions about:

http://www.pcp.ch/ASUS-Silent-Square-EVO-1a12328822.htm http://www.pcp.ch/Cooler-Master-Hyper-TX2-1a12299842.htm http://www.pcp.ch/Scythe-Andy-Samuray-Master-1a12296300.htm http://www.pcp.ch/Thermaltake-BigTyp-120-1a12174381.htm

And plenty more of the same kind.

I had in mind simple, cheap, solutions.

That all depends of course, but if the system is e.g. 1000, and the necessary CPU-fan is 64, it doesn't seem too bad.

Above solution is way better, but I haven't seen retailers using them.

Well, they're probably way overkill unless you're into overclocking and all that.

The most often used is a big hole on side panell with funnel right above CPU fan.

Yep, I've got that too. Works quite well. /Per Jessen, Zürich -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Basil Chupin wrote:

Aaron Kulkis wrote:

...

Per Jessen wrote:

...

Dave Plater wrote:

...

It seems the cpus are outpacing heatsinks 60c is a very unstable temperature, think about it, you almost have 4 cpus in one chip carrier, except for the interface with the outside world. All this in a package smaller than a P4. The heatsink supplied is most probably made for a single or dual core anyway. Heat is the enemy of electronic reliability. Dave

I've tried to find some recommendations on the CPU-temperature, but I'm not having much luck. The warning levels in the BIOS are 60/70/80. Also, I would have thought the heatsink supplied by AMD for the Phenom would be sufficient for 4 cores - but maybe not.

You would think, but I'm not wild about the fan-speeds neeed on the manufacturer-supplied heatsink with its dinky little fan.

I started using after-market fans several years ago for both better cooling and lower (quieter) fan-speeds that result from using larger-diameter fans.

The bigger diameter of the fan the bigger diameter is the spindle, the bigger is the 'dead spot' where no air is being blown -- and this is also the where the heat-dissipating part of the cpu under the heatsink is located.

And these heat-sinks using the larger sized fans are designed to take that into account. The "dead zone" doesn't extend forever; it is almost or entirely gone one "spindle diameter" past the fan blades. -- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org

Dave Plater wrote:

Per Jessen wrote:

...

Per Jessen wrote:

...

I'm quite tempted to move the system outside tomorrow - in the early morning we're a couple of degrees below - with a huge external fan, I should be able to keep the whole thing _very_ cold. Mind you, I will be disappointed if it then starts to work. The amount of cooling I've got should be enough.

OK, yesterday morning I moved the box outside, hooked it up with network and power. Outdoor temperature was -1C .

Temperature rose to about 6 in the afternoon, but the machine actually ran pretty stable throughout the day. I had one lockup which needed a hard reset, but I haven't been able to reproduce that.

Whilst running 4 x mprime, the temperatures as seen by sensors stabilised at this:

work System CPU temp3 Outdoor 4 mprime 39 51 81 5 4 mprime 35 50 80 6

So it's all about cooling. The interesting thing is - the machine uses the AMD supplied CPU heatsink and fan. It's got additional fans and heatsinks to improve the air/heat flow. How do I improve on the cooling without looking at water-cooling and such?

Today I've removed the left cover and set up a large fan (similar to the one shown in the picture on the right hand side here: http://en.wikipedia.org/wiki/Fan_%28mechanical%29)

work System CPU temp3 Room 1 mprime 35 46 80 21 2 mprime 35 51 81 21 3 mprime 36 56 81 21 4 mprime 38 60 81 21

Well - the system remains unstable and automatically reboots after a while, maybe 10-15mins, of running 4 x mprime. I've already got an RMA# so I think this board is going back after all.

/Per Jessen, Zürich

Can you still get peltier devices,

Or better yet -- DON'T! Peltiers are only good in situations where you want to use 2W of ADDITIONAL electricity consumption for every 1W used. To cool his 95W Phenom would require an ADDITIONAL 180W of power consumption by the Peltier cooler. There's much more efficient ways of disippating 95W of heat....You can get decent water-cooling with only 10-20W if you're willing to use an fanless external tower radiator.

they used to use them with early P1s and 486s. Their main use nowadays is in 12v cool boxes. They only have a 50% cooling efficiency so 100W of cooling takes 200W of power but they don't need all the pipes of a water cooled system.

And where would this peltier device be dumping that 300W of heat?... Oh that's right ... the ambient air right around the CPU and all the other heat-stressed components like the north bridge chip, and disk drives. Good idea -- NOT!

Also a possibility that your vcore is too high, that causes excessive heat.

Very true. and Overclocking is just begging for trouble AND significantly shortened CPU life.

Dave

-- To unsubscribe, e-mail: opensuse+unsubscribe@opensuse.org For additional commands, e-mail: opensuse+help@opensuse.org