To avoid dust and overheating I had mounted my MB right on the wall above the table. It is now easy to maintain and to install-remove components. No RFI since I use only internet-radio. The only headache is PSU - it is closed from the production.
Cooling? Consider all the effort in fans for getting heat out of a case. Why keep it in the case in the first place? Take the cover or side panel off and leave it off. I have been doing that for 25 years and never had a problem. This way there is no "inside" temperature of the case. It all convects away.
If that still bothers you set up a small fan to blow on the open motherboard. I mean like a cheap six inch fan from the local drug store. I have never found a need for it but if it makes you feel better it will guarantee the air temp around the parts is always the same as room temperature. You can't get any better than that.
My SUN M20 Manual says not to take away the side panel for more than 10 minutes. My front panel is a mesh like a radiator in an old Lancia car. The engineering design of this WS is superior to that of the average PC. It has been running 24/7 since January 2008. I only added RAM, a second disk and a RW/DVD. But I am thinking of upgrading its CPU with a faster chip I have already bought from a relative, and so I was wondering about a heatsink. No AC.
Tullio
To repeat, 25 years and not a problem, to emphasize, not a single problem.
SUN certainly has well designed hardware. How can it be better than no case and a room fan? There is nothing better than room temperature. That is the way it is.
Cooling? Consider all the effort in fans for getting heat out of a case. Why keep it in the case in the first place? Take the cover or side panel off and leave it off. I have been doing that for 25 years and never had a problem. This way there is no "inside" temperature of the case. It all convects away.
If that still bothers you set up a small fan to blow on the open motherboard. I mean like a cheap six inch fan from the local drug store. I have never found a need for it but if it makes you feel better it will guarantee the air temp around the parts is always the same as room temperature. You can't get any better than that.
What do you do to prevent things like dust-build-up in your rig? Or do you clean it very regularly? And do you have any pets that shed hair, and themselves :) , all around the house?
I'm saying there could be other reasons to leave the side panel on.
Excuse but are their dust filters on the intakes of your case? What intakes do you have? Do you clean those filters regularly?
As you have neither intakes nor filters on them why would it be any different with or without the side panel? What dust build up have you found inside your machine?
My SUN M20 Manual says not to take away the side panel for more than 10 minutes. My front panel is a mesh like a radiator in an old Lancia car. The engineering design of this WS is superior to that of the average PC. It has been running 24/7 since January 2008. I only added RAM, a second disk and a RW/DVD. But I am thinking of upgrading its CPU with a faster chip I have already bought from a relative, and so I was wondering about a heatsink. No AC.
Tullio
IF it has a windchill-design, you are right about not taking off the panels!
And the main reason for keeping an ordinary ATX case on is to minimise RFI from getting into and escaping out of the case. Thats why there are blanking plastes over the slots at the back and the CD slots at the front. Sometimes the cables to the frontpanel buttons and LEDs have ferrite filters too.
I have a portable radio beside one of my machines and if the lid's removed I can instantly tell what type of BOINC task is running by the type of nasty buzzing on top of the local radio station.
Bingo! The real reason. RFI. However if you hear a buzz that is 60Hz and no case in this reality it going to shield it.
What you will hear is elevated noise cutting the S/N of the radio signal which is what the case does shield.
Back when Limbaugh was entertaining, pre-pills, I ran a shielded, grounded cable out the window and a long wire outside. FM is balanced. Use an old style twin-lead antenna with the crossbar of the T outside the window.
Better yet, as you have a computer, use internet radio.
I just built a system with an I7-2600K Z68 chipset, GTX560, 16GB RAM and had some real temperature issues with the CPU. Running E@H at 100% had the core temps 94-96C with the alarms set at 98C. No OC, it's running 3.5GHz.
I replaced the stock cooler with a CoolerMaster 212+ and they are now running 50-56C. The only issue is the cooler is huge about 6" tall so it takes a big case. I'm using a ThermalTake V series full tower with lots of fans.
I'm sure that in a significant percentage of cases (no pun intended), just re-appying the thermal interface grease stuff would improve cooling. E.g. last time I transplanted a CPU between boards, I could see that the thermal grease was really covering the whole surface of the heat spreader (instead of just closing the little scratches in it), actually insulating the CPU and the heat sink. It's amazing how careless this stuff is applied in off the shelf PCs.
It's also amazing how much of the goop that the CPU manufacturers put on the die when installing the heat spreader. I've popped the top on a number of old processors that were past any financial value, and both brands put quite a lot on before putting the spreader on. When I put thermal compound on before installing heatsinks, I take the edge of one of the package lids the CPU came in and smooth out and scrape off any extra. My i5-2400 went from 81 degrees C on the hottest core with the stock fan, heatsink, and thermal compound down to 52 with XIGMATEK Gaia SD1283 and Arctic 5. I also changed cases to an Antec Nine Hundred Two V3 at the same time, which may also have some effect with 200MM top fan and 120's in the front and back. I have to clean dust out about once a month from the filters since my office is in the mudroom where everyone comes into the house.
(..snip..) When I put thermal compound on before installing heatsinks, I take the edge of one of the package lids the CPU came in and smooth out and scrape off any extra. (..snip..)
i hang on to old credit cards (after running a couple neodymium magnets accross the strip) and use those, nice flat "sharp" edge. has worked perfectly for years :)
seeing without seeing is something the blind learn to do, and seeing beyond vision can be a gift.
One thing that I keep wondering about is the thermal stability of some of the cheaper CUDA enabled consumer graphics card.
I have a 9800 GT and a GT 240, and both were causing some troubles. The first would consistently produce only invalid results, so I considered it damaged and replaced it with the second, which would eventually lock up sometimes during task initialization.
Both problems were obviously correlated to ambient room temperature (problems started in hot conditions), but even before the fan would reach even 75 % capacity or the GPU temperature would reach any value close to the max spec value.
Recently I got a nice old, used workstation, a Dell Precision T4300 with a Core 2 Duo, E8400 @ 3 GHz. Very unlike the cheap consumer cases in my other PCs, this one has apparently thought-thru air ducts, big, quiet fans, perforated slot blinds etc, so the air really flows thru the case effectively. Installed in this case, my 9800GT which was already declared dead by me works without flaw so far. And after putting the PC with the GT 240 in a cooler room (and it's colder anyway now here), also the GT 240 has stopped locking up.
So I suspect that some (!) cards may have thermal trouble even before the card's cooling is at max. Perhaps it's not the GPU itself that is overheating, but some other parts (voltage regulators???). Both cards use only the PCI-E slot to draw power (no extra connectors). Maybe it's even the voltage regulators on the mother boards and not the card itself?? Or the PSU?
Anyway, if you see invalid results or other strange behavior with your CUDA WUs ==> improve the cooling!
When working at SGS (now STMicroelectronics), which produced printed circuit boards based on the Z80 microprocessor, I was astonished seeing how the technicians would test the boards connected to a display. They used a phon (hairdryer) aimed a every integrated circuit on the board. If the board crashed while the hairdryer was aimed at a given chip, they would substitute it. Very high tech solution.
Tullio
To avoid dust and overheating
)
To avoid dust and overheating I had mounted my MB right on the wall above the table. It is now easy to maintain and to install-remove components. No RFI since I use only internet-radio. The only headache is PSU - it is closed from the production.
RE: RE: Cooling? Consider
)
To repeat, 25 years and not a problem, to emphasize, not a single problem.
SUN certainly has well designed hardware. How can it be better than no case and a room fan? There is nothing better than room temperature. That is the way it is.
RE: RE: Cooling? Consider
)
Excuse but are their dust filters on the intakes of your case? What intakes do you have? Do you clean those filters regularly?
As you have neither intakes nor filters on them why would it be any different with or without the side panel? What dust build up have you found inside your machine?
RE: RE: RE: My SUN M20
)
Bingo! The real reason. RFI. However if you hear a buzz that is 60Hz and no case in this reality it going to shield it.
What you will hear is elevated noise cutting the S/N of the radio signal which is what the case does shield.
Back when Limbaugh was entertaining, pre-pills, I ran a shielded, grounded cable out the window and a long wire outside. FM is balanced. Use an old style twin-lead antenna with the crossbar of the T outside the window.
Better yet, as you have a computer, use internet radio.
I just built a system with an
)
I just built a system with an I7-2600K Z68 chipset, GTX560, 16GB RAM and had some real temperature issues with the CPU. Running E@H at 100% had the core temps 94-96C with the alarms set at 98C. No OC, it's running 3.5GHz.
I replaced the stock cooler with a CoolerMaster 212+ and they are now running 50-56C. The only issue is the cooler is huge about 6" tall so it takes a big case. I'm using a ThermalTake V series full tower with lots of fans.
Hi! I'm sure that in a
)
Hi!
I'm sure that in a significant percentage of cases (no pun intended), just re-appying the thermal interface grease stuff would improve cooling. E.g. last time I transplanted a CPU between boards, I could see that the thermal grease was really covering the whole surface of the heat spreader (instead of just closing the little scratches in it), actually insulating the CPU and the heat sink. It's amazing how careless this stuff is applied in off the shelf PCs.
HBE
It's also amazing how much of
)
It's also amazing how much of the goop that the CPU manufacturers put on the die when installing the heat spreader. I've popped the top on a number of old processors that were past any financial value, and both brands put quite a lot on before putting the spreader on. When I put thermal compound on before installing heatsinks, I take the edge of one of the package lids the CPU came in and smooth out and scrape off any extra. My i5-2400 went from 81 degrees C on the hottest core with the stock fan, heatsink, and thermal compound down to 52 with XIGMATEK Gaia SD1283 and Arctic 5. I also changed cases to an Antec Nine Hundred Two V3 at the same time, which may also have some effect with 200MM top fan and 120's in the front and back. I have to clean dust out about once a month from the filters since my office is in the mudroom where everyone comes into the house.
RE: (..snip..) When I put
)
i hang on to old credit cards (after running a couple neodymium magnets accross the strip) and use those, nice flat "sharp" edge. has worked perfectly for years :)
seeing without seeing is something the blind learn to do, and seeing beyond vision can be a gift.
One thing that I keep
)
One thing that I keep wondering about is the thermal stability of some of the cheaper CUDA enabled consumer graphics card.
I have a 9800 GT and a GT 240, and both were causing some troubles. The first would consistently produce only invalid results, so I considered it damaged and replaced it with the second, which would eventually lock up sometimes during task initialization.
Both problems were obviously correlated to ambient room temperature (problems started in hot conditions), but even before the fan would reach even 75 % capacity or the GPU temperature would reach any value close to the max spec value.
Recently I got a nice old, used workstation, a Dell Precision T4300 with a Core 2 Duo, E8400 @ 3 GHz. Very unlike the cheap consumer cases in my other PCs, this one has apparently thought-thru air ducts, big, quiet fans, perforated slot blinds etc, so the air really flows thru the case effectively. Installed in this case, my 9800GT which was already declared dead by me works without flaw so far. And after putting the PC with the GT 240 in a cooler room (and it's colder anyway now here), also the GT 240 has stopped locking up.
So I suspect that some (!) cards may have thermal trouble even before the card's cooling is at max. Perhaps it's not the GPU itself that is overheating, but some other parts (voltage regulators???). Both cards use only the PCI-E slot to draw power (no extra connectors). Maybe it's even the voltage regulators on the mother boards and not the card itself?? Or the PSU?
Anyway, if you see invalid results or other strange behavior with your CUDA WUs ==> improve the cooling!
HBE
When working at SGS (now
)
When working at SGS (now STMicroelectronics), which produced printed circuit boards based on the Z80 microprocessor, I was astonished seeing how the technicians would test the boards connected to a display. They used a phon (hairdryer) aimed a every integrated circuit on the board. If the board crashed while the hairdryer was aimed at a given chip, they would substitute it. Very high tech solution.
Tullio