Linux - HardwareThis forum is for Hardware issues.
Having trouble installing a piece of hardware? Want to know if that peripheral is compatible with Linux?
Notices
Welcome to LinuxQuestions.org, a friendly and active Linux Community.
You are currently viewing LQ as a guest. By joining our community you will have the ability to post topics, receive our newsletter, use the advanced search, subscribe to threads and access many other special features. Registration is quick, simple and absolutely free. Join our community today!
Note that registered members see fewer ads, and ContentLink is completely disabled once you log in.
If you have any problems with the registration process or your account login, please contact us. If you need to reset your password, click here.
Having a problem logging in? Please visit this page to clear all LQ-related cookies.
Get a virtual cloud desktop with the Linux distro that you want in less than five minutes with Shells! With over 10 pre-installed distros to choose from, the worry-free installation life is here! Whether you are a digital nomad or just looking for flexibility, Shells can put your Linux machine on the device that you want to use.
Exclusive for LQ members, get up to 45% off per month. Click here for more info.
i installed the 'sensors' package to see what kind of temps i was getting, and whether or not to bother using the arctic silver 5 i bought. i used it anyway, and my temperatures didnt improve much.
anyway, i'm running FC4 on a dual opteron setup. i dont have any intake fans (yet), but i have one large 120 mm fan at the rear, and of course a psu fan. i also have a fanless video card, which gets quite hot. i tried leaving my case sealed to get better airflow, and my temps are still pretty outrageous. i know you guys are going to jump on my back about the intake fans, and rightly so, once you see my temps. but the strange thing is, i can put my finger RIGHT at the base of the heatsink/fan, and the thing is barely warm. none of the fins of the hsf are remotely warm, but the very base of it (solid metal, no fins) is hardly warm at all.
i was very happy with the system temp through inspection, until i got 'sensors' running. first of all, i cant determine where my real CPU temps are. i would think since i have 2 cpus, tat would mean 2 temp sensors? i'm starting to question my common sense, here.
here is the output of sensors.
Code:
[cjreynolds@localhost ~]$ sensors
w83627hf-isa-0290
Adapter: ISA adapter
VCore 1: +1.54 V (min = +1.42 V, max = +1.57 V)
VCore 2: +1.55 V (min = +1.42 V, max = +1.57 V)
+3.3V: +3.38 V (min = +3.14 V, max = +3.47 V)
+5V: +5.11 V (min = +4.76 V, max = +5.24 V)
+12V: +4.99 V (min = +10.82 V, max = +13.19 V)
-12V: -1.42 V (min = -13.18 V, max = -10.80 V)
-5V: -3.74 V (min = -5.25 V, max = -4.75 V)
V5SB: +5.64 V (min = +4.76 V, max = +5.24 V)
VBat: +3.02 V (min = +2.40 V, max = +3.60 V)
fan1: 0 RPM (min = 0 RPM, div = 2)
fan2: 0 RPM (min = 0 RPM, div = 2)
fan3: 0 RPM (min = 0 RPM, div = 2)
temp1: +36°C (high = -1°C, hyst = -1°C) sensor = thermistor
temp2: +38.5°C (high = +80°C, hyst = +75°C) sensor = thermistor
temp3: +39.0°C (high = +80°C, hyst = +75°C) sensor = thermistor
vid: +1.500 V (VRM Version 2.4)
alarms: Chassis intrusion detection ALARM
beep_enable:
Sound alarm disabled
eeprom-i2c-0-54
Adapter: SMBus nForce2 adapter at 5000
Memory type: DDR SDRAM DIMM
Memory size (MB): 1024
eeprom-i2c-0-50
Adapter: SMBus nForce2 adapter at 5000
Memory type: DDR SDRAM DIMM
Memory size (MB): 1024
adt7463-i2c-1-2e
Adapter: SMBus nForce2 adapter at 5040
V1.5: +2.594 V (min = +1.42 V, max = +1.58 V) ALARM
VCore: +1.292 V (min = +1.71 V, max = +1.89 V) ALARM
V3.3: +3.330 V (min = +3.13 V, max = +3.47 V)
V5: +5.143 V (min = +4.74 V, max = +5.26 V)
V12: +12.078 V (min = +11.38 V, max = +12.62 V)
CPU_Fan: 2631 RPM (min = 4000 RPM) ALARM
fan2: 0 RPM (min = 0 RPM)
fan3: 0 RPM (min = 0 RPM)
fan4: 0 RPM (min = 0 RPM)
CPU: +50.75°C (low = +10°C, high = +50°C)
Board: +37.75°C (low = +10°C, high = +35°C) ALARM
Remote: +55.50°C (low = +10°C, high = +35°C) ALARM
CPU_PWM: 255
Fan2_PWM: 255
Fan3_PWM: 255
vid: +1.800 V (VRM Version 9.1)
ok, here is what makes me doubt my results.
Code:
+12V: +4.99 V (min = +10.82 V, max = +13.19 V)
-12V: -1.42 V (min = -13.18 V, max = -10.80 V)
-5V: -3.74 V (min = -5.25 V, max = -4.75 V)
with those numbers, the system wouldnt even post. gotta be BS. the nForce2 adapter's voltages are the ones i trust. until you look at the V1.5 and VCore. the strange thing is, the ISA adapter seems to have the correct VCore and V1.5, but incorrect +12, -12, and -5. it seems that both adapters have both good and bad values. i'm hoping the ISA cpu temps are correct. my cpu is idle, i think 50.75°C is a little too high for idle. my bios temps arent much better, they are around 63°C idle. one time, i restarted, went into bios immediately, and cpu0 showed 128°C, the other showed 66°C. yes, celsius, not fahrenheit.
Re: motherboard temp sensors, shoudl i trust them?
Quote:
Originally posted by santaliqueur my bios temps arent much better, they are around 63°C idle. one time, i restarted, went into bios immediately, and cpu0 showed 128°C, the other showed 66°C. yes, celsius, not fahrenheit.
i'm very confused, and i know this isnt really a linux specific question, but if anyone who can help me make sense of all this, i'll be very thankful.
CJ
I have seen the same type of results from lm_sensors, which I
assume is what you mean by the "sensors package." From my
experience, with 5 comps, the only thing that works is ACPI on
one laptop. Though I have 3 PCs with ACPI boards, nothing at
all works on them (there is nothing in /proc/acpi/ related to the
thermal zone stuff).
Therefore, I don't really trust anything but the BIOS. I have used
the ACPI readings with GkrellM so that I can see it constantly, but
that's only CPU temp, and it is bogus most of the time. I keep it
set to Celcius, and sometimes it drops to 8C. Restarting GkrellM
will change it, sometimes for seconds, sometimes longer. I finally
quit running GkrellM, because it also filled syslog with errors.
The only thing that works here is to cat it to stdout, so I have this
running all the time in a term. That "-n 1" tells it to run every one
second, which is what I'm using to get reliable results.
ACPI doesn't work properly in *nix on this box, no matter what I
have done. It does on Windoze, so it's a problem with the Linux
kernel, not the hardware.
Hope this helps...but certainly you have some incorrect results.
I have been using the drivers in the 2.6 kernel for a good long time and Gkrellm to monitor them. I do have a problem with the drivers not changing the values after I have been up a long while but Gkrellm works fine. I made my drivers as modules and then wrote a script that unloads them then reloads them which sort of resets them.
When the values are changing though and getting updated, they are very accurate for my mobo. The chip on mine actually puts out the temp in Celcius just like in the BIOS. I have read where some mobos have to go through some math to even get Celcius figured out.
I'm sure my drivers are different from yours so they will work different but there is some that work like a champ though.
I'm with the other guy though, trust the BIOS for sure and certain. That should be dead on. I have two sensors on mine, one CPU, one ambient. I took a fan and blew air across it and checked the ambient air temp with a known accurate meter then checked the ambient value on my rig. It was only off a degree or so which is really close. Both sensors are the same type so the CPU one should be OK too.
I think if you were running that hot, you would be seeing smoke. I hope you can get that right though. You may want to search for your mobo on google4linux and see what others are getting.
I'm with the other guy though, trust the BIOS for sure and certain
i normally would agree with both of you guys, but i saw a temp of 128°C after rebooting one time. the heatsink/fan was cool to the touch, even right at the base. that kind of temp would have melted my finger to the aluminum, if it hadnt split the cpu into pieces and sent it shooting out of the socket. i'm thinking there is something wrong with the sensors themselves. or lm_sensors, or both.
Location: 1st hop-NYC/NewJersey shore,north....2nd hop-upstate....3rd hop-texas...4th hop-southdakota(sturgis)...5th hop-san diego.....6th hop-atlantic ocean! Final hop-resting in dreamland dreamwalking and meeting new people from past lives...gd' night.
Distribution: Siduction, the only way to do Debian Unstable
Posts: 506
Rep:
One sensor on my board shows 125degrees C. but the cpu is 37 and the chipset is 23 degrees C.I believe its labled as power,so I was assumeing it was the core pwr supply temp or something similar or one of the temps didnt convert to C from F.
For the most part lm sensors is about on par if not exactly on par with my bios and various windows tools I used to have on this box.
Thermal paste is not fool proof to apply. I suggest thermal tape because it is fool proof. It does provide better heat transfer than thermal paste over time. Artic Silver is different from one manufacture to another. One thing that could not hurt is to make the bottom of the heatsink smooth. Do not take the time to make it shiny because you are going to make it dirty again.
i'm not worried about my thermal grease job. i put a very small amount on the cpu and smoothed it out with a razor blade until it was almost transparent. a bad thermal grease job isnt going to make a cpu temp 128 degrees. the sensor HAS to be faulty.
Location: 1st hop-NYC/NewJersey shore,north....2nd hop-upstate....3rd hop-texas...4th hop-southdakota(sturgis)...5th hop-san diego.....6th hop-atlantic ocean! Final hop-resting in dreamland dreamwalking and meeting new people from past lives...gd' night.
Distribution: Siduction, the only way to do Debian Unstable
Posts: 506
Rep:
Thermal paste is fine and is recommended by all overclockers as myself.
The tape is crap.
Your advice on smoothing the heatsink or "lapping" as we say in the overclocking world is a great way to make the cooling process by the heatsink fan much better.There are plenty of websites with tutorials on whats needed and the proper procedure to do so.
I use ceramique from artic silver guys but artic silver 5 has also given great results.
Originally posted by ironwalker Thermal paste is fine and is recommended by all overclockers as myself.
The tape is crap.
Your advice on smoothing the heatsink or "lapping" as we say in the overclocking world is a great way to make the cooling process by the heatsink fan much better.There are plenty of websites with tutorials on whats needed and the proper procedure to do so.
I use ceramique from artic silver guys but artic silver 5 has also given great results.
Tape is not crap. Paste becomes crap when first applying it to the processor. Paste is not fool proof. Amplifier amature builders uses tape or mica because it is less fuss and the builder does not have to question themselves "Did I put too much or too little". If one output transistor is too hot the distortion of the amplifier will be higher than expected. Tape has similar heat transfer as Artic Silver and is more reliable than you might think.
I have used the tape on transisters a lot. I would never do that on a CPU though. Tape is not as thermally conductive as a good paste. If it is a transistor that disapates a good bit of heat, I use the paste and a mica insulator. If you use mica, you still have to use paste, on both sides. Mica is a electrical insulator, not a thermal conductor. It isolates the tab from the heatsink. Some tabs connect to the collector and they are usually connected to the power supply which depending on the voltage can provide a shock. It can cause a short as well. Most heat sinks are grounded. Triac tabs are usually attached to MT2 which in a controller for a 120 or 240 volt device can have the full 120 or 240 volts, AC to boot which usually means a lot of current capability. I have never tested but have read that AS5 is conductive. Be carefull if you use it on transistors. Normal paste is not conductive.
Yes, I have a good bit of experience with electronics. I took my courses back in the mid - late 80's. I also have designed a lot of amps, power supplies and such as that. I have been messing with electronics for over 20 years. I have built a 12 volt power supply that is regulated and puts out 80 amps at about 13 volts. It doubled as my battery charger. The transformer alone weighs about 70 lbs, thank god for wheels. Tape has its place but is not as good as the paste. I have taken transisters off that I used paste on and has been installed for many years and the paste has never dried out or broken down in any way. I use ECG424 paste made by Phillips. I bought a large tube of it from Bluff City Electronics.
All that said, I have the genuine AS5 on mine and I polished the heatsink until it looked like a mirror. I could read the newspaper in the reflection it was so smooth. It made my temp drop a good bit. Right now, ambient: 72.5F CPU: 91.4F. Fan speed is about 3100RPMs. That is at full load. I run folding on this rig and the difference between ambient and CPU is rarely more than 20F, usually when I forget and leave the A/C off. At idle the difference is about 10F. My CPU fans adjusts its speed according to CPU temp by a temp sensor mounted to the bottom of the CPU itself. That makes it not change a whole lot.
My advice, get some genuine AS5, polish the heatsink to at least a 1200 or 1400 grit sandpaper grit. That will smooth it up a lot. I got my sandpaper at a local auto parts store. Get the wet/dry kind, it lasts longer. If you put some water, oil, or alcohol on the sandpaper it will sand better and faster. It took me about a hour to do mine. Be carefull to keep it flat if you decide to do this. There are instructions on the overclockers forums on how to do this. I would not use tape on anything but some of the slower CPUs that don't generate as much heat.
Distribution: RH 6.2, Gen2, Knoppix,arch, bodhi, studio, suse, mint
Posts: 3,304
Rep:
I think your isa sensor is showing the correct cpu temp.
Run a cpu intensive app, like lame encoding the same mp3 over and over and watch your sensors to see which temp goes up the most. Stop the app and watch for the sensor to go back down. That's how you'll find which one is really showing your cpu temp.
LinuxQuestions.org is looking for people interested in writing
Editorials, Articles, Reviews, and more. If you'd like to contribute
content, let us know.