Batteries
in [Mandrake]
|
From: Moe Trin on 5 Oct 2007 16:10 On Fri, 05 Oct 2007, in the Usenet newsgroup alt.os.linux.mandrake, in article <cRiNi.145986$bO6.59282(a)edtnps89>, Unruh wrote: >ibuprofin(a)painkiller.example.tld (Moe Trin) writes: >>Intel builds the sensor (a diode on the die) as part of the CPU itself. >>What you are measuring is the forward voltage drop across the diode >>under a "known" current. That translates directly into temperature at >>that spot on the die (which probably isn't the hot spot on the die). >>See the Intel spec sheets for the CPU for details. The words you are >>looking for are "Thermal Diode". The problem is measuring that >>voltage accurately, and converting that digital value into an assumed >>die temperature. > >But since I suspect that the transverse conductivity is at least as good >as the vertical, I would suspect this temp is close to the mean of the >die. From the Celeron spec sheet (others are similar if not identical): 6.1 Thermal Diode The mobile Intel Celeron processor has an on-die thermal diode that can be used to monitor the die temperature (T J). A thermal sensor located on the motherboard, or a stand-alone measurement kit, may monitor the die temperature of the processor for thermal management or instrumentation purposes. Table 33 and Table 34 provide the diode interface and specifications. Note: The reading of the thermal sensor connected to the thermal diode will not necessarily reflect the temperature of the hottest location on the die. This is due to inaccuracies in the thermal sensor, on-die temperature gradients between the location of the thermal diode and the hottest location on the die, and time based variations in the die temperature measurement. Time based variations can occur when the sampling rate of the thermal diode (by the thermal sensor) is slower than the rate at which the T J temperature can change. >If this thing is reporting temperatures out by 20C it is not worth >anything. Intel CPUs report their temperatures far more accurately than >that. My cpy even not running is at around 38C under low loads. I simply >do not believe 27 unless you are pouring liquid N2 over the chip. The CPU doesn't report the temperature. You have a device that is off the CPU that is measuring the voltage drop across the thermal diode (hopefully at the specified current flow). You then have software or firmware elsewhere that translates that measured voltage into a die temperature at the location of the thermal diode. Also, don't forget that the voltage drop is a function of temperature measured in Kelvins, so the change from 30C to 40C is only around a third of a percent. Old guy
From: Unruh on 5 Oct 2007 20:49 ibuprofin(a)painkiller.example.tld (Moe Trin) writes: >On Fri, 05 Oct 2007, in the Usenet newsgroup alt.os.linux.mandrake, in article ><cRiNi.145986$bO6.59282(a)edtnps89>, Unruh wrote: >>ibuprofin(a)painkiller.example.tld (Moe Trin) writes: >>>Intel builds the sensor (a diode on the die) as part of the CPU itself. >>>What you are measuring is the forward voltage drop across the diode >>>under a "known" current. That translates directly into temperature at >>>that spot on the die (which probably isn't the hot spot on the die). >>>See the Intel spec sheets for the CPU for details. The words you are >>>looking for are "Thermal Diode". The problem is measuring that >>>voltage accurately, and converting that digital value into an assumed >>>die temperature. >> >>But since I suspect that the transverse conductivity is at least as good >>as the vertical, I would suspect this temp is close to the mean of the >>die. >From the Celeron spec sheet (others are similar if not identical): > 6.1 Thermal Diode > > The mobile Intel Celeron processor has an on-die thermal diode that can > be used to monitor the die temperature (T J). A thermal sensor located > on the motherboard, or a stand-alone measurement kit, may monitor the > die temperature of the processor for thermal management or > instrumentation purposes. Table 33 and Table 34 provide the diode > interface and specifications. > > Note: The reading of the thermal sensor connected to the thermal diode > will not necessarily reflect the temperature of the hottest location > on the die. This is due to inaccuracies in the thermal sensor, on-die > temperature gradients between the location of the thermal diode and > the hottest location on the die, and time based variations in the die > temperature measurement. Time based variations can occur when the > sampling rate of the thermal diode (by the thermal sensor) is slower > than the rate at which the T J temperature can change. Agreed, but, I would expect it to be far more accurate as a reflection of even the hottest portion of the die than is a motherboard mounted thermometer. >>If this thing is reporting temperatures out by 20C it is not worth >>anything. Intel CPUs report their temperatures far more accurately than >>that. My cpy even not running is at around 38C under low loads. I simply >>do not believe 27 unless you are pouring liquid N2 over the chip. >The CPU doesn't report the temperature. You have a device that is off >the CPU that is measuring the voltage drop across the thermal diode >(hopefully at the specified current flow). You then have software or >firmware elsewhere that translates that measured voltage into a die >temperature at the location of the thermal diode. Also, don't forget >that the voltage drop is a function of temperature measured in Kelvins, >so the change from 30C to 40C is only around a third of a percent. I agree, but that voltage drop is a function of the temperature on the die. And a thermometer that is out by that much is pretty useless. And I still do not believe that the temp of his cpu is 27C
From: Peter D. on 6 Oct 2007 03:07 on Saturday 06 October 2007 07:54 in the Usenet newsgroup alt.os.linux.mandrake Adam wrote: [snip] > Moe Trin wrote: [snip] >>> Went outside and timed stator of watthour meter with printer >>> unplugged (1:23, 1:14, 1:16) and plugged in but switched off (1:49, >>> 1:52, 1:56) -- that's a minute and some seconds for the stator to >>> make one revolution. Apparently I use /less/ power with it plugged in! >> >> Geez, get 20 of them, and you may not have to pay anything for electrons. > > Yesterday morning I measured about 42.5W total household power > consumption with the printer unplugged, and about 28.5W with it plugged > in but switched off, using method #3 on > http://michaelbluejay.com/electricity/measure.html . Something isn't > right there. (Everything was switched off except for the clocks and the > things that have a "standby" state, like the VCR and the touch-on lamp.) [snip] Maybe.. Something you don't know about is switching on/off in response to a thermostat (the fridge?) or a timer (the VCR?). You have a highly reactive load and a cheap power meter. It is plausible that your power meter is actually an Amp meter, so plugging in a device that corrects the power factor will reduce your electricity costs. You are going crazy. -- sig goes here... Peter D.
From: Adam on 7 Oct 2007 20:23 Unruh wrote: > And a thermometer that is out by that much is pretty useless. > And I still do not believe that the temp of his cpu is 27C Thanks! I don't believe it either. Details in the "Temperature Sensors" thread in this NG (a.o.l.mandrake). Does CPU 43, MB 30 for idle system and CPU 45, MB 40 after 60 seconds of "stress" sound more plausible? The most important point is that, since the flaky DIMM was removed (your suggestion, and a good one), I haven't found any problems relating to temperature. Adam
From: Jim Beard on 7 Oct 2007 20:32
Adam wrote: > Unruh wrote: >> And a thermometer that is out by that much is pretty useless. >> And I still do not believe that the temp of his cpu is 27C > > Thanks! I don't believe it either. Details in the "Temperature > Sensors" thread in this NG (a.o.l.mandrake). Does CPU 43, MB 30 for > idle system and CPU 45, MB 40 after 60 seconds of "stress" sound more > plausible? The most important point is that, since the flaky DIMM was > removed (your suggestion, and a good one), I haven't found any problems > relating to temperature. I recently started looking at alt.comp.hardware.x86-64, and found a few things under a CPU Temperature thread that may bear repeating: > > I have a Athlon 64x2 4600+(65) windson AM2,ASUS M2N-E MB,ASUS EN7600GS > > video card,and 2G memory. I have the stock heat sink with a 5000 rpm fan > > on it with Arctic silver paste. The CPU temperature runs around 36c and > > the MB at 34.Room temperature is around 80F. I mainly use Foxfire on > > the Inter net. Is this temperature ok are do I need to get a better heat > > sink. If so what is a good one to get . Thanks Johnny > > Gainesville Fl --- 36C is great especially in an 80F room. --- 36C sounds like the idle temp, if that's a loaded temperature then it's even better. Generally you want your loaded temperatures to be in the 50s or less but the CPU is probably speced to run up to 75C (it could be a little more or a little less) so you are running at a very comfortable temp. --- Every CPU has a different max temp allowed, you can look up the max temp for AMD desktop chips here, http://www.amdcompare.com/us-en/ --- Cheers! jim b. -- UNIX is not user-unfriendly; it merely expects users to be computer-friendly. |