Timeline for Power apps for S5R4 Run?

Randall  McPherson
Randall McPherson
Joined: 14 Mar 05
Posts: 6
Credit: 8,341,604
RAC: 0

RE: One other possible

Message 83502 in response to message 83500

Quote:

One other possible explanation comes to mind. If the system was indeed in some sort of ON-DEMAND mode, then ordinary ongoing BOINC activity (at lowest priority) might have allowed it to slip back to power-saving speed. But as soon as you started to investigate the problem, downloading and then running tools like CPU-Z, the foreground activity might have been enough to tickle it up to full speed for the duration.

Truly a quantum effect, where the act of observation determines the state of the entity being observed!

That's an interesting possibility. It seems impossible, given the dismal performance figures, that the clock was actually running at rated speed during BOINC processing. Either the tools were inaccurate, or some effect like the above was causing it to be unobservable.

As for the temperature readings, the heat sink on the processor right now is actually fairly sizable and does not feel terribly warm to the touch. I am a little skeptical of the accuracy of the temperature readings, as Core 0 (the first core) has a temperature reading four degrees higher in Real Temp than the other three cores. This seems unlikely to be the reality since they are all running E@H WUs. In Speedfan, the same disparity exists but all the temperatures are 10 degrees higher (e.g. 77 instead of 67). So, I'm not sure what to believe about the temperature readings. The clock speed does not seem to be getting throttled back though.

archae86
archae86
Joined: 6 Dec 05
Posts: 2,741
Credit: 2,903,002,054
RAC: 3,307,410

RE: As for the temperature

Message 83503 in response to message 83502

Quote:
As for the temperature readings, the heat sink on the processor right now is actually fairly sizable and does not feel terribly warm to the touch. I am a little skeptical of the accuracy of the temperature readings, as Core 0 (the first core) has a temperature reading four degrees higher in Real Temp than the other three cores. This seems unlikely to be the reality since they are all running E@H WUs. In Speedfan, the same disparity exists but all the temperatures are 10 degrees higher (e.g. 77 instead of 67). So, I'm not sure what to believe about the temperature readings. The clock speed does not seem to be getting throttled back though.


The temperature sensors are actually junctions, and on any given chip not perfectly matched. The truly attentive temperature mavens advocate calibrating the individual sensors on each chip (and Speedfan actually gives you a place to do it{Readings|Configure|Advanced|Chip:Intel CORExxxxx|Temperature n offset|set to +/-n|remember it).

The 10 degree offset thing is another matter. Intel actually does not support the interface people use to report core temperatures for this purpose. So people resort to backdoor strategems to deduce one crucial reference. Particularly for Q6600 parts and their close relatives, some of the available temperature-reporting third party aps make different assumptions than others by stepping.

There is supposed to be a user-accessible register which reports this offset, though Intel (is this getting familiar) does not support this use and will not commit to its accuracy. It changed value between two Q6600 steppings, leading to widespread reports of an absolutely massive change in operating temperature at the same operating conditions. This was absolutely preposterous, particularly since (literally) the same dice are used in Duo parts, for which not such miracle was reported.

So, if you really want to know calibration offset for the on-die sensors, try to arrange to put them at a known temperature.

My efforts on these lines involved setting all the fans in my desktop PC to max, removing a side panel, and blowing air from a two-foot square window fan into the case from a few inches away, for half an hour with the system powered off. After that soaking, I booted to an idle state, and watched the temps decay back down from the brief pulse of power to boot.

I'm much more confident of the offsets than the absolute value. But to help make the case that the sensors are not perfect,I'll just note than on my E6600 one of the sensors reported a temperature substantially below ambient to this test.

On a less jolly note, I think the sensors may drift differentially with time. My cores at idle are no longer very well-matched, but I'm not motivated enough to drag out the fan and redo the calibration.

Comment viewing options

Select your preferred way to display the comments and click "Save settings" to activate your changes.