Stress-test SW, Windows 7 and "BSOD"

[Idontcare]

an ASUS tech-rep announced that "CPU" is indeed "TCASE." So the oddball "core" temperature I'd noticed -- pushing closer to 75 to 80C at higher clocks above 4.7 Ghz, would just be an anomaly for the overall "package" or TCASE of around 70C.

How can ASUS automagically detect and report TCASE when measuring Tcase requires milling your IHS and placing a thermocouple between the IHS and HSF?

 

[BonzaiDuck]

How can ASUS automagically detect and report TCASE when measuring Tcase requires milling your IHS and placing a thermocouple between the IHS and HSF?

I could ask the same question. I suspect that they actually use some algorithm based on the core [TJunction] temperatures. It probably factors in information about the thickness of the IHS and copper thermal properties. But the shareware softwares, like AIDA64, HWMonitor, etc. AS WELL AS ASUS Monitor somehow "got it wrong." It defies logic that their "package" or "CPU" temperatures should be equal or greater (by a degree or so) than the higher of the core values.

I had take the approach of watching the core values AS IF they were "TCASE," so my current clock and voltage settings are [probably] more than safe.

Perfectionists tend to ruin things for themselves. I've now got it in mind to install a "T-Balancer Big-NG" fan controller. The problem: the software won't read CPU and mobo temperatures on the new mobos without additional software -- which -- to the mCubed [T-Balancer] people -- can only be MBM. And as we all know, MBM is pretty much defunct -- won't run under Win 7. It's the equivalent of a software kludge, even if you could use it in Win 7.

So I'm thinking about dremeling a quarter-inch-long, 2mm-deep slot in the NH-D14 heatsink base -- to get an analog sensor "between" IHS and HSF base, or between the IHS/HSF and the HSF's heatpipes. P . . I . . . T . .. A.

You'd wonder if it wouldn't be too much trouble for mobo makers to design the circuitry underneath the center of the processor so they could provide you a 2mm hole in the mobo for inserting an analog sensor that way. I think with the newer socket-designs, it's either impossible or asking for trouble to try getting a sensor under the processor with the sensor wires between the gold pins . . .

 

[Idontcare]

I could ask the same question. I suspect that they actually use some algorithm based on the core [TJunction] temperatures. It probably factors in information about the thickness of the IHS and copper thermal properties. But the shareware softwares, like AIDA64, HWMonitor, etc. AS WELL AS ASUS Monitor somehow "got it wrong." It defies logic that their "package" or "CPU" temperatures should be equal or greater (by a degree or so) than the higher of the core values.

I had take the approach of watching the core values AS IF they were "TCASE," so my current clock and voltage settings are [probably] more than safe.

Perfectionists tend to ruin things for themselves. I've now got it in mind to install a "T-Balancer Big-NG" fan controller. The problem: the software won't read CPU and mobo temperatures on the new mobos without additional software -- which -- to the mCubed [T-Balancer] people -- can only be MBM. And as we all know, MBM is pretty much defunct -- won't run under Win 7. It's the equivalent of a software kludge, even if you could use it in Win 7.

So I'm thinking about dremeling a quarter-inch-long, 2mm-deep slot in the NH-D14 heatsink base -- to get an analog sensor "between" IHS and HSF base, or between the IHS/HSF and the HSF's heatpipes. P . . I . . . T . .. A.

You'd wonder if it wouldn't be too much trouble for mobo makers to design the circuitry underneath the center of the processor so they could provide you a 2mm hole in the mobo for inserting an analog sensor that way. I think with the newer socket-designs, it's either impossible or asking for trouble to try getting a sensor under the processor with the sensor wires between the gold pins . . .

Doesn't your mobo have a socket thermo-couple? I thought they were all required to have a thermo-couple located in the vicinity of the socket so that the "package" temps could be monitored?

I don't really know the answers, just vocalizing some preconceptions I had/have about the topic.

Personally I would only worry myself over the Tcase temps if I were actually concerned about the longevity of the CPU. However, I really have no desire to have the CPU live more than 3yrs.

So instead I focus on DTS temperature as that number plays a critical role in the Vcc vs. GHz stability line which in turn determines my OC speed vs. HSF noise levels.

 

[BonzaiDuck]

Doesn't your mobo have a socket thermo-couple? I thought they were all required to have a thermo-couple located in the vicinity of the socket so that the "package" temps could be monitored?

I don't really know the answers, just vocalizing some preconceptions I had/have about the topic.

Personally I would only worry myself over the Tcase temps if I were actually concerned about the longevity of the CPU. However, I really have no desire to have the CPU live more than 3yrs.

So instead I focus on DTS temperature as that number plays a critical role in the Vcc vs. GHz stability line which in turn determines my OC speed vs. HSF noise levels.

"thermocouple located in vicinity of socket . . . " I wouldn't know.

What is "DTS temperature?" Either my memory escapes me, or I need your informative assistance.

I understand "where you're comin' from" on the longevity issue. I try to find some OC settings that wouldn't affect lifespan so much, and then work from there. But, perhaps even in violation of our forum precept that "OC'd systems shouldn't be used for mission-critical work . . . " I use my system for less-than-mission critical and want to avoid unexpected unhappiness or confusion when the CPU starts going south.

Even so, my OC'd Q6600 B3 seems to continue ticking along after four years with a fixed VCORE. My sibling brother may get another few years out of it without dropping the clock or voltage . . .

 

[Idontcare]

DTS is the sensor responsible for reporting the "distance to TJmax" value.

Basically it is as close as you are going to get in terms of monitoring the temperatures that the xtors and wiring inside the chip are experiencing, precisely the same temperature that determines the S/N ratio and electrical stability of the circuits themselves.

 

[john3850]

Bonzai here is a good read

http://www.techreaction.net/2009/10/14/guide-to-understanding-intel-temperatures/

 

[BonzaiDuck]

Bonzai here is a good read

http://www.techreaction.net/2009/10/14/guide-to-understanding-intel-temperatures/

Thanks . . . that is, indeed, a good summary. I may be already familiar with much of the info, but a good reference to store in my "SB Project" PDF library.

Some info here or there on the forums suggests that temperature is less of a threat than voltage, but the game I'm playing with myself is to eliminate all sources of uncertainty about my i7-2600K, even in light of a rumored 2700K and an IB socket-1155 in the near future. If my "TCASE" ["CPU"] temperature reported by ASUS Monitor is close to "reality," I've got some room left. But my voltages are close enough to what we'd consider safe. Maybe . . . maybe . . . just another 0.01V, though . . .

 

[tweakboy]

My advice is, use your computer and get stuff done, play games, do your normal thing. And if you can do all this without a BSOD, then your comp is fine. The stress test and your cpu aint playing nice with each other imo.

 

[BonzaiDuck]

My advice is, use your computer and get stuff done, play games, do your normal thing. And if you can do all this without a BSOD, then your comp is fine. The stress test and your cpu aint playing nice with each other imo.

We're way beyond that since I started this thread. You could say the thread was "derailed" but it really just took a detour. Much of the information here (even mine) may be useful to a lot of SB over-clockers who want to see what others are doing.

There are many minor related issues. "Does my monitoring software report voltages and temperatures with reasonable accuracy?" "What sort of voltage adjustments can I make?" "What do I do with PLL over-voltage?" "What do I do with LLC?" "What differences to expect between air-cooling and water?" etc. etc.

I looked to see if I could change the thread title for the parts "Win 7" and "BSODs," but can't find the capability for that . . .

 

[Idontcare]

I looked to see if I could change the thread title for the parts "Win 7" and "BSODs," but can't find the capability for that . . .

In vB there's a timelimit beyond which the thread's OP can't change the title. But a mod can. If you happen to know a mod you could pm them with a request for a title change, which if they approve and accept then they can/will usually implement the change for you. 😉 😉

 

[BonzaiDuck]

In vB there's a timelimit beyond which the thread's OP can't change the title. But a mod can. If you happen to know a mod you could pm them with a request for a title change, which if they approve and accept then they can/will usually implement the change for you. 😉 😉

Let me analyze and formulate a bit, and I'll get back . . . .