Best program to tell temperatures?

My M11x R1 seems to have gotten very hot and has had some over heating issues in L4D2. I'm trying to find out what my actual heat temperatures are.

 

Ive tried a few and realtemp works really well for me. It not only tells your temperatures actively, it also monitors clock speeds and records min and max temps. Its a very lightweight program. It will also tell you your gpu temp if you set it to.

 

I use core temp:

Core Temp

It may not be the best out there, but it works!

 

RealTemp or ThrottleStop are both accurate. Stay away from SpeedFan as it's not accurate for either the R1 or R2.

 

I like HWmonitor:

CPUID - System & hardware benchmark, monitoring, reporting

 

HWmonitor is what Dell loads remotely on your computer when they do long range diagnostics. I suppose that may or may not mean it's good. lol

 

Is there a program to use that will not use alot of CPU % while running it?

 

None of them should be using a lot of CPU cycles or anything beyond a few percent.

 

I use HWiNFO32 for all my temperature needs.

No installation, GPU and CPU, small footprint, and can sit nicely on your taskbar.

 

Small footprint compared to what? Bigfoot?



Dell probably got in the habit of installing HWMonitor because for the vast majority of 45nm Core 2 mobile CPUs, it uses the wrong TJMax value so the core temperature is reported 5C lower than the actual core temperature.

 

CoreTemp + a gadget called All CPU Meter

 

HWMonitor has been around for a while. Wouldn't that have been solved by now?

 

I just downloaded the latest version, 1.17.0 from the HWMonitor website and it continues to use TJMax=100 for my T8100 even though the Intel website clearly shows it should be using 105C.

Intel® Core 2 Duo Processor T8100

It seems to be using the wrong value for the majority of 45nm mobile CPUs. When software assumes the wrong TJMax value, the reported temperatures will be off by the amount of error in TJMax.

If you realize this, you can go to the Intel website and look up your processor and then you can edit the hwmonitorw.ini configuration file and change this value manually from 100 to 105.

CPU_0_TJMAX=105.0

That's not a big deal but if you don't know about this then HWMonitor will be reporting your core temperatures incorrectly. If you are using HWMonitor on your M11x-R1 with a SU7300, you will also need to check and fix this error.

Intel® Core 2 Duo Processor SU7300

Edit: In just over 2 hours, HWiNFO32 has more than doubled its memory consumption. When memory usage continuously increases, that's usually a sign of a memory leak.
I'll have to download the latest version tomorrow and let Martin know if this is a problem still.

 

Kevin, please don't make harsh conclusions.
HWiNFO32 memory allocation grows over time when sensors are active because all values are saved in memory if the user decides later to create a logfile with all values sampled. This can easily be disabled if you click Configure in the sensors screen and turn off "Log all vales...".
Regarding the TJmax value... you just put too much importance on the DTS sensor, when if fact it's IMPOSSIBLE to determine a true TJmax value for CPUs prior to Nehalem. Even if you think that 105 C is correct on your CPU, I'm sure I could find 1000 same and other such CPUs where it's different. The other point is the ACCURACY of the DTS sensor and for this I'd suggest to read the " CPU Core temperature measuring via DTS - Facts & Fictions" thread on the HWiNFO32 Forum where I tried to explain this and I published the numbers (see my sig)..

 

I just downloaded and ran HWiNFO32 v.3.70 and Private memory usage was 100% stable at 16,504K for a couple of hours with it minimized to the system tray.



I was previously using HWiNFO32 v.3.55 and memory usage had grown to over 80,000K and was continuing to grow. I couldn't find an option in v.3.55 to prevent that from happening so that's why I disagreed with WaR when he said it had a "small footprint".

Intel publishes a TJMax specification for the majority of their 45nm Core 2 mobile processors and that value is 105C. I don't have a lab or 1001 CPUs of each model to verify the accuracy of Intel's specifications so I usually go along with their numbers unless proven otherwise.

Core Temperature = TJMax - Digital Thermal Sensor reading

AIDA64, Core Temp, Everest, HWiNFO32, RealTemp and ThrottleStop all use the Intel specified value when determining an approximate core temperature. I use the word approximate because we both know that Intel's temperature sensors are not 100% accurate and the TJMax spec is not 100% consistent from core to core or from one CPU to the next. It's not a perfect spec but it's probably closest to the actual number for the majority of these CPUs.

HWMonitor continues to use a value of 100C for most 45nm Core 2 CPUs so it reports core temperatures 5C cooler than all of the other programs out there. All that does is create confusion for users. Anyone with a Core 2 mobile CPU, like the SU7300 in the M11x-R1 need to be aware of this problem. I also showed how this can be corrected by editing the TJMax value in the HWMonitorw.INI configuration file.

 

Hm, confusion That's what Intel created with their info about DTS.. I have other docs which say if a certain bit in MSR is set, then Tjmax = 90 there.. And for dual-cores of that family 100 C.. and in some cases different again, etc.. Remember the IDF-Taipei TPWS002 when they wanted to help temperature monitoring developers and published Tjmax specs for several CPU models? Then they renamed it to TJ Target and Tj.. And last but not least most values published there had been proven totally wrong... this is confusion

EDIT: And many users don't realise the fact, that especially on older CPU families the DTS was not meant for temperature measuring (only to catch the state of T=Tjmax) and that on many of them when the temperature is < 50 C the sensor returns unusable temperature values.

 

So...

Can we agree that the TJMax values that AIDA64, Core Temp, Everest, HWiNFO32, RealTemp and ThrottleStop are using should be considered OK whereas those who are using some other value aren't?

 

No, we can't assume anything Depends on CPU model and how the tool is written. Some tools use a single value for a whole CPU family, whereas for example HWiNFO32 uses more detailed values per CPU model (which for some of them might not be valid).
There's no way how to determine the true Tj,max value for CPUs prior to Nehalem. We can only assume that for Nehalem and later where it's possible to read the Tjmax from a register all tools are consistent. But again - even some of those family (like the Gulftown) seem to have extremely inaccurate sensors..
Again, I advise to read the "DTS..." post on my forum..

 

With Nehalem and newer CPUs, the value in MSR 0x1A2 is not TJMax even though monitoring software assumes it is. Intel calls it MSR_TEMPERATURE_TARGET so that value is no more accurate than the previous CPU generations. It's just a target.

Intel never published TJMax information for their Core 2 desktop CPUs but they did publish this information in the datasheet for their Core 2 mobile CPUs. The majority of the 45nm Core 2 mobile CPUs are TJMax=105C. If it is in the datasheet, monitoring software should be using that value. HWMonitor has decided to ignore the datasheet and uses 100C instead.



The TJ numbers released at the IDF conference in document TPWS002 did not seem to be based on any engineering data. Those TJ numbers were so bad and so unbelievable that Intel had to release an updated version of the truth a month later that wasn't much more accurate than the original document.

passive101: Hopefully you can find a program you like from the list I posted. They all have strengths and weaknesses.