Hardware info 7.02 broken boost clocks and old bug still here

D

Deleted member 16431

Guest
Hi, I will go straight to the point.
On hardware info 7.02 a strange behaviour which started in version 6.x and was fixed in the last 2 releases of version 6 and was not present in version 7.00 has appeared again: when at idle my cpu usually idles and 599-499 mhz on all 4 cores, but one version 7.02 and some older 6.x versions the following happens and happened: the cpu will go down to 599mhz but in steps in the span of 5 seconds it will boost to 699-799-1100-997 in this exact sequence and the cpu will settle idle at 997. This as I said only happens on some hardware info versions and I am pretty sure Hwinfo is the culpirit of this behaviour.
The system affected by this is a core i7-8565u based laptop (asus x512fj-ej021t).

The other bug is regarding hwinfo in both 64 and 32 bit versions and is found both in windows xp and windows 10. It affects my core duo e4300 desktop mounted on an msi-g41tm p33 motherboard. Hardware info reports temperature for both the cpu cores which are too low, every other monitoring software reports exactly 20 C more on each cpu core and therefore 20 C less on the margin of temp before thermal throttling is reached. This behaviour has been there since at least version 5.9. At first I was unsure whether it was hwinfo or other software fault but I think it is hardware info, at idle the temps reported can be even 1-2 C lower than ambient and they are lower than the temperature reported by the motherboard when looking at cpu temp in bios, also they are lower compared to any other sensor on the motherboard. Also (don't blame me) I am pretty sure that the cpu cannot be near ambient temperature because it is on a wraith like amd cooler which has never seen a repaste in the past 16 yrs so i think that the idles temps aroun 36-40 C are correct instead of 20C.

This last bug does not seem to affect another system I have which is based on a core e7600 mounted still on another msi mobo but both in windows xp and windows 10 the temperature reading seems correct, around 5-6 C more compared to ambient on a 1 yrs old repasted amd wraith like cooler so I think that the temp reported with that cpu is not affected by some strange bug.

Last stupid question, is there a chance or even a way to get some data from the gpu on those cpus? It is still the old approach of having the gpu on the motherboard and not inside the cpu package so I do not know if it is possible.

I hope someone can illuminate me about this.
In the meantime
Have a good day
 

Martin

HWiNFO Author
Staff member
I think the first issue with clock is due to the observer effect - the CPU is very sensitive and just a slight load causes it to ramp up the clocks. It's very tough to determine what exactly is causing this (might be just some minor feature or enhancement that is driving this). Disabling monitoring of some sensors not needed might help. Anyway, the actual clocks you see are at such level only for a very brief period of time exactly when they are being detected. I'd suggest to rather rely on the Effective Clock values as per: https://www.hwinfo.com/forum/threads/effective-clock-vs-instant-discrete-clock.5958/

Regarding the temperature on E4300 this is certainly due to a different assessment of Tj,max for that model. It cannot be precisely determined as described here: https://www.hwinfo.com/forum/threads/cpu-core-temperature-measuring-facts-fictions.148/
You can adjust the Tjmax to a value that you believe is reasonable by right-clicking on the DTS sensor.
 
D

Deleted member 16431

Guest
I think the first issue with clock is due to the observer effect - the CPU is very sensitive and just a slight load causes it to ramp up the clocks. It's very tough to determine what exactly is causing this (might be just some minor feature or enhancement that is driving this). Disabling monitoring of some sensors not needed might help. Anyway, the actual clocks you see are at such level only for a very brief period of time exactly when they are being detected. I'd suggest to rather rely on the Effective Clock values as per: https://www.hwinfo.com/forum/threads/effective-clock-vs-instant-discrete-clock.5958/

Regarding the temperature on E4300 this is certainly due to a different assessment of Tj,max for that model. It cannot be precisely determined as described here: https://www.hwinfo.com/forum/threads/cpu-core-temperature-measuring-facts-fictions.148/
You can adjust the Tjmax to a value that you believe is reasonable by right-clicking on the DTS sensor.
Thank you for the information provided.
I did not know that core duo cpus had no reliable measurements thermal wise. That is pity, I need to use my core duo with windows 10 and that cpu overclocks hard i can push it at 2.4 without touching the voltage before not being able to boot or crashing, but if cannot get reliable temperatures the last thing I want is to burn the chip because it does not know what is his tj max :(
 

Martin

HWiNFO Author
Staff member
The CPU knows what its Tj,max is but it won't tell that :) So each software trying to report the actual temperature is guesstimating what the Tj,max might be (based on various sources), hence resulting in reporting different temperatures.
 
Top