[SOLVED] What Is The Best CPU Temp Monitor for my Ryzen 7 1700

[DeathHadArrived]

What Is The Best CPU Temp Monitor for my Ryzen 7 1700

 

[DMAN999]

Ryzen Master:
https://www.amd.com/en/technologies/ryzen-master
or
HWInfo:
https://www.hwinfo.com/download/

 

[DeathHadArrived]

wait up is HWInfo good or not

 

[CountMike]

wait up is HWInfo good or not

Is.

 

[mamasan2000]

Hwinfo64 for sure.
I can't remember the name of the other ones, maybe Coretemp? Because I've never used them. Can't trust them at all when it comes to Ryzen. Reddit is full of posts about those monitoring programs that fail to read correctly. Overclocking subreddit.

 

[drea.drechsler]

What Is The Best CPU Temp Monitor for my Ryzen 7 1700

Solid vote for HWInfo. It gets frequent updates for new processors and improved readings, has a lot of sensors none others do (it even shows Ryzen sleep state for each core) as well as a very complete system information tree. And temp readings include both the instantaneous CPU temp for each die (good for multi-die CPU's) as well as an 'average' temp for true thermal state of the processor.

Ryzenmaster reports only the average package temp, good to know the true thermal state of the processor but leaves you wondering why the fans pulse so much as it ignores instantaneous local temp when a core boosts. Ryzenmaster's true purpose is for overclockers and that's what AMD has said. That's also why it overclocks after boot-up as most processors have a 'cold bug' in that they won't boot up at an overclocked frequency (which BIOS overclocking requires) when below 0 degrees C.

HWMonitor (and maybe Coretemp) might be good if I ever figured out which of the sensor readings meant what as they obviously aren't labeled right. But since the only way to figure that out would be to calibrate against HWinfo's readings it kind of becomes absurd to even try...just use HWInfo.

 

[DeathHadArrived]

I am using HWinfo and it is not telling me the temperatures of the individual die on the Ryzen chip.

 

[punkncat]

I have not been able to use HW monitor reliably with Ryzen.
Core Temp

 

[drea.drechsler]

I am using HWinfo and it is not telling me the temperatures of the individual die on the Ryzen chip.

Assuming you've the right version of HWInfo you only see it with 3000 CPU's as they use chiplets. 1700's are monolithic die.

 

[Karadjgne]

HWInfo64 : Yes.

Not to be confused with HWMonitor, which is a resounding : Absolutely Not.

Just have to understand that software is software and runs as it will, not as you expect it to.

For instance, HWInfo reads single temps periodically. It's not a constant flow. Ryzen master reads temps and averages every 3 seconds. Both are accurate, but can be different.

Let's say over a 6 second period, the cpu actually had 50-60-70-40-50-50. HWInfo reads every 2 seconds, so you'd see a 60, 40, 50. You'd not see the others, including the spike to 70 unless running a graph. Ryzen master would show 2 readings, 60 as an average of the first 3 seconds, 47 as an average of the second 3 seconds. So 60-40-50 vs 60-47. And that can confuse people as the temps are somewhat different. Yet both are accurate.

It's all on the software, the time periods sampled, the time run. AMD doesn't really care about a single spike to 70, that's temporary and expected behavior during workloads, but an averaged reading will give a better idea where the cpu really is all around. Other software like HWInfo, is more accurate in exactness, but can be a little misleading because it is exact, but only temporarily, a spike not being the whole sum of the cpu.

With HWMonitor, can't even be sure it's the cpu being read, or if it's a single hottest core, how average temps are assumed or if it's not an anomoly and confusing the PCH with core temp. It's just very unreliable.

 

[CountMike]

In HWinfo it's possible to set polling interval to much shorter, I set mine to 500 ms,

 

[Karadjgne]

Yep. It's nice that that's possible, but that was a hypothetical bunch of numbers meant to convey the principle, not fact. And polling and reporting times also differ, if you had to try and read temps changing 2x a second, the number would look just like digital '88' blur.

So there's a lot of 'give and take', but just like thermal margins, the exact number is rarely important, what's important is what the number represents. I could care less if my cpu was 52 or 57, it's in the 50s and well below throttle, and it was the same last week. If tomorrow it was in the 60s, my cooler needs cleaning. If it hit the 80s, time to repaste. But it'd not make a difference if it was 81 or 89, it's in the 80s and supposed to be in the 50s.

What's best about HWInfo is its reliability more than its accuracy. It's constantly updated, tweaked, bug checked, the author working with Amd engineers to get the best possible results, every single poll. So it it says it's in the 50's, I can believe it's in the 50s. If it spikes to 90, I had a spike to 90 and it wasn't some funky anomoly. And I can trust that cpu, pch, vrm etc are what they are, with tmpin#1-4 there's no guarantee whatsoever, they changed with different motherboards.

 

[drea.drechsler]

....
What's best about HWInfo is its reliability more than its accuracy. It's constantly updated, tweaked, bug checked, the author working with Amd engineers to get the best possible results, every single poll.
...

I haven't used HWInfo for quite some time, but when I did I was always dismayed when confronted by a bunch of sensor readings with labels that made no sense. I had to 'calibrate' against HWInfo to decipher what something like "TmpSense1", for instance, meant.

Took me a bit, but I got to the point it was like...why not just use the reference? And it just got better when HWInfo added things like an 'average' temp reading that better shows the Ryzen thermal state. And a C-State Residency readout for each core that gives you a good idea of when the processor is dropping into 'deep sleep'...or not. Those are just some of the truely useful readouts not normally available.