Temperature Monitoring Utilities - What year/century are they in?

[britechguy]

Over the past 10 years, and particularly the last 5 years, I've noticed that virtually any temperature monitoring utility I'm familiar with has "not kept up with the times, or temperatures" and pretty much all of them consistently sound alarms when none are necessary.

One example (and it's only one) is SpeedFan. It sets the default "ideal" temperature for the cores on an i5 12th Gen at 40 degrees C and the alarm temperature at 50. Both of these figures are insanely low. At rest, complete rest, the core temps tend to be above 40 and with just the slightest bit of work they head up into the high 50s and low 60s. T-max for this processor is 100 degrees C. I wouldn't blink an eye until at least 85 degrees, if not 90, was reached, particularly under heavy load.

And for those that monitor SSD temperatures, they seem to expect those to be much lower than within normal operating limits for modern NVMe drives, too.

If you use any of these monitoring utilities, have you noticed the same thing? I routinely have to go in and adjust the temperatures up based on the spec sheets for the CPU/APU and/or SSD.

 

[Sky-Knight]

I just don't use this stuff anymore. I use the utility that comes with the main board if I use anything at all. Speedfan is fun... but it suffers from opensourceitis while not being open source. Yet I don't know of a solid solution that meets that reality either.

 

[britechguy]

Even the manufacturers don't seem willing to give anything approaching a "real world operating temperature range" when it comes to devices anymore.

The Silicon Power UD80 NVMe SSD I bought shows operating range (and I know this means *can* operate) of 0 - 70 degrees C. When you are doing anything, and I mean anything at all, with an ambient temperature of around 75 degrees F, the actual operating temperature is always right around 60 C, minimum, and when you put it under load it always is up in the high 60 degree range, but has never yet hit 70.

Realistically, normal actual operating temperature range for this thing should be noted as 55 to 70 degrees. And I suspect its Tmax is well above 70 degrees C.

 

[Sky-Knight]

And I don't like seeing anything get that hot no matter what. NVME lifespans plummet when you let them get >70C.

 

[Porthos]

I used copper heatsinks on every one I installed. https://www.amazon.com/gp/product/B07XD8M43P/ref=ppx_yo_dt_b_search_asin_title?ie=UTF8&psc=1

 

[cypress]

I haven't used SpeedFan in years. I wanna say about 10 years + what I am using right now is CPUID HWmonitor.

 

[britechguy]

I haven't used SpeedFan in years. I wanna say about 10 years + what I am using right now is CPUID HWmonitor.

They all seem to suffer from inaccurate alarm levels. When it comes to CPUs/APUs in particular, what was "unacceptably hot" in 1992 is solidly "within typical operating limits" now.

The alarm level (or whatever the utility might call it) for "hot alarms" is being given when these devices are nowhere near to their Tmax. Something running at 80 degrees C with a Tmax of 100 is NOT running hot. It's not even close to running hot. That's "just warm" for something so designed. Twenty degrees under Tmax is akin to the distance between NYC and Chicago - they're nowhere near to close.

Back when I was a moderator at BleepingComputer I was constantly having to deal with people having conniption fits about their computer "running hot" when it was not doing anything of the sort. And much of that came from these damned temperature monitoring utilities using figures for "ideal" and "warning" that were 10s of degrees too low.

I always "look to the specs" and I wish the folks that put together these utilities would, too. There is no generic ideal or warning temperature. Both are determined by the hardware in use.

 

[fincoder]

I haven't used SpeedFan in years. I wanna say about 10 years + what I am using right now is CPUID HWmonitor.

Exactly the same here. HWwonitor seems the most reliable, but I don't use any alarm features.

 

[britechguy]

but I don't use any alarm features.

To be clear, I don't mean literal alarms. But most of these things use little icons to indicate when they think the computer is "burnin' hot." And more than 9 times out of 10, the utility's assessment of "burnin' hot" is so far off the spec sheets, we're talking tens of degrees C below Tmax. That simply is not hot.

The device makers mean that Tmax is the top operating temperature where NO DAMAGE OR THROTTLING is necessary. That means it's within normal limits. Giving "hot" indication at anything that's not within 5 degrees C of that value is just ridiculous. It makes people very panicky for no reason whatsoever.

 

[NviGate Systems]

The hobbyist market for this is pretty much dead and little to no money for developing apps like this. Plus, many vendors simply don't give full details on temps.

For example, temps that do show are not on die temps, those are locked behind vendor "gates" or whatever proprietory system they have. If it's something that is negotiable, usually the vendor will have an app to let you set fan curve or see it. Only when money was involved, say with GPU Mining, did suddenly new apps spring up and new BIOS files etc. But for typical use, it's slim pickings like you said. I am currently running a low TDP CPU and Low TDP chips have a lower MAX Temp, but most utilities icorrectly list it as the typical 75/99 Intel Specs...so it goes both ways.