Ryzen isn't Intel. Even if results are similar, the way amd does things has been inherently different to Intel for more years than I can remember (which is a lot).
For Intel, they work by dropping voltages and frequency at idle, all cores remain active, and core temp access is every 256ms on all cores, the single core temp being either the hottest temp read or a set core, depending on the software used. This leaves pc's at idle speeds at @ 10°C ± from ambient.
Amd, of course, is different. Voltages and corresponding frequency aren't dropped to save power, much, instead parts of or entire cores are shutdown. So where Intel is all cores running low, amd has all but 1 core basically off. With any background task running, that puts the entire load on that single core, at its operating voltages, and you get a single core running at 20-30°C above ambient.
Which naturally freaks ppl out when idle temps are in the 50's, not realizing it's only one core, because they are indoctrinated with Intel temps being low, on all cores.
Because of the way amd does this, taking almost instant readings of cores would be semi-pointless, temps would fluctuate like an equalizer bars readout. So instead Ryzen Master uses an average, taken from the last few readings, still totalling less than a second. This will put RM at odds with other software as each has its own polling time, some reading every second, some reading every half second etc, so visible temps might be different by a good margin. Say a core goes from 30 to 50 and back to 40, RM would read that as 40, the average. Another software might poll on the half second, and catch the 50. Another might poll on the full second and catch that final 40. All 3 software are accurate, as far as that goes, but the half-second poll isn't as realistic as temps in a core can vary drastically depending on the load.
It's far more important to see what the cpu is doing overall, not so much exactly. The issue facing most software isn't the sensors, isn't the polling times, it's the reader. The value you see has to be put on screen, and left there long enough to actually read it, if temps were announced every read, the numbers would blur into digital 88. Which puts software showing highest reads in any given time period, and if the cores were reading all 30's and 40's and one read of 50, you get the impression the core really is at 50, when in reality it spent the last 2-3 seconds at 30-40. Which puts RM as a far more realistic approach than most, regardless of actual accuracy.
Facebook
Twitter
Instagram