![]() I cannot tell you how many times (bordering on a hundred?) I've heard people say "well XYZ app says my CPU is at 30C, but then ABC app says it's at 47C? Which is right?", when really XYZ is looking at DTS and averaging them, while ABC is looking at a HW monitoring IC, but both apps say "CPU Temperature". It's one reason why these generic Windows hardware monitoring applications are ridiculously stupid - they really don't teach this to end-users for reasons I just do not understand. Therefore: "sensor" can refer to one of several different things, with the source data coming from one of several different places. Then there's PECI, and ACPI thermal zones, and. And back on the DTS side, TjMax and all that horseshit makes things crazy. ![]() ![]() None of *that* discusses per-vendor implementation differences, such as when they choose to stick a resistor between the thermistor and the chip, changing the formulas you have to use (in software) to calculate accurate values (so what the chip manufacturer says to use for a formula is no longer accurate, and rarely if ever do the actual motherboard manufacturers public disclose such resistors/changes - it often varies per motherboard too). And that's all on-die - none of that is related to hardware monitoring ICs (sometimes part of a Super I/O chip) which provides data from external sources (particularly thermistors placed on the motherboard near/around components of interest). For example, Intel CPUs (both mobile and desktop) include DTSs, but where they choose to use a DTS (and what it associates with) is their choice model X may have a DTS per CPU core as well as 2 in the on-die GPU, but another model may have only 1 for the GPU. But really it doesn't matter, because it all boils down to choices the manufacturers make when it comes to designing the product. AMD and Intel do things very differently.
0 Comments
Leave a Reply. |