I have some questions that may have been asked but my searched have come up empty.
Everyone knows (or should) that a CPU that gets too hot will start to throttle its frequency of operation to get a temperature back under control, my questions stem from this basic understanding.
1. In a multi core CPU is one core gets too hot does ( and you have frequency set per core) does the single hot core downclock or are all cores affected?
2. What time period (e.g. 1ms, 5ms 100 or 500ms) has to elapse at that trigger temp (e.g. 100C) before the downclock is initiated by the CPU protection?
3. Asked another way if I saw a spike of e.g. 102C within a polling period of 500ms but gone within 1000ms in the monitoring software would that trigger a momentary throttling?
I have no understanding of the intricacies of CPU throttling (other than the universal wish to avoid its causes at all costs) and any answers to these questions from a noob would be much appreciated?
Thank you for bringing this to my attention, let me help you on this matter.
I understand you have some questions about the processor heat generated when starting to throttle.
I would like to know if you have one processor model in mind. Please run the following https://downloadcenter.intel.com/download/25293/Intel-System-Support-Utility-for-Windows- tool and reply attaching the results. The utility will provide full details about your hardware configuration.
Thank you for your time.
I am using an i9-7900x and under load I see momentary spikes to over 100c on 2-3 cores that do not last past 1 polling cycle ( software poll is set down to 500ms) curious as to whether such momentary spike would trigger a throttling of the core if they hit the 105c temp. The spikes always seem to disappear within two polls. I would like to understand better how the throttling is activated and deactivated over a time curve. Average temps are low as the system has a 420 radiator looking after it.
Thank you for the information.
TDP is a power dissipation and component temperature operation condition, this sometimes exceeds for short periods of time or if running high power workload. I would recommend checking the following document on pages 80-99.