This time i have something that should be more interesting!
First, some info:
- TEMP: Temperature in Celsius
- C0% (AFAIK):
- C states are indicators of utilisation
- There’s C0 - C10
- C0 is CPU doing mucho work, C10 is nacho work.
- POWER: CPU Package power draw in watts
- MULTI: Simply, MULTI x 100 = Ghz, i.e. 24 Multi means 2.4Ghz.
If you compare the benchmark graph above with the other graphs, you’ll realise they have nothing in common - the benchmark has my CPU pinned at 100% whereas in all other scenarios its spiky as hell.
I thought that we might be able to make more useful observations if i smoothed out the graphs a bit, and that the process in itself would be a pretty good “real world work” scenario to examine, so i logged the data for this afternoon, where (among other things) i worked on that.
TBH it should be quite obvious where i actually spent time on it and where i was kind of just derping around - see where the C0% line is consistently pretty high - at the beginning of P1 and toward the end of P2.
I eventually decided on using a moving average which perhaps is not the best option, but its what i know and i ain’t going learn stuff just to draw graphs.
Calculating a moving average is a pain in the rear I/O ports, usually pretty heavy on the CPU (especially when you’re really just pretending to know how to computer; efficiency = lol) - again, another good example of a heavy “real world work” scenario - so i set throttlestop to capture extra logs (8/sec instead of 1/sec) and let it run on a dataset of about 50000 records.
I stopped it after about 30 minutes - i’ll let it finish later - but i think what i got is still pretty interesting:
So what does this all mean? Nothing that we weren’t already told, actually; what this says to me is:
if CPU utilisation is not 100% (i.e. it isn’t already drawing a constant 7W), the CPU will still boost to above 7W (within RAPL constraints) to get higher speeds
The i7 V is 100% able to sustain operation at 7W, maintaining a max of ~80C.
For all interested, these were the ambient conditions:
- google says ambient temperature is 28C
- Singapore is humid as hell, feels hotter than that
- V sitting on the sofa, no particular care taken with regard to cooling
So i’d say that this is actual proof, reinforcing whats already been said: