link626 wrote:
I'd be really surprised if they undervolted across the board, since not every cpu undervolts the same, and can cause instability.
Also, if it were really undervolted, hwinfo64 or similar apps would say so.
When you're loading the cpu 100% by video encoding, for example, it's going to run as hot as it can, and any Intel microcode optimization won't prevent it from running as hot as possible. This is where the fans come into play.
Spinning the fans as fast and loud as possible at 100% cpu load will bring temps down more.
The real test is to load up hwinfo64 and see what the max temps are under full load, and note if there is any throttling, and how long the cpu stays in full turbo boost mode.
The cpu alone won't throttle much, but when run simultaneously with gpu, could result in throttling.
If it were me, I would spend 15 minutes on each firmware back to back and run the same benchmarks and compare results.
You are right in some points:
- Fans will help, but the noise will increase significantly;
- Pushing the CPU to max alone does not bring throttling, but when coupled with load on the GPU, the temps increase and throttling happens.
But regarding undervolting, no, programs like that would not show.
If Asus undervolted e.g. -30mV, many will not feel a difference and I will risk saying every CPU will handle it.
Furthermore, this new voltage would be the reference, i.e. with 0mV offset.
I am not saying it is something an OEM would do, but it is possible.
At 100%, depending on the code, the CPU can handle thing differently.
Intel has found a bug on the code of Kaby-Lake chips. See here:
http://www.theregister.co.uk/2017/06/25/intel_skylake_kaby_lake_hyperthreading/I would not spend time making a downgrade on the BIOS firmware, especially not knowing what the modifications are.
