Question Laptop CPU Package Power under full load - Can someone enlighten me?

[qubeex]

I've been looking for an explanation all over the internet.

So basically there are two laptops with the same CPU (i7-8750h).

In a Cinebench run, both score pretty much the same. Both sustain 3.9 Ghz throughout the benchmark.

WHY one shows in HWinfo CPU Package Power: 70W and the other 45W?

I understand the power limit implementation of the OEM, but I don't understand HOW they can have the same scores / clock speeds?

I always thought that if the CPU Package Power draw is higher it means that the system is cooled better therefore the power limit has been set higher, therefore the performance will be better.

But how come in this case a 45W CPU is performing the same as a 70W CPU? Also, which one would be a better choice?

 

[ElFenix]

I'm guessing it's actually a bug in hardware info

 

[killster1]

i suppose the 70wvs 45w limit wouldn't matter if it hit its thermal limit 80C? 90C? 100C? before that anyway right? So a cooler slower cpu could actually be faster (in a laptop)

what does the intel extreme tuning thing say? have you tried lowering the offset? i think its fun to use .90v less but sometimes it doesnt change alot.

 

[qubeex]

During Cinebench they reach:
45 W = 85 degree
70 W = 80 degree

None of them hit the thermal throttling limit (90-95).

Weirdly enough, the 70W laptop is cooler during the benchmark. It is thicker and has better fans though.

But the scores are still the same.

XTU states the same TDP as HWinfo. They are at stock uv and offsets. I just want to understand why one consumes 70W and the other 45W, considering they are the exact same CPU, same core speed, same cb score. hmm

 

[killster1]

During Cinebench they reach:
45 W = 85 degree
70 W = 80 degree

None of them hit the thermal throttling limit (90-95).

Weirdly enough, the 70W laptop is cooler during the benchmark. It is thicker and has better fans though.

But the scores are still the same.

XTU states the same TDP as HWinfo. They are at stock uv and offsets. I just want to understand why one consumes 70W and the other 45W, considering they are the exact same CPU, same core speed, same cb score. hmm

sorry im reallllllly sleepy. but how can you tell how many watts its consuming? you have a watt meter plugged in? 70w could be thermal limit to 80? i can make all kinds of silly turbo wattage settings but it doesnt mean it will really use that amount.

 

[Zucker2k]

sorry im reallllllly sleepy. but how can you tell how many watts its consuming? you have a watt meter plugged in? 70w could be thermal limit to 80? i can make all kinds of silly turbo wattage settings but it doesnt mean it will really use that amount.

This. Having a higher Platform (Package) TDP setting doesn't mean the cpu is necessarily consuming as much.

 

[coercitiv]

I just want to understand why one consumes 70W and the other 45W, considering they are the exact same CPU, same core speed, same cb score. hmm

They're consuming roughly the same amount of power, you just convinced yourself the software is giving you the correct current package power reading, which it obviously isn't.

 

[qubeex]

I'm even more confused now...

So basically CPU Package Power, is not a representation of the current power draw of the CPU (cores, cache, and other uncore)?

It is just a number table specified by the bios to which Throttling limits to happen?

Say, CPU Temp reached the CPU Throttling Temp -> CPU Package Power to get lower? Together with the PL1 and PL2 limits, which are time based constraints.





But more importantly, can I decide based on this 45W vs 70W limits, which laptop performs better on the long run? Is one more 'efficient' than the other?

 

[Bouowmx]

If both processors are at same frequency (yes) for the duration of the benchmark, then power consumption won't affect performance. The 70-W has higher core voltage.

Cinebench R15 doesn't run long enough to test power limits (because Turbo Boost short power limit). Try Prime95.

Normally, a laptop with higher power limit is desirable, if cooling is capable

 

[DrMrLordX]

you have a watt meter plugged in?

Seconded. Use an external meter like a Kill-a-Watt and see what it's pulling from the wall.

 

[qubeex]

So here's a graph of what I am referring to.
Three laptops with the exact same CPU rendering the exact same scene.

As you can see the RED one is pulling almost 80W to run at 3.9 Ghz, whilst the green and blue is pulling 50-60W to run at the same speed. And this is what I don't understand.

- How can they run at the same clock speeds whilst pulling different CPU Power amounts? (especially in the first part of the render)

- Does it mean that the RED one has more cooling capacity therefore the PL1 is set higher? But why would you do that if it runs at the same clock speed?

Benchmark wise, RED finished first, then the BLUE, followed by the GREEN (which was power limit throttling at second 40).