benchmark thermal performance

[PeachGr]

Before few months, i swapped my AMD stock cooler to a bequiet dark rock slim cooler for my Ryzen 7 3800x
At first i was disapointed. The noice was good, basicaly emits no sound compare to the loud stock, but thermals...
I was expecting lower temps but for a few days, temprature of all cores (measured on AMD ruzen master) was at the same level (82-85C)
after a week or so i checked again. To my surprise i ve had 75-77C without any change to hardware or software.
I didn't know why but as i discovered recently, as i watched a Jayztwocents video, he said that thermalpaste works better after some time
Everything made sence to the point that i don't know why nobody else was talking about that.
So when a reviewer is testing thermals by swapping one cooler after another, isn't that questioning the reliability of the benchmark resaults?

[Bombastinator]

1 hour ago, PeachGr said:

Before few months, i swapped my AMD stock cooler to a bequiet dark rock slim cooler for my Ryzen 7 3800x
At first i was disapointed. The noice was good, basicaly emits no sound compare to the loud stock, but thermals...
I was expecting lower temps but for a few days, temprature of all cores (measured on AMD ruzen master) was at the same level (82-85C)
after a week or so i checked again. To my surprise i ve had 75-77C without any change to hardware or software.
I didn't know why but as i discovered recently, as i watched a Jayztwocents video, he said that thermalpaste works better after some time
Everything made sence to the point that i don't know why nobody else was talking about that.
So when a reviewer is testing thermals by swapping one cooler after another, isn't that questioning the reliability of the benchmark resaults?

No, because they’re comparing each to each other.  It would only not work if one was left on for a week but another was not. 

[PeachGr]

16 minutes ago, Bombastinator said:

No, because they’re comparing each to each other.  It would only not work if one was left on for a week but another was not. 

I get what you're saying, but I don't mean who's better than the next one. I mean as a buyer, if I buy one of them, I wanna know what temperature to expect

[Roswell]

54 minutes ago, PeachGr said:

I get what you're saying, but I don't mean who's better than the next one. I mean as a buyer, if I buy one of them, I wanna know what temperature to expect

That’s impossible to know because your ambient temperature will not be the same as their ambient temperature. Your ambient temperature will also change considerably throughout the year.

 

The only things that that you should be paying attention to in the reviews are delta over ambient temperature and noise levels.

[PeachGr]

1 minute ago, Vitamanic said:

That’s impossible to know because your ambient temperature will not be the same as their ambient temperature. Your ambient temperature will also change considerably throughout the year.

 

The only things that that you should be paying attention to in the reviews are delta over ambient temperature and noise levels.

that argument is what i hear a lot. Ambient temprature is +-5C.
People use Airconditioners, and the cpu temp of mine is the same +-1 all year.

[Roswell]

3 minutes ago, PeachGr said:

that argument is what i hear a lot. Ambient temprature is +-5C.
People use Airconditioners, and the cpu temp of mine is the same +-1 all year.

It’s not an argument, it’s fact. Many people also have more than a 5c variation in ambient temperatures, not everyone wastes that much money to keep their home/office at the same temperature 24/7 every day of the year.

 

The only solution to your issue is to contact reviewers and demand that they change their thermostat to the same setting as yours before they run their tests.

[Bombastinator]

It’s not even just the ambient temperature.  Chips vary a little bit between each other too.

 

the traces on chips are really really tiny.  As tiny as they can possibly be made.  This frequently means they vary a lot.  Some chips run really efficiently, some dont.  The concept is called the silicon lottery.  I’ve got an old 4770k that just isn’t very good. It’s got more variance in it than average.  The result is most people with 4770ks can overclock their chips to 4.5ghz or more.  Some with “golden chips” get 5ghz.  I can only get 4.0. It’s called getting a lead chip.  Sometimes it happens. Reviewers can’t account for that either.  They don’t know if the chip they use to test with is gold or lead or somewhere in between.  They’d have to test thousands of chips to know and they only get a one.  Or maybe a couple.  Not enough.  Looking at a review can tell you very roughly where it might be, but the only way to get an accurate number is to put it in your machine with your chip at your house and try it. 

Edited October 24, 2020 by Bombastinator

[PeachGr]

The point is that nobody mentions the difference of new, and a few days old thermal paste

[Bombastinator]

Just now, PeachGr said:

The point is that nobody mentions the difference of new, and a few days old thermal paste

They don’t have time.   They’re reviewers.  They build a machine, test it, take it apart and build another one. They’ve got lots of stuff to review. You seem to me to be saying that saying one paste is better than another when new isn’t accurate because one paste may age better than another. Normally it just doesn’t change that much.  There are exceptions. This is true of some pastes.  Liquid metal, (indium/gallium) for example is known to be corrosive and one has to be very very careful what one puts it on and even then it needs to be changed frequently. Most pastes are made of pretty similar stuff: silicon oil and some sort of very finely granulated filler. Graphite, silver, industrial diamonds, etc..  they’re all going to settle a bit over time.  Some pastes will dry and crack.  Some won’t.  
The problem is you seem to want more accuracy than is available.  It would be nice, but nobody has it.