Commander Kinsey wrote:
On Sat, 15 May 2021 08:07:15 +0100, Paul wrote:

Commander Kinsey wrote:
On Fri, 14 May 2021 15:05:45 +0100, Andy Burns
wrote:

Commander Kinsey wrote:

All product specs are lies. The CPU cooler in this computer claims it
will cool a 300W CPU. Mine gives off only 140W,

The TDP figures given by Intel and AMD may be given in "Watts" but they
don't correspond to actual watts of power consumed by the processor,
they're calculated mumbo-jumbo, then the cooler manufacturers base
their
recommendations on the CPU manufacturer numbers, so it's lies all the
way down ...

How hard can it be to quote the TDP as the amount of power the CPU uses
(which is the same as the amount of heat it gives off)?

It is used for engineering as well as public relations purposes.

What part of engineering requires a number that is meaningless to be
stated? It's neither the heat given off or the power used.

The coolers are "honest" in that their theta can be
verified by FrostyTech and they'd look bad if they
fudged the number.

What is happening today, is "industrialized overclocking".

Are you saying the TDP of my Ryzen is actually 105W if I forbid it to up
the clock? That's like selling a car that will go 100mph and only make
a noise of 50dB. But not telling you that you can't do both at once.
Actually I've seen that kind of bull****, I bought an HP (spit!) laser
printer which quoted pages per minute. It was referred to (hidden away
in the smallprint) as "blank page speed". If you printed blank pages,
it would achieve that speed. But a full page of text (what most people
print) was nowhere near as fast. What they'd done is tell you how fast
it could spin the roller.


There are a few web articles about all the power states now,
and all I can say with any certainty is "TDP applies to base clock".
It's intended as a long term number, where the cores drop
to base clock, because the all-cores turbo table doesn't
allow anything higher than base clock.

Some 3.7GHz processors, do 5.1GHz on one core, 5.0Ghz
on two cores. And on all-cores, the multiplier drops
back to the base clock value. And then there's a
remote chance that TDP is the power value at that
point.

But if you make power measurements, you'll be seeing
all sorts of values. The odds of observing 95W on
a 95W processor are pretty slim, and you're just
as likely to see it drawing 140W.

My processor isn't "ancient", but it also doesn't
have the latest turbo features, so the power
model for it is simpler. It seems to draw a bit
more than TDP, but not outrageously so. And that
would be TDP on all-cores at base clock (possibly
3.4GHz).

Paul