iMore's reply to Linus on why macs are slower than PCs

[jagdtigger]

Just now, GoldenLag said:

Intel TDP is mostly an issue due ti the fact mobo vendors tend to break it out of the box. 

We talk about laptops here, meaning they have the control over every bit of the HW including the cooler and the FW. If a laptop runs hot(like in this case) its the manufacturers fault and  nobody else...

[GoldenLag]

7 minutes ago, jagdtigger said:

We talk about laptops here, meaning they have the control over every bit of the HW including the cooler and the FW. If a laptop runs hot(like in this case) its the manufacturers fault and  nobody else...

If it goes below advertised base speed. 

 

As long as it runs within spec they should be fine

[jagdtigger]

1 minute ago, GoldenLag said:

If it goes below advertised base speed. 

 

As long as it runs within spec they should be fine

So its fine that it runs at 90+ as long as its above or at base...

[GoldenLag]

Just now, jagdtigger said:

So its fine that it runs at 90+ as long as its above or at base...

Technically yes. Its not super healthy, but its fine. 

 

Not that id be glad having a laptop behaving that way

[jagdtigger]

Just now, GoldenLag said:

echnically yes. Its not super healthy, but its fine. 

Not really fine, you wouldnt be able to use the laptop because of the hot chassy....

[Trixanity]

Intel is totally not fiddling with their TDP despite moving the goal posts to fit within arbitrary numbers. If TDP is pointless and arbitrary why not kill the thing. Only manufacturers need to know anyway and they should test their products to deliver a decent end result.

[Master Delta Chief]

4 hours ago, GoldenLag said:

Technically yes. Its not super healthy, but its fine. 

 

Not that id be glad having a laptop behaving that way

I wouldn't call that fine at all and many others can agree on that. It's basically become a portable heater (Good for in the winter btw) and that's not good at all. It is not comfortable using your laptop when it starts overheating. On the long term, your processor dies out simply due to the amount of heat it tried to withstand that it put out. Thermal design is always an important topic to ensure the durability is good enough so that the device can remain active for a long time. 

[Drak3]

1 hour ago, Master Delta Chief said:

It's basically become a portable heater

Regardless of cooling solution or TDP measuring method, this will be true. It's the job of a cooler to take heat away from the component being cooled and dissipate it inti the atmosphere.

[mr moose]

8 hours ago, GoldenLag said:

Yes the ammount of heat a cooler cab dissipate can be calculated accuratly to a high margin.

I know this.

Quote

The TDP of CPU is supposed to tell how much cooling capacity is needed. If the TDP isnt indicative of how much cooling is needed, then its a "false" description.

It is indicative though.   the description and criteria are clearly listed.  If cooling engineers and laptop engineers can work it out. The spec is not for consumers like yourself, in fact it's barely a spec for anyone beyond system designers.

Quote

People should allways look at reviews, but that doesnt excuse bad TDP rating

Again the spec isn't "bad", wanting it to be some thing different or cover areas of a CPU operation that are beyond the necessary operating scope for engineering doesn't make it a bad spec.  It is fully in line with how most of the industry spec out their silicon data sheets.

 

[GoldenLag]

27 minutes ago, mr moose said:

Again the spec isn't "bad"

Its "bad" if it doesnt describe reality. Like any scientific model. 

28 minutes ago, mr moose said:

It is indicative though.   the description and criteria are clearly listed.

If its able to indicate, then its descriptive. But if its not descriptive enough to be used directly. Then its not satisfactory. 

 

Intel TDP isnt great, but at least its somewhat accurate when you use components that dont deliberatly run out of the box spec. In terms of powerdraw the 95 watt CPUs do hold themself to 95 watts power consumption (roughly). 

[Dabombinable]

10 minutes ago, GoldenLag said:

Its "bad" if it doesnt describe reality. Like any scientific model. 

If its able to indicate, then its descriptive. But if its not descriptive enough to be used directly. Then its not satisfactory. 

 

Intel TDP isnt great, but at least its somewhat accurate when you use components that dont deliberatly run out of the box spec. In terms of powerdraw the 95 watt CPUs do hold themself to 95 watts power consumption (roughly). 

My 4790K holds itself to 88W (it fluctuates between 86-91W) when under a full load (including memory controller) when locked to 4GHz. With boost enabled, the package power climbs to 96W as it maintains its boost of 4.2GHz, even on a business class motherboard with the voltage locked to 1.2V max.
And it runs quite hot as well, though that'd be the thermal compound's fault as it has relatively poor thermal conductivity of 3.8W/(mK) - the cooler can dissipate up to 130W (full load at around 4.5-4.6GHz is around 120W).

[thorhammerz]

17 hours ago, jagdtigger said:

So its fine that it runs at 90+ as long as its above or at base...

> Base clock (or higher)
> Under TjMax (typically around 100 c)
Within spec.

 

Intel laptop (desktop ones too) CPUs will happily run in the 90 degrees centigrade range (just below Thermal Junction) for years on end (at least, until the warranty expires after 3-5 years) as long as you do not run them above spec (aka "overclocking"; forcing turbo boost multipliers to apply beyond power/thermal/time limits is technically a user-end overclock as well, for that matter).

 

If anything is going to die as a result of thermal fluctuations, the minimally-spec'ed motherboards are much more likely to kick the bucket first.

[Hellion]

Apple tools, gonna apple tool.

 

More to come at 5:00.

[Loote]

18 hours ago, jagdtigger said:

Not really fine, you wouldnt be able to use the laptop because of the hot chassy....

There are separate tests regarding chassy temps because depending on materials used, it may get hot or not. My GF has laptop on her knees while browsing, but when she games or uses Autocad, it's on a pad which makes for better cooling and only keyboard temps stay relevant.

That said, metal cases are pretty, but transfer heat to your legs easily.

[mr moose]

11 hours ago, GoldenLag said:

Its "bad" if it doesnt describe reality. Like any scientific model. 

If its able to indicate, then its descriptive. But if its not descriptive enough to be used directly. Then its not satisfactory. 

 

Intel TDP isnt great, but at least its somewhat accurate when you use components that dont deliberatly run out of the box spec. In terms of powerdraw the 95 watt CPUs do hold themself to 95 watts power consumption (roughly). 

It describes reality specifically and accurately that all the profession system designers out there to work with, it seems to be only forum enthusiasts who have an issue with it. 

 

 

[GoldenLag]

4 minutes ago, mr moose said:

It describes reality specifically and accurately that all the profession system designers out there to work with, it seems to be only forum enthusiasts who have an issue with it. 

intel TDP is fine once you realize that its for baseclock. though i still dislike their TDP rating, especially on Z series plattform where its instantly broken by anyone  with a competent VRM. 

 

not intel`s fault directly, but its rather deceptive in ways. 

[mr moose]

6 minutes ago, GoldenLag said:

intel TDP is fine once you realize that its for baseclock.

Anyone who wants to read and use the TDP spec should find out what it means, there isn't a single engineer on the planet who would look up a spec like TDP but not look up the specific criteria behind it.    As I said before, it is not really a spec for consumers.   

 

6 minutes ago, GoldenLag said:

though i still dislike their TDP rating, especially on Z series plattform where its instantly broken by anyone  with a competent VRM. 

 

not intel`s fault directly, but its rather deceptive in ways. 

Why is it deceptive though?  failing to read the data sheet on an IC then wondering why you keep burning them out is not the result of deceptive practices, it is the result of the user not educating themselves and guessing which parts they need.

[GoldenLag]

2 minutes ago, mr moose said:

Why is it deceptive though?  failing to read the data sheet on an IC then wondering why you keep burning them out is not the result of deceptive practices, it is the result of the user not educating themselves and guessing which parts they need.

its also ecouraged to break TDP spec using high end mobos, their own mobo will then "perform" better than the competition. 

 

the CPU isnt running according to spec while an average user is wondering why the CPU is overheating while having a 150 watts capable cooler on a 95 watt TDP CPU. 

[mr moose]

2 minutes ago, GoldenLag said:

its also ecouraged to break TDP spec using high end mobos, their own mobo will then "perform" better than the competition. 

 

the CPU isnt running according to spec while an average user is wondering why the CPU is overheating while having a 150 watts capable cooler on a 95 watt TDP CPU. 

What do you mean the CPU isn't running according to spec?   Why would you use a cooler only capable of dealing with 150Watts when the CPU can dissipate upto 300watts on full load all cores at max clock?   I'm not sure why the mistake of the users is a spec problem?

[GoldenLag]

1 minute ago, mr moose said:

What do you mean the CPU isn't running according to spec?   Why would you use a cooler only capable of dealing with 150Watts when the CPU can dissipate upto 300watts on full load all cores at max clock?   I'm not sure why the mistake of the users is a spec problem?

Because TDP. You should in theory only need a 95 watt cooler according to Intel themselves.

 

A 150w cooler is 50% more than you "need"

 

That is the problem with the spec and people breaking the spec

[mr moose]

2 minutes ago, GoldenLag said:

Because TDP. You should in theory only need a 95 watt cooler according to Intel themselves.

 

A 150w cooler is 50% more than you "need"

 

That is the problem with the spec and people breaking the spec

 

Are you being serious?  several posts ago you acknowledged that the TDP spec was for base clock, now you are trying to argue because it says 95Watt that that is the only size cooler you need. 

 

 

[GoldenLag]

2 minutes ago, mr moose said:

 

Are you being serious?  several posts ago you acknowledged that the TDP spec was for base clock, now you are trying to argue because it says 95Watt that that is the only size cooler you need. 

 

 

On paper it says "95 watt heat dissipation is required for this CPU".

 

I know its only for baseclock, because i live and breath tech.

 

So you are admitting to it not being a 95 watt TDP CPU and that it doesnt describe reality?

 

Because according to spec, a 95 watt TDP cooler is sufficient. 

[leadeater]

58 minutes ago, GoldenLag said:

On paper it says "95 watt heat dissipation is required for this CPU".

 

I know its only for baseclock, because i live and breath tech.

 

So you are admitting to it not being a 95 watt TDP CPU and that it doesnt describe reality?

 

Because according to spec, a 95 watt TDP cooler is sufficient. 

It's sufficient only if you want base clocks. If a laptop maker only wants to target that then that's all they need. That doesn't mean cooling solutions for laptops need to be 95W, they need to be what the manufacturer is targeting. It's not hard to know what the power usage per core is at different clock speeds and combinations of core utilization, package input current or VRM output current will tell you that plus I'm willing to bet there's more detailed specification supplied to system designers with that kind of information in it.

 

Problem is you can know all that information without knowing what the baseline is, I can know what the CPU will use at given points but where is the default point, the base? That's what the Intel TDP is. What the CPU will use under load thermally constrained at TjMax or Tcase Max depending on package type. It's a baseline, it's all it's been for ages.

 

Edit:

Also the boost wattages are settable and viewable and the duration of each. All the data is there to make an adequate cooling solution, all that information is not the product sheet TDP.

[Dabombinable]

Here it is. This is how a CPU (with a cooler that is capable of working well above Intel's TDP rating - earlier revisions were rated at 130W TDP) behaves over the course of a day with a wide variety of uses. The mobo is rather simple+cheap, and goes as far as to have parallel and serial ports:

Spoiler

 

[mr moose]

11 hours ago, GoldenLag said:

On paper it says "95 watt heat dissipation is required for this CPU".

 

I know its only for baseclock, because i live and breath tech.

 

So you are admitting to it not being a 95 watt TDP CPU and that it doesnt describe reality?

 

Because according to spec, a 95 watt TDP cooler is sufficient. 

 

After this discussion you are being disingenuous about TDP.  It's "Thermal" design power, not "total" design power.

 

I am saying that if you put a 95watt cooler on an Intel CPU with a TDP of 95watts you the CPU will run almost full load at base clock maintaining TjMAX.  If you want to know what size cooler you need for boost clocks and power settings then you'll have to do some research and maybe even some math.

 

But as I said before, it's much easier to look at power draw benchmarks and reviews than it is to go that route, even for enthusiasts.