Images of pins for z370 & z270

[Mira Yurizaki]

  On 10/3/2017 at 8:33 PM, Bananasplit_00 said:

dosent change my conclusion that these extra pins are not needed, oh and Kabylake at stock sits in its power draw more or less smack dab on that TDP, and moveing forward  id expect the same from them.

Anandtech measures system power consumption.

 

Otherwise 90W at idle? Dafuq Intel?

[Bananasplit_00]

  On 10/3/2017 at 8:37 PM, M.Yurizaki said:

Anandtech measures system power consumption.

acording to them this was as acurate they could get for the CPU only, while messuring system power. calculateing for PSU losses, GPU, and rest of system, this 90W figure is what they put out. i dont think its entirely accurate but compared to the I7 4790K the Kabylake chip is clearly closer to its rated 90W figure

[AresKrieger]

  On 10/3/2017 at 8:33 PM, Bananasplit_00 said:

dosent change my conclusion that these extra pins are not needed, oh and Kabylake at stock sits in its power draw more or less smack dab on that TDP, and moveing forward  id expect the same from them.

Ok so look at your graph and you'll see that an i5 with a tdp of 91W has a draw of 68W and that the i7 has 90W so TDP only correlates with that chart (also that says idle which isn't really relevant) well the 8700k would be 1.5X the i7's number assuming power efficiency  remained the same, and given the clock speed increase it is doubtful the chips will use less power even if they are more efficient so I fail to see how your conclusion makes any sense, unless you assume intel managed to add two more cores without increasing power draw.

[Mira Yurizaki]

  On 10/3/2017 at 8:39 PM, Bananasplit_00 said:

acording to them this was as acurate they could get for the CPU only, while messuring system power. calculateing for PSU losses, GPU, and rest of system, this 90W figure is what they put out. i dont think its entirely accurate but compared to the I7 4790K the Kabylake chip is clearly closer to its rated 90W figure

However, you still can't really say that TDP as a "good enough" estimate of power consumption given that there are processors on that list aren't close to their TDP versus what they measured.

 

I mean they measured 110W for an 88W TDP part. How do we know this won't be the same thing on Coffee Lake? Not to mention that unless Intel has done something drastic to reduce per-core power consumption, it doesn't make sense that electrical power only increased 4W, if you wanted to go that route, on a14nm process. You're telling me Intel managed to get per-core power consumption down by 43% using the same process node. Even if they made refinements, I wouldn't expect more than say 10% or so.

 

Not to mention those pins are tiny in terms of gauging. Having any extra power concentrated in fewer pins isn't a good idea.

[AresKrieger]

Just for reference to everyone discussing whether this is necessary from a TDP point of view TDP is a metric of wasted power (in the form of heat) not used power, it is akin to the temperature gauge on a car engine, which doesn't tell you how much gasoline you're using nor does it tell you the horsepower of the engine, it just tells you how hot the block is getting.

[samcool55]

  On 10/3/2017 at 8:54 PM, AresKrieger said:

Just for reference to everyone discussing whether this is necessary from a TDP point of view TDP is a metric of wasted power (in the form of heat) not used power, it is akin to the temperature gauge on a car engine, which doesn't tell you how much gasoline you're using nor does it tell you the horsepower of the engine, it just tells you how hot the block is getting.

Thing is, with tech it's basically 100% electricity to heat.

With a car engine there's movement, air resistant and other things, with a cpu that's not the case. There are basically no variables compared to a car engine.

 

Also TDP is not a temperature thing, it's watts which is a form of energy...

If it would be a temperature thing it would be 80°C TDP which is not the case.

[Bananasplit_00]

  On 10/3/2017 at 8:43 PM, AresKrieger said:

Ok so look at your graph and you'll see that an i5 with a tdp of 91W has a draw of 68W and that the i7 has 90W so TDP only correlates with that chart (also that says idle which isn't really relevant) well the 8700k would be 1.5X the i7's number assuming power efficiency  remained the same, and given the clock speed increase it is doubtful the chips will use less power even if they are more efficient so I fail to see how your conclusion makes any sense, unless you assume intel managed to add two more cores without increasing power draw.

got a better number to use? i cant find any actiual CPU power draw figures that are any better or more reliable for Kabylake. 

  On 10/3/2017 at 8:48 PM, M.Yurizaki said:

However, you still can't really say that TDP as a "good enough" estimate of power consumption given that there are processors on that list aren't close to their TDP versus what they measured.

my point with that graph was that Intel was right on with their current CPUs, and were not before, therefor partialy legitamising my usage of the TDP number

  On 10/3/2017 at 8:48 PM, M.Yurizaki said:

I mean they measured 110W for an 88W TDP part. How do we know this won't be the same thing on Coffee Lake? Not to mention that unless Intel has done something drastic to reduce per-core power consumption, it doesn't make sense that electrical power only increased 4W, if you wanted to go that route, on a14nm process. You're telling me Intel managed to get per-core power consumption down by 43% using the same process node. Even if they made refinements, I wouldn't expect more than say 10% or so.

even if we add in those extra 50W power, lets just say 150W or something then for the 8700K, the socket would still hold that, you are pushing 1.17W per pin in the Kabylake socket which is, obviously a 50% increase in power per pin, going from 0.5A to 0.75A assumeing it keeps the 1.5V CPU input voltage(got a better number on that one? information was scarce). which really i can not see as being a problem. i run 2V input voltage, and if we assume the same power draw that would be 1W per pin unless im thinking about this wrong.

  On 10/3/2017 at 8:48 PM, M.Yurizaki said:

Not to mention those pins are tiny in terms of gauging. Having any extra power concentrated in fewer pins isn't a good idea.

its not that they are bad in any way, they are good in every way, exept for the fact that Coffee Lake would most likely run just fine on the current Kabylake LGA 1151 socket. it would be really nice if you could drop in Kabylake in a Z370 motherboard, which my guess is, would be compleatly fine functionally, the extra pinns added here are probably wired up to the FIVR on Kabylake aswell in just the same way

[zberry7]

  On 10/3/2017 at 9:16 PM, samcool55 said:

Thing is, with tech it's basically 100% electricity to heat.

With a car engine there's movement, air resistant and other things, with a cpu that's not the case. There are basically no variables compared to a car engine.

 

Also TDP is not a temperature thing, it's watts which is a form of energy...

If it would be a temperature thing it would be 80°C TDP which is not the case.

Tempature is a form of energy, do you know what TDP stands for?

 

THERMAL design power

 

No one in here actually was involved with the R&D of this chip, So I’m curious as to how everyone seems to be so sure these CPUs would run just fine on Z270.

 

They changed the pin layout, so obviously now, there is no cross compatibility, it’s not like the reserved pins on Z270 can magically start delivering power. And if anyone noticed, they didn’t just change the reserved pins.. there were pins that were in use on Z270 that they changed.

[Mira Yurizaki]

  On 10/3/2017 at 9:22 PM, zberry7 said:

Tempature is a form of energy, do you know what TDP stands for?

 

THERMAL design power

Thermal design power is talking about the amount of heat generated whereas temperature is the average amount of heat throughout the system. You can have a higher TDP rating but the same temperature as a lower TDP rated part, but it just means you need a better cooler to achieve the temperature.

[AresKrieger]

  On 10/3/2017 at 9:21 PM, Bananasplit_00 said:

got a better number to use? i cant find any actiual CPU power draw figures that are any better or more reliable for Kabylake. 

Wish I did as it is hard to prove a point to someone without them

 

  On 10/3/2017 at 9:16 PM, samcool55 said:

Thing is, with tech it's basically 100% electricity to heat.

With a car engine there's movement, air resistant and other things, with a cpu that's not the case. There are basically no variables compared to a car engine.

 

Also TDP is not a temperature thing, it's watts which is a form of energy...

If it would be a temperature thing it would be 80°C TDP which is not the case.

It's not 100%, heat (keep in mind I never said temperature when referring to tdp only in my car analogy) can be measured in watts as it is energy, the lack of variables doesn't change anything it simply makes it easier to calculate power use if you know the efficiency (which we don't) and that was merely an example to get a point across not a literal 1 to 1 comparison

 

Regardless another factor to consider is that increasing the number of power pins might allow for easier increase of clock speed, that depends on intel's design so I haven't a clue to what extent it could help but I'm certain it would as it would make it easier to feed the chip during operation.

[AnonymousGuy]

I was saying 3 months ago that the pinout was different to support DDR4-2666 and 2 more cores.  The name change from Z270->Z370 alone was done to indicate they wouldn't be compatible with each other.

 

Why is no one applying common sense and just hoping that compatibility is going to magically happen.

[zberry7]

  On 10/3/2017 at 9:27 PM, M.Yurizaki said:

Thermal design power is talking about the amount of heat generated whereas temperature is the average amount of heat throughout the system. You can have a higher TDP rating but the same temperature as a lower TDP rated part, but it just means you need a better cooler to achieve the temperature.

Yes. My point is TDP isn’t the same as power draw, just because we use watts as a way to measure it

 

[Mira Yurizaki]

  On 10/3/2017 at 9:36 PM, zberry7 said:

Yes. My point is TDP isn’t the same as power draw, it’s dissapated heat

It isn't the exact value, however every electronic component dissipates power as heat. The problem is everyone assigns blanket TDP values and even then, those TDP values are based on some arbitrary thing the manufacturer decided that they hope covers most use cases. Like you're going to thermal throttle an Intel processor using a stock heatsink if you feed it Prime95, because Intel doesn't think that's a realistic workload (and it really isn't).

[Lurick]

  On 10/3/2017 at 9:35 PM, AnonymousGuy said:

I was saying 3 months ago that the pinout was different to support DDR4-2666 and 2 more cores.  The name change from Z270->Z370 alone was done to indicate they wouldn't be compatible with each other.

 

Why is no one applying common sense and just hoping that compatibility is going to magically happen.

Because Intel is the devil and everything they do is pure evil, duh!

[TheRandomness]

Well, income the stickers. Just trace out the pins for power and split them evenly to the new pins.

[SquintyG33Rs]

  On 10/3/2017 at 6:20 PM, DocSwag said:

they increased number of pins for stuff like ground and vcore. Which makes sense as they need better power delivery for 6 core cpus.

that's not a factor there's a bunch of unused pins on the socket that can be assigned to that.

[Mira Yurizaki]

  On 10/3/2017 at 10:32 PM, SquintyG33Rs said:

that's not a factor there's a bunch of unused pins on the socket that can be assigned to that.

That's what they did.

 

The real question is then why was there all those unused pins in the first place?

[SquintyG33Rs]

  On 10/3/2017 at 10:34 PM, M.Yurizaki said:

That's what they did.

 

The real question is then why was there all those unused pins in the first place?

they leave overhead for future use. which is exactly what they're doing now which is also why people are complaining that their existing boards don't work

[goodtofufriday]

Couldn't a bios update turn those pins on? 

[Hunter259]

  On 10/3/2017 at 10:58 PM, goodtofufriday said:

Couldn't a bios update turn those pins on? 

Can't just "turn things on" if the traces were never created in the first place.

[DocSwag]

  On 10/3/2017 at 10:32 PM, SquintyG33Rs said:

that's not a factor there's a bunch of unused pins on the socket that can be assigned to that.

I'm afraid I don't see any unused pins... Correct me if wrong.

[AnonymousGuy]

  On 10/3/2017 at 11:03 PM, DocSwag said:

I'm afraid I don't see any unused pins... Correct me if wrong.

RSVD

[goodtofufriday]

  On 10/3/2017 at 11:01 PM, Hunter259 said:

Can't just "turn things on" if the traces were never created in the first place.

At least on AMD side they are usually connected just waiting to be used. 

[DocSwag]

  On 10/3/2017 at 11:23 PM, AnonymousGuy said:

RSVD

Does RSVD stand for reserved?

[Hunter259]

  On 10/3/2017 at 11:27 PM, goodtofufriday said:

At least on AMD side they are usually connected just waiting to be used. 

Only if the motherboard makers know that. It can be completely up to them whether or not to put traces on unused pins.