True nature of E-cores and how effective are they?

[AusWolf]

None of which contradicts what I said. As I said twice "Designed to target a max of 95c, assuming other boost variables allow as I stated earlier".



They also are CPU cooler reviews and not a analysis of how easy a given chip is to cool. None of the prior metentioned variables or questions I mentioned earlier are answers by CPU cooler reviews



What are you talking about, the u12a can adequately cool every Zen 4 chip. To say is can't do it for ANY zen 4 chip is obviously false.



People, specifically you, don't seem to understand that there's no precice defintion of what "easier to cool" means. There's no official metric to define your subjective opinion.

I define "easier to cool" as
1. Allowing the CPU to use more power while running at the same temperature with the same cooler, or
2. The CPU needing a less capable cooler to operate with the same power consumed at the same temperature.

In this regard, larger, less dense, monolithic dies are better. My 7700X runs at 91-92 °C with a 280 mm AIO in Cinebench multi-core at its 142 W PPT limit. An Intel CPU limited to 142 W wouldn't even be close in temperature with the same cooler.

This is not to say that Intel CPUs are better. They're just better at transferring heat, which shouldn't be a question, imo.

 

[oxrufiioxo]

I don't understand why amd/intel fanboys are fighting over the thermal characteristics of two completely different architectures on two completely different processes nodes it's pretty much useless.

Also comparing user temps is pretty much useless my last intel chip the 9900k was notoriously difficult to cool yet I had no issues keeping it under 80c at 5ghz like 60 somthing c while gaming same with my 5800X some people struggle to keep it under 90c mine never exceeds 75c... There are way too many variables from ambient temp to case used to quality of the fans to motherboard voltage differences so any temps I'm getting are pretty much useless to anyone else unless they have identical hardware and are using the exact cpu not just the same model due to tolerance and silicon lottery. They also likely have to be in the same room or at the very least the room would need to be climate controlled.

A good example is at stock my 5800X runs 5c cooler under full load in my X570 Crosshair 8 hero vs my Aorus Master while scoring within margin of error all other things being equal.

All that matter is how low can you go on a cooler on a given processor while getting the maximum performance out of the box anyone buying a 7950X or 13900k can afford decent cooling and anyone going with a 13500 or 7700/7600 doesn't need to.

It isn't super hard these days to look at a graph see how a given cpu performs see what cooler was used to get a general idea of what you need to cool the cpu.

Who cares what is easier to cool people buy the performance they need and work the rest out themselves based on a lot more variables than what cooler a cpu requires life goes on.

While I don't condone this a user can litterally buy 4-5 coolers within a specific budget on amazon and keep what's best people need to stop trying to overcomplicate somthing in the name of trying to make their own purchasing decision look better nobody cares. Except fanboys they love to fabricate stuff in their own minds to try and make somthing they've purchased look better smh.

 

[JustBenching]

I define "easier to cool" as
1. Allowing the CPU to use more power while running at the same temperature with the same cooler, or
2. The CPU needing a less capable cooler to operate with the same power consumed at the same temperature.

In this regard, larger, less dense, monolithic dies are better. My 7700X runs at 91-92 °C with a 280 mm AIO in Cinebench multi-core at its 142 W PPT limit. An Intel CPU limited to 142 W wouldn't even be close in temperature with the same cooler.

This is not to say that Intel CPUs are better. They're just better at transferring heat, which shouldn't be a question, imo.

Just tested it out of curiosity, 12900k @ 140 watts on a u12a sits just below 60c (57c to 59c)

 

[AusWolf]

I don't understand why amd/intel fanboys are fighting over the thermal characteristics of two completely different architectures on two completely different processes nodes it's pretty much useless.

Also comparing user temps is pretty much useless my last intel chip the 9900k was notoriously difficult to cool yet I had no issues keeping it under 80c at 5ghz like 60 somthing c while gaming same with my 5800X some people struggle to keep it under 90c mine never exceeds 75c... There are way too many variables from ambient temp to case used to quality of the fans to motherboard voltage differences so any temps I'm getting are pretty much useless to anyone else unless they have identical hardware and are using the exact cpu not just the same model due to tolerance and silicon lottery. They also likely have to be in the same room or at the very least the room would need to be climate controlled.

A good example is at stock my 5800X runs 5c cooler under full load in my X570 Crosshair 8 hero vs my Aorus Master while scoring within margin of error all other things being equal.

All that matter is how low can you go on a cooler on a given processor while getting the maximum performance out of the box anyone buying a 7950X or 13900k can afford decent cooling and anyone going with a 13500 or 7700/7600 doesn't need to.

It isn't super hard these days to look at a graph see how a given cpu performs see what cooler was used to get a general idea of what you need to cool the cpu.

Who cares what is easier to cool people buy the performance they need and work the rest out themselves based on a lot more variables than what cooler a cpu requires life goes on.

While I don't condone this a user can litterally buy 4-5 coolers within a specific budget on amazon and keep what's best people need to stop trying to overcomplicate somthing in the name of trying to make their own purchasing decision look better nobody cares. Except fanboys they love to fabricate stuff in their own minds to try and make somthing they've purchased look better smh.

I can't speak for others, but I'm not a fan of any company. I'm arguing over the definitions of heat transfer and heat density, not over which manufacturer is better.

Sure, you can buy a 7700X or 7950X and run it at 95 °C with an el cheapo cooler (or a 13900K at 100 °C as a matter of fact), sacrificing some of its potential performance, there's nothing wrong with that. All I'm saying is that the fact that it's OK to run a modern CPU at its temperature limit because it was designed that way doesn't mean that it's "easy to cool".

I love my 7700X, but I'd never say that it's easy to cool. Easy to run at 95 °C, sure, but if you want all the performance it's got to offer, you'll have to do better and give it as beefy a cooler as you can, while you don't necessarily have to do that with an equal Intel part.

 

[oxrufiioxo]

I can't speak for others, but I'm not a fan of any company. I'm arguing over the definitions of heat transfer and heat density, not over which manufacturer is better.

Sure, you can buy a 7700X or 7950X and run it at 95 °C with an el cheapo cooler (or a 13900K at 100 °C as a matter of fact), sacrificing some of its potential performance, there's nothing wrong with that. All I'm saying is that the fact that it's OK to run a modern CPU at its temperature limit because it was designed that way doesn't mean that it's "easy to cool".

I love my 7700X, but I'd never say that it's easy to cool. Easy to run at 95 °C, sure, but if you want all the performance it's got to offer, you'll have to do better and give it as beefy a cooler as you can.

Fair points, but most are just using temperatures as an excuse to fuel their fanboy make believe war over two companies that don't give two $h!+s about you and just want your money lol.

 

[Vario]

Just tested it out of curiosity, 12900k @ 140 watts on a u12a sits just below 60c (57c to 59c)

I believe it, from my experience, 100C+ is attained at 280+ watts on 12900KS/NHD15S.

 

[TheoneandonlyMrK]

E core's, remember them sigh

 

[JustBenching]

I believe it, from my experience, 100C+ is attained at 280+ watts on 12900KS/NHD15S.

Yeah, around the 280w is where my U12A hits a wall with the 12900k

 

[OkieDan]

Pretty sure (from reading or watching Zen 4 reviews in the past) Intel CPUs hitting 100 C causes thermal throttling whereas AMD CPUs hitting 95 C doesn't throttle, as reflected in HWInfo. Also the Intel throttling caused frame time inconsistencies that were not apparent when Zen 4 hit 95 C.

 

[AusWolf]

Pretty sure (from reading or watching Zen 4 reviews in the past) Intel CPUs hitting 100 C causes thermal throttling whereas AMD CPUs hitting 95 C doesn't throttle, as reflected in HWInfo. Also the Intel throttling caused frame time inconsistencies that were not apparent when Zen 4 hit 95 C.

I think that comes from the different definitions of "thermal throttling".

Intel CPUs work with set boost stages (x.8 GHz with 8 cores, x.4 with 6 cores, x.2 with 4 cores loaded, etc.). If the CPU cannot achieve that for thermal reasons, it'll report thermal throttling.

AMD CPUs only have one base and one maximum boost clock, and their speed varies in between due to several factors (voltage, load, thermals, silicon lottery, etc.). When the CPU clocks lower due to one or more factors, it's only called "dynamic adjustment", not thermal throttling. Therefore, AMD CPUs are a lot better at handling these variables, while Intel CPUs need to be either allowed to run to the max, or constrained by power limits or clock speeds.

On Intel, you'll see maximum boost until you hit a wall and your frame rates dip. Your goal as a system builder is to eliminate that wall by a quality motherboard, good enough cooling, etc. On AMD, you'll see real time adjustments to whatever your circumstances are. If your cooling isn't good enough, the CPU won't even try to boost too high to begin with, so there's no frame rate dip, just a slightly lower overall performance.

 

[Mussels]

Meaning the cooler is Not dissipating well 140w.

Not correct at all.

These CPU's only run hot because the heat never reaches the cooler

Other people covered this already, but this is the problem with getting used to how one brand works: You see high temps on intel and it's always because the cooler cant handle the wattage.


So you see a high temperature anywhere else and suddenly - that false logic appears of "the cooling isnt good enough"

AM4 and AM5 have heat density as the problem where you need really good contact from the CPU die to the IHS, then the IHS to the heatsink (including a proper thermal paste application) - you get the same temps on a 120mm air cooler as you do a custom water loop and it has *nothing* to do with the capacity of the cooling

 

[AusWolf]

Not correct at all.

These CPU's only run hot because the heat never reaches the cooler

Other people covered this already, but this is the problem with getting used to how one brand works: You see high temps on intel and it's always because the cooler cant handle the wattage.


So you see a high temperature anywhere else and suddenly - that false logic appears of "the cooling isnt good enough"

AM4 and AM5 have heat density as the problem where you need really good contact from the CPU die to the IHS, then the IHS to the heatsink (including a proper thermal paste application) - you get the same temps on a 120mm air cooler as you do a custom water loop and it has *nothing* to do with the capacity of the cooling

Well said. If one still doesn't get it, I recommend looking at the first post in my build log - link in my signature.

 

[OkieDan]

The IHS has a bit to do with it as well, but it doesn't even make sense to go direct die on Zen 4 because there's very little performance to gain going past 220-230W where we can already get beyond without custom loops. I've often wondered why AMD didn't use a thicker PCB so the IHS could be thinner, seems like a valid option.

 

[Deleted member 185158]

Not correct at all.

These CPU's only run hot because the heat never reaches the cooler

Other people covered this already, but this is the problem with getting used to how one brand works: You see high temps on intel and it's always because the cooler cant handle the wattage.


So you see a high temperature anywhere else and suddenly - that false logic appears of "the cooling isnt good enough"

AM4 and AM5 have heat density as the problem where you need really good contact from the CPU die to the IHS, then the IHS to the heatsink (including a proper thermal paste application) - you get the same temps on a 120mm air cooler as you do a custom water loop and it has *nothing* to do with the capacity of the cooling

Uh huh, maybe in your house. With your environment.

My place I can run any cpu I want down to -195c.

And everything in-between.

What was that you where saying about cooling again?

Not correct at all.

These CPU's only run hot because the heat never reaches the cooler

Other people covered this already, but this is the problem with getting used to how one brand works: You see high temps on intel and it's always because the cooler cant handle the wattage.


So you see a high temperature anywhere else and suddenly - that false logic appears of "the cooling isnt good enough"

AM4 and AM5 have heat density as the problem where you need really good contact from the CPU die to the IHS, then the IHS to the heatsink (including a proper thermal paste application) - you get the same temps on a 120mm air cooler as you do a custom water loop and it has *nothing* to do with the capacity of the cooling

I suppose a more serious reply.

This same conversation for years.
Temp limits haven't changed much.
Top chips are hot.

That means top chips give even custom loops a challenge.

Users today have better access to obtain lower temps than ever before. IE: de-lid tools, motherboard flex brackets and so forth.

Temps are under better control today then any other period of time since PGA and lga became a thing after slotted chips.

The problem is actually understanding all this isn't a new conversation for me, cause we had it at 14nm 32nm 45nm 65nm 90nm 130nm and so forth.

And yet we still pursue max performance. So we buy the 140w chips. And they run hot.

No, they run within their temperature envelope.

We just want the better IPC. Fuck the thermals, that aint news here. These processors don't run 95c with games using 4 cores. 8 cores..... out of 16/32 available threads. And STILL within the thermal envelope.

The difference is today the chips overclock for you. Then you tweak single digit gains and complain about heat.

Yeah, 5800x has more transistors than a 3800X. It's hotter as a result, but operates within a similar temp envelope. You just get more bang for your buck = IPC.

What temps? 90c since first gen Ryzen. It's a big deal suddenly? 100c since first gen i7. Same old same old.

 

[AusWolf]

This same conversation for years.
Temp limits haven't changed much.
Top chips are hot.

That means top chips give even custom loops a challenge.

Users today have better access to obtain lower temps than ever before. IE: de-lid tools, motherboard flex brackets and so forth.

Temps are under better control today then any other period of time since PGA and lga became a thing after slotted chips.

The problem is actually understanding all this isn't a new conversation for me, cause we had it at 14nm 32nm 45nm 65nm 90nm 130nm and so forth.

And yet we still pursue max performance. So we buy the 140w chips. And they run hot.

No, they run within their temperature envelope.

We just want the better IPC. Fuck the thermals, that aint news here. These processors don't run 95c with games using 4 cores. 8 cores..... out of 16/32 available threads. And STILL within the thermal envelope.

The difference is today the chips overclock for you. Then you tweak single digit gains and complain about heat.

Yeah, 5800x has more transistors than a 3800X. It's hotter as a result, but operates within a similar temp envelope. You just get more bang for your buck = IPC.

What temps? 90c since first gen Ryzen. It's a big deal suddenly? 100c since first gen i7. Same old same old.

You're right in your points, but that doesn't make Mussels' point wrong. Heat transfer does differ when you compare Intel and AMD due to the density and location of compute dies, IHS design, etc. Sure, every CPU is pushed to the max these days, but you can correct that with sensible power limits and/or undervolting, PBO tuning and such. You will still see a difference between an X Watt AMD and Intel CPU, though.

 

[Deleted member 185158]

You're right in your points, but that doesn't make Mussels' point wrong. Heat transfer does differ when you compare Intel and AMD due to the density and location of compute dies, IHS design, etc. Sure, every CPU is pushed to the max these days, but you can correct that with sensible power limits and/or undervolting, PBO tuning and such.

Yes, I get all that.

I have a Morgan Silver dollar lapped both sides I've used many times instead of a copper IHS plate..

It comes down to which one of you guys has put forth this kind of effort cooling a processor.

I ran successfully a 2700X at 4ghz 1.190v after a delid under a peltier 12715.

Shall we discuss transistor leakage and why people have a decent experience with "under-volt" as well? Because the cooler temps reduce leakage. This is not an entire thread Mussles himself did create?

Oh I get it. No worries. I just don't think it all has to be that mission critical, so to speak.

 

[Mussels]

Deleted member didn't like being disagreed with.
Obviously if you run sub-ambient cooling you'll still get lower temperatures, just like you would from cooler ambients.

It's got nothing to do with how much money you've thrown at things with a peltier and all about basic science - and it's also irrelevant to the thread and intels E-cores