Discussion Intel Meteor, Arrow, Lunar & Panther Lakes Discussion Threads

[Tigerick]

As Hot Chips 34 starting this week, Intel will unveil technical information of upcoming Meteor Lake (MTL) and Arrow Lake (ARL), new generation platform after Raptor Lake. Both MTL and ARL represent new direction which Intel will move to multiple chiplets and combine as one SoC platform.

MTL also represents new compute tile that based on Intel 4 process which is based on EUV lithography, a first from Intel. Intel expects to ship MTL mobile SoC in 2023.

ARL will come after MTL so Intel should be shipping it in 2024, that is what Intel roadmap is telling us. ARL compute tile will be manufactured by Intel 20A process, a first from Intel to use GAA transistors called RibbonFET.



Comparison of upcoming Intel's U-series CPU: Core Ultra 100U, Lunar Lake and Panther Lake

Model	Code-Name	Date	TDP	Node	Tiles	Main Tile	CPU	LP E-Core	LLC	GPU	Xe-cores
Core Ultra 100U	Meteor Lake	Q4 2023	15 - 57 W	Intel 4 + N5 + N6	4	tCPU	2P + 8E	2	12 MB	Intel Graphics	4
?	Lunar Lake	Q4 2024	17 - 30 W	N3B + N6	2	CPU + GPU & IMC	4P + 4E	0	8 MB	Arc	8
?	Panther Lake	Q1 2026 ?	?	Intel 18A + N3E	3	CPU + MC	4P + 8E	4	?	Arc	12

Comparison of die size of Each Tile of Meteor Lake, Arrow Lake, Lunar Lake and Panther Lake

	Meteor Lake	Arrow Lake (20A)	Arrow Lake (N3B)	Arrow Lake Refresh (N3B)	Lunar Lake	Panther Lake
Platform	Mobile H/U Only	Desktop Only	Desktop & Mobile H&HX	Desktop Only	Mobile U Only	Mobile H
Process Node	Intel 4	Intel 20A	TSMC N3B	TSMC N3B	TSMC N3B	Intel 18A
Date	Q4 2023	Q1 2025 ?	Desktop-Q4-2024 H&HX-Q1-2025	Q4 2025 ?	Q4 2024	Q1 2026 ?
Full Die	6P + 8P	6P + 8E ?	8P + 16E	8P + 32E	4P + 4E	4P + 8E
LLC	24 MB	24 MB ?	36 MB ?	?	8 MB	?
tCPU	66.48
tGPU	44.45
SoC	96.77
IOE	44.45
Total	252.15

Intel Core Ultra 100 - Meteor Lake



As mentioned by Tomshardware, TSMC will manufacture the I/O, SoC, and GPU tiles. That means Intel will manufacture only the CPU and Foveros tiles. (Notably, Intel calls the I/O tile an 'I/O Expander,' hence the IOE moniker.)

 

[igor_kavinski]

Which law of electronics is this?

Effect of Temperature on Power-Consumption with the i7-2600K

Building on a previous thread that was directed more towards power-consumption scaling with CPU clockspeed, of the many things uncovered in that thread the one thing that just never sat right with me, or with many other members, was the impact of the operating temperature on power-consumption...

forums.anandtech.com

The reason why the power-consumption is increasing with temperature has to do entirely with the static leakage power consumption of the CPU (it is independent of clockspeed, solely dependent on temperature and voltage) and a phenomenon known in physics as the Poole-Frenkel effect.

Thank you for asking. So now I know about the physics behind it too.

 

[igor_kavinski]

By the way, temperature is also the reason extreme overclockers need to use LN2 to hit otherwise impossible frequencies. At subzero temperatures, leakage current is reduced, less power is wasted as heat so more of it can be used to drive higher frequencies.

 

[mikk]

Here is a Cinebench R23 result from an Ultra 7 155U, unfortunately they haven't logged the power usage.

 

[naukkis]

By the way, temperature is also the reason extreme overclockers need to use LN2 to hit otherwise impossible frequencies. At subzero temperatures, leakage current is reduced, less power is wasted as heat so more of it can be used to drive higher frequencies.

Actually higher leakage in cpu means higher clocking potential. Cpu manufacturers sometimes give golden overclocking samples away that are way too leakage to be sold. Metal resistance will grow with temperature so keeping silicon as cold as possible gives transistors more current to switch - basically rising their max operating speed.

 

[SiliconFly]

Which law of electronics is this?

Basic physics. Efficiency = work done/energy spent

Waste heat drastically reduces efficiency. This they usually teach in 8th grade.

 

[poke01]

More temperature would reduce the efficiency and waste more power. People were not too happy with Zen 4 bumping up against 95C. Apple engineers must be smoking something to let their CPU go above 100C. Either make a better, heftier thermal solution to sustain such temperatures or clamp down on them before they go out of hand.

I think those sensors are heat spots don’t mean much. Why? Because I had an M3 MacBook Pro machine to test and when it go to 100C, the back of the laptop was still a lot cooler than say a 12th gen Intel U series laptop. I seen Intel laptops that go to 90C that are much hotter on the backside.

The MacBook Air doesn’t go past 45C with regards to the surface temperature. It looks like Apple is making sure that to stay below 45C. I

 

[Hitman928]

Basic physics. Efficiency = work done/energy spent

Waste heat drastically reduces efficiency. This they usually teach in 8th grade.

Temperature isn’t 1:1 with waste heat though. You can have a super efficient processor that runs at 100 C and a horribly inefficient one that runs at 65 C. You have to take into account the effectiveness of the cooling solution and how much work is being done (i.e., how high of performance is being reached)

Basing a processor’s efficiency on its operating temperature makes no sense.

 

[coercitiv]

Basic physics. Efficiency = work done/energy spent

Waste heat drastically reduces efficiency. This they usually teach in 8th grade.

Except you weren't talking about waste heat with your original post, but rather package temperature (or hotspot temp), which is not exactly the same thing. Meanwhile @igor_kavinski was probably talking about the relation between chip temperature and leakage power, which is wasted energy that scales with temperature delta over ambient.

I think those sensors are heat spots don’t mean much. Why? Because I had an M3 MacBook Pro machine to test and when it go to 100C, the back of the laptop was still a lot cooler than say a 12th gen Intel U series laptop. I seen Intel laptops that go to 90C that are much hotter on the backside.

The MacBook Air doesn’t go past 45C with regards to the surface temperature. It looks like Apple is making sure that to stay below 45C. I

Indeed, modern laptops use external sensors in combination with package readings. Allowing the chip to reach very high temps has the disadvantage of increasing waste heat (see above) but the major advantage of increasing heat transfer. One could call it "cooling overclocking" as the cooling solution will dissipate more heat when the chip is hot, therefore allowing for longer burst performance while the chassis isn't warm enough to trigger throttling.

As an interesting finding, I noticed my Lenovo laptop become noticeably more aggressive in ramping the fans after installing their power management app (or updating the UEFI, not sure). At first I was annoyed, it seemed like such a random thing to do, even when CPU temps were very low. After more observations through trial and error, I noticed the fan was more likely to ramp up when battery temp was high, the app even has color coding for a battery temp sensor. It turns out they might be trying to keep the battery at low temps to prolong it's life. Interesting if true, though I never found out if this is the only reason for the change in cooling behavior.

 

[Doug S]

Temperature isn’t 1:1 with waste heat though. You can have a super efficient processor that runs at 100 C and a horribly inefficient one that runs at 65 C. You have to take into account the effectiveness of the cooling solution and how much work is being done (i.e., how high of performance is being reached)

Basing a processor’s efficiency on its operating temperature makes no sense.

The only useful measure of waste heat in a laptop is its power draw, because every watt the laptop (CPU, GPU, DRAM, NAND, display, etc) draws is emitted as heat.

A laptop drawing 5 watts could have a CPU hitting 110C if all that power is going into a single core, while a laptop drawing 50 watts might never exceed 50C depending on sensor placement and how much of the chip (other stuff like DRAM & NAND) is drawing lots of power.

Ideally you'd design it to disperse the heat evenly. That's not practical to do (i.e. having the display surface emit equal heat to the bottom of the laptop) but if you did you'd be able to draw more power without it seeming uncomfortably warm to the touch. Or warm at all, since basically anything under body temperature will be removing heat from your body not adding to it.

 

[SiliconFly]

Basing a processor’s efficiency on its operating temperature makes no sense.

It's very simple. As temperature increases, a processor's efficiency decreases. Cos' some of the power thats consumed by the cpu is converted into waste heat instead of getting work done.

In simple words, let us say a cpu is running at 40°C. When the cpu has more work to do, it speeds up which in turn draws more power. If the temperature increases to 100°C, some of the power that is drawn by the cpu is actually converted to waste heat (60°C in this case) instead of getting useful work done. This reduces efficiency.

Efficiency = work done/energy spent

In other words, for every watt the processor consumes, it's expected to get certain amount of work done. When the cpu is running slower, most of the energy consumed is actually used to get useful work done and very less energy is converted to waste heat. When it's running very fast, a lot more energy is converted to waste heat compared to the amount of energy used to get useful work done. More the heat, less efficiency.

 

[SiliconFly]

All this cos I posted a link saying a M3 MacBook Air runs hotter (at 114°C) compared to competition! 🤔

Maybe we should just call it MacBook Air Fryer or MacBook HotAir.

Just a thought.

 

[Hitman928]

It's very simple. As temperature increases, a processor's efficiency decreases. Cos' some of the power thats consumed by the cpu is converted into waste heat instead of getting work done.

In simple words, let us say a cpu is running at 40°C. When the cpu has more work to do, it speeds up which in turn draws more power. If the temperature increases to 100°C, some of the power that is drawn by the cpu is actually converted to waste heat (60°C in this case) instead of getting useful work done. This reduces efficiency.

Efficiency = work done/energy spent

In other words, for every watt the processor consumes, it's expected to get certain amount of work done. When the cpu is running slower, most of the energy consumed is actually used to get useful work done and very less energy is converted to waste heat. When it's running very fast, a lot more energy is converted to waste heat compared to the amount of energy used to get useful work done. More the heat, less efficiency.

No, you don’t understand how it works. You have several replies after your original post about the M3 temperature that explain it, I suggest you re read those replies.

Hint: two CPUs that consume the same amount of power are producing the same amount of waste heat, the amount of work they are doing makes no difference to how much waste heat is being produced.

 

[SiliconFly]

No, you don’t understand how it works...

Actually you don't. Where do you think the heat is coming from?

 

[DavidC1]

Actually, it is not completely unrealistic to extrapolate that if Skymont is 20% higher ST perf than Crestmont - which to me is a reasonable assumption - then Lion Cove will have to be roughly 25% higher ST perf than Redwood Cove. If Lunar Lake is indeed 1.5 times faster at CB R23 MT than MTL-U that is.

Lunarlake's Lion Cove P core does not have Hyperthreading.

Plus it says on the leaked slides that the "E cores are prioritized for lower priority tasks", and doesn't even share the L3 cache with the P cores so the P might have to do extra work.

One thing is they can improve greatly at the 15-20W mark where Meteorlake severely underperforms.

It doesn't have to be all about the architecture.
-Next gen process + TSMC's process more suited for lower power
-Delays on Meteorlake and sub-par tile implementation
-Improved uarch

Complex silicon chips have variability beyond the silicon itself. The unknown variable that we cannot calculate for exists in the capability of the team and it's management to carry out the goals of the project.

 

[poke01]

All this cos I posted a link saying a M3 MacBook Air runs hotter (at 114°C) compared to competition! 🤔

You seem to confusing CPU hotspots with the surface temps of a laptop. The M3 Macbook Air will be much cooler than any MTL laptop when resting on a lap/holding it, despite the "higher" CPU temps.

You can have a CPU that goes to 100C(in this case 114C) and uses only 22watts and on the other hand have an Intel CPU got to 90C but it will be hotter to handle because of more energy/watts it uses. I think MTL chips can go up to 65 watts and be under 100C. You cannot determine efficiency from temperature but from energy used.

We had the same song and dance with the M2 Airs and frankly Apple's throtting will kick in before any real damage happens. I have seen no reports of anyone getting burnt from a M chip.

 

[Hitman928]

Actually you don't. Where do you think the heat is coming from?

You’ve already got this information if you’d read through all the replies, but here’s a link to a free MIT lesson on the subject:

https://ocw.mit.edu/courses/6-884-complex-digital-systems-spring-2005/a5b7f72687b2d269fed647330d539f48_l11_power.pdf

Efficiency does not decrease all that significantly with increased temperature and there is no correlation between operating temperature and efficiency between different chips.

 

[maddie]

It's very simple. As temperature increases, a processor's efficiency decreases. Cos' some of the power thats consumed by the cpu is converted into waste heat instead of getting work done.

In simple words, let us say a cpu is running at 40°C. When the cpu has more work to do, it speeds up which in turn draws more power. If the temperature increases to 100°C, some of the power that is drawn by the cpu is actually converted to waste heat (60°C in this case) instead of getting useful work done. This reduces efficiency.

Efficiency = work done/energy spent

In other words, for every watt the processor consumes, it's expected to get certain amount of work done. When the cpu is running slower, most of the energy consumed is actually used to get useful work done and very less energy is converted to waste heat. When it's running very fast, a lot more energy is converted to waste heat compared to the amount of energy used to get useful work done. More the heat, less efficiency.

You always have wasted energy but the % increases as the temps increase. It's not simply that you waste more as more work is done, which by itself does not lead to less efficiency. Electron mobility and such.

 

[DavidC1]

We would still have to account for the OS.

The performance/watt doesn't matter as much as the battery life.

In the Kabylake generation, ARM platforms were roughly 50% ahead in battery life per WHr. This is the critical part they need to improve. They fell in efficiency with the post-Tigerlake parts.

They may never be able to reach MacOS levels of battery, but reaching WoA levels for browsing/video playback would be an important step, dispelling the widespread erroneous notion that ISA is somehow responsible for battery life differences(yet again).

It's very simple. As temperature increases, a processor's efficiency decreases. Cos' some of the power thats consumed by the cpu is converted into waste heat instead of getting work done.

What are you talking about? Processors are by definition 100% "waste heat". They are same as electrical heaters. The definition of "efficiency" for a CPU is performance per watt, or how well it performs per watt.

The M3 Pro absolutely KILLS it in that regard. It uses power like a "U" while performing like an H in both single and multi-thread. The gap is akin to Prescott Pentium 4 versus Pentium M,

-Except in this case the "Pentium M" also takes the performance lead.
-Except that the "Pentium M" is also beating the second best(AMD) in perf/watt by a huge margin too. Third obviously being Intel.

 

[igor_kavinski]

https://ocw.mit.edu/courses/6-884-complex-digital-systems-spring-2005/a5b7f72687b2d269fed647330d539f48_l11_power.pdf

What I was referring to but as pointed out in previous replies, the increased temp might be just at the particular hotspot being measured so maybe the entire chip isn't getting that hot.

 

[FlameTail]

Basic physics. Efficiency = work done/energy spent

Waste heat drastically reduces efficiency. This they usually teach in 8th grade.

I know that. No, my question was- what is reason efficiency decreases with increasing temperature. Igor Kavinski gave the answer.

 

[DavidC1]

FinFET CMOS has lot less leakage than in previous generations.

For a simple example my R9 290 mining system I could reduce power consumption down by 30W when decreasing the temperature from 95C to 70C.

On the "Polaris" RX 4xx/5xx GPUs, decreasing the temperature by 20C reduced power by less than 5W.

Intel results show that by 90nm, gate leakage was already becoming a significant factor, hence the introduction of High-K dielectric at 45nm which reduced gate leakage to almost nothing.

The SOI process in AMD processors brought benefits, but FD-SOI was the real kicker. But instead of FD-SOI, the industry went FinFET, which allowed most of the depleted channel benefits of FD-SOI while improving characteristics + performance vastly in other metrics while being easier to implement.

The final play in this journey is the Gate-All-Around, which encapsulates the channel on all four sides, versus FinFET, which does it for three, hence the "most of the benefits".

What I was referring to but as pointed out in previous replies, the increased temp might be just at the particular hotspot being measured so maybe the entire chip isn't getting that hot.

If people think the MBA reaching 115C is a problem with such an efficient CPU, I'd love to see how it performs if we replace it with an AMD one, or worse Intel H!

Apple may have screwed up on the laptop part, but the CPU is top notch.

 

[SiliconFly]

You seem to confusing CPU hotspots with the surface temps of a laptop.

Doesn't matter. Higher operating temperatures reduces efficiency. @igor_kavinski 's link spells it out clearly. Transistors get slower when hotter. Every transistor matters.

Efficiency does not decrease all that significantly with increased temperature and there is no correlation between operating temperature and efficiency between different chips.

Efficiency decreases & there is a very direct correlation. See above. It spells it out clearly.

You always have wasted energy but the % increases as the temps increase. It's not simply that you waste more as more work is done, which by itself does not lead to less efficiency. Electron mobility and such.

Your 1st statement & 2nd statement contradict, but I think I understand what you're trying to say.

Igor Kavinski gave the answer.

Yep. His answer was simple & elegant.

Apple may have screwed up on the laptop part...

Apple definitely screwed up the MacBoor Air M3 (running @ 114°C). It's a total piece of ****. Thats why I wanted to call it MacBook HotAir. And hence all this...

 

[coercitiv]

Your 1st statement & 2nd statement contradict, but I think I understand what you're trying to say.

They don't contradict. Leakage power is always there, but increases as chip temperature increases (even hot spots matter). More work does not directly imply more leakage power (waste), it can only lead to higher leakage if the cooling is not good enough. It's also the reason why leakage power matters little when a CPU is under very high load since dynamic power is usually overwhelmingly higher, whereas it may arguably be somewhat of a problem if the chip is under low load and running on battery. (though usually this is followed by a rapid drop in temps, and thus drop in leakage as well). This is why we have variable cooling in consumer products, we aim for a certain chip temperature and the cooling system (hardware and software) does it's best to stay within spec.

I really think you should stop lecturing folks about these things, and start reading. The Wikipedia page on processor power dissipation does a great job of presenting the important facts in a condensed format.

 

[SiliconFly]

They don't contradict. Leakage power is always there, but increases as chip temperature increases (even hot spots matter). More work does not directly imply more leakage power (waste), it can only lead to higher leakage if the cooling is not good enough. It's also the reason why leakage power matters little when a CPU is under very high load since dynamic power is usually overwhelmingly higher, whereas it may arguably be somewhat of a problem if the chip is under low load and running on battery. (though usually this is followed by a rapid drop in temps, and thus drop in leakage as well). This is why we have variable cooling in consumer products, we aim for a certain chip temperature and the cooling system (hardware and software) does it's best to stay within spec.

I really think you should stop lecturing folks about these things, and start reading. The Wikipedia page on processor power dissipation does a great job of presenting the important facts in a condensed format.

I'm not. You're the one lecturing now. I've done a fair bit of reading myself. Claiming high temperature doesn't affect performance is outright silly (just cos it's a Mac!). Some want to believe that even when the MacBook Air is running at 114°C, it's still efficient. And let them believe what they want, it's their choice. But I don't.

 

[coercitiv]

I'm not. You're the one lecturing now. I've done a fair bit of reading myself. Claiming high temperature doesn't affect performance is outright silly (just cos it's a Mac!). Some want to believe that even when the MacBook Air is running at 114°C, it's still efficient. And let them believe what they want, it's their choice. But I don't.

Several people in this thread have already tried to help you, yet you insist with the same misconceptions. Good luck digging yourself out of this.