News Intel Arrow Lake and Panther Lake CPU power profiles allegedly surfaced — leak details Intel Baseline, Performance, and Extreme profiles for next-g...

[Admin]

Arrow Lake-S, Arrow Lake mobile, and Panther Lake Intel power profiles have allegedly been uncovered, featuring big changes compared to Raptor Lake.

Intel Arrow Lake and Panther Lake CPU power profiles allegedly surfaced — leak details Intel Baseline, Performance, and Extreme profiles for next-g... : Read more

 

[KyaraM]

The Extreme profile features a 125W PL1 rating, 295W PL2 rating and 400W IccMAX amperage limit.

I think you mean 400A here, not 400W, right? And in the next paragraph, PL2 is certainly not raised to 25W, but rather to 250W...

Anyways, baseline and Performance sound fine to me. Extreme is pretty heavy on the power draw, but I guess it's at least honest if these limits are enforced. Will be interesting to see how they affect performance. My guess is that Performance will be the most sensible balance here, but we will have to see how much these chips will actually consume as well, since as we know that can differ.

 

[Lavadog321]

400 amps??? My entire house only has 40 amp service from the power company. Is this real? What am I misunderstanding?

 

[KyaraM]

400 amps??? My entire house only has 40 amp service from the power company. Is this real? What am I misunderstanding?

You are missing that it's your PSU's job to convert the current and amperage from your power plug into the values your CPU can withstand. Or do you think your CPU gets the full 110V or whatever it is in the US from the wall plug?

I (A) = P (W) / V (V)

This is the formula. In easy terms, Ampere is Watts divided by Volts. If your CPU pulls 295W with 1.5V, then the Amperage is just short of 200A, at ca. 196,7A. At 1V, it would be 295A. It feels like the limit is simply set higher so that it won't conflict with the rest of the stats, maybe allow for loss or whatever. I'm not an electrical engineer, though, so others here can likely explain more comprehensively. But the short version is, your PSU converts V and A received from the network into what your components can take.

 

[Kamen Rider Blade]

I thought Intel was trying to reduce power draw, not just use the head room to push more power through?

 

[baboma]

>Anyways, baseline and Performance sound fine to me. Extreme is pretty heavy on the power draw, but I guess it's at least honest if these limits are enforced.

'Extreme' only shows up for the top part, which makes sense for the best-perf-bar-none crowd. When you're playing top dollar for top perf, power efficiency isn't exactly top of mind.

That, and benchmark wins are still needed for bragging rights. Like it or not, it's essential for marketing. Enthusiasts and fanboys alike tend to fixate on the flagship winning or losing.

Anyway, the mere presence of an "Intel Default Setting" should suffice to bring an end to OEMs' unlimited power defaults, especially in light of the fallout of the instability issue. Intel should've done this a long time ago.

>but we will have to see how much these chips will actually consume as well, since as we know that can differ.

Interesting to note that short-duration power is more finely gradated to PL2/3/4.

I've no doubt ARL will consume less, if only because PL1 will now be 125W (or 65W) vs unlimited as before.

Regardless, on desktop, performance takes precedence over power efficiency. I don't see people complaining about 4090's 450W power draw, or 5090's reported 500W.

On that tangential note,

https://wccftech.com/intel-arc-batt...e2-gpu-cores-12-gb-192-bit-bus-19-gbps-memory

 

[SunMaster]

I think you mean 400A here, not 400W, right? And in the next paragraph, PL2 is certainly not raised to 25W, but rather to 250W...

PL4 is the max the chip can get before shutting down.

 

[Neilbob]

Regardless, on desktop, performance takes precedence over power efficiency. I don't see people complaining about 4090's 450W power draw, or 5090's reported 500W.

Not for everyone. Particularly when the so-called high performance can typically only be observed via benchmark numbers, I'd far rather go for efficiency all day long, especially here in the UK (and many other countries as I understand it) where electricity prices are kind of obnoxious. The majority of the time it's impossible to tell the difference.

As for the 4090 etc, people don't complain about that because they are purposefully buying it knowing the top-tier FPS comes at the cost of power (including melting power connectors. Is that still a thing amidst all this Intel nonsense?)

Personally, even if I could justify splurging such a huge sum of money just for gaming, I wouldn't specifically because I don't want anything in my PC guzzling power like that. Same with this CPU; I'd purposefully go for the lowest power setting. I kind of hate that the Extreme, heavy-power slurping setting is even still an option.

 

[KyaraM]

PL4 is the max the chip can get before shutting down.

???
What are you even talking about? I'm talking about this graph from the article:
https://cdn.mos.cms.futurecdn.net/55TKSW5k9yeFGcPRcvs5Wa-970-80.jpg
No mention of PL4, no mention of 400W anywhere, and to boot, my comment spans errorenous text about two different chips, divided by paragraphs. I'm not sure what you even want to say.

 

[SunMaster]

I think you mean 400A here, not 400W, right? And in the next paragraph, PL2 is certainly not raised to 25W, but rather to 250W...

I was referring to this. PL4 is 400w, which to my understanding is the same as Iccmax. Not 400A.

 

[Guardians Bane]

You are missing that it's your PSU's job to convert the current and amperage from your power plug into the values your CPU can withstand. Or do you think your CPU gets the full 110V or whatever it is in the US from the wall plug?

I (A) = P (W) / V (V)

This is the formula. In easy terms, Ampere is Watts divided by Volts. If your CPU pulls 295W with 1.5V, then the Amperage is just short of 200A, at ca. 196,7A. At 1V, it would be 295A. It feels like the limit is simply set higher so that it won't conflict with the rest of the stats, maybe allow for loss or whatever. I'm not an electrical engineer, though, so others here can likely explain more comprehensively. But the short version is, your PSU converts V and A received from the network into what your components can take.

Thanks for that quick lesson on electrical engineering! I had to learn all that back in the early 90's for electrical work on the AH-64 Apache when I was in the army. I haven't used that knowledge since the 2000's and forgot most of it!!!! Lol ... Anyway, good looking out.

 

[thestryker]

I'm hoping that if Intel is dumping PL1==PL2 that means they've got a lot more confidence in the architecture overall and especially the scheduling. Personally speaking I don't particularly care about maximum power consumption but the thing I kept going back to with 12th+ is that they could be really efficient, but you couldn't really configure them to act that way across the board easily. This is one place AMD has absolutely nailed when it comes to Zen 4/5 (aside from some idle power issues, but that's a whole other thing).

 

[Guardians Bane]

>>on desktop, performance takes precedence over power efficiency.

>Not for everyone.

Generalizations, such as above, are for the main, not for the entirety. The above isn't universally true, but it is true enough. To wit, the Ryzen 9K's public shellacking in reviews. Its power efficiency can't compensate for the weak performance.

>Personally, even if I could justify splurging such a huge sum of money just for gaming

Everybody has different expectations, different budgets, different use cases. If bang/buck is high on your list, then ARL won't make the cut, or Ryzen 9K for that matter.

Even moreso than Ryzen, ARL will be an entirely new platform, with new motherboards that will come at a premium. It will have glitches that all new platforms have. It's fun and exciting to beeswax about, but the value sweet spot will always be on the previous gen's markdowns.

I'm not currently in the market for another desktop, having built one recently. But if I were, it will almost certainly be a Raptor Lake, which I expect to see substantial markdowns in the coming months. Being an enthusiast means I'm fully apprised of its minimal risk, and bargain hunting always require ignoring the prevailing "Intel sux" groupthink and going your own way.

There's nothing stopping you from getting a RPL part and running it at 65W. That's actually what I did. For me, Intel's lack of power limits is an advantage, not a liability.

I absolutely agree on your last gen comment. For me I am going Ryzen though. Microcenter has a bunch of good bundle deals for Ryzen 7000 with mobo, RAM, and CPUs. Those last gen prices from both Intel and AMD are slowly getting better and better! Honestly I was hoping 9000 would perform. But we'll, it came out and I've been underwhelmed. Price to performance is def last gen from both.

 

[Amdlova]

People looking at max power if you guys don't know the alder lake or raptor lake can idle at 2w on desktop. I use the igpu yo save anothers 10 w from the gtx 4060ti 16gb

You can use your intel profile or adjust what your need on the fly with xtu.

 

[Guardians Bane]

>I'm hoping that if Intel is dumping PL1==PL2 that means they've got a lot more confidence in the architecture overall and especially the scheduling.

It all boils down to performance. ARL needs a significant perf improvement over RPLR, or it'll face the same reception Ryzen 9K did, regardless of whatever architectural improvement.

More than Ryz 9K, ARL not only need to reduce power, but also compensate for forgoing HT. If it can get decent gain while achieving both, ARL would well deserve the win for this gen.

Lunar Lake's 8C/8T performance will be the bellwether for ARL. We'll know about that on Sep 3.


>For me I am going Ryzen though. Microcenter has a bunch of good bundle deals for Ryzen 7000 with mobo, RAM, and CPUs.

Yeah, Ryzen 7K is fine as well. I favor RPL for its productivity edge, but ultimately it doesn't really matter which you buy. Both Intel & AMD parts are comparable. They both have similar bang/buck.

But you may be able to stick with your first choice, a Strix Point miniPC. Supposedly such is in the works.

https://videocardz.com/newz/amd-strix-point-expected-to-debut-in-october-claims-aoostar

Yeah, agree on your reply. Both do offer some good perks on each platform. I've just used Intel for so long, and my 6900hx is going damn good, I decided to go all out on my next build. BUT I still plan on checking out the Strix based minis for sure! Aoostar has already said they have a Strix mini coming out in either Sept or Oct (I forget which. Probably Oct). What I'm really excited for is a Halo based mini! Since I'll have a full on DT with dGPU for all the really heavy stuff and games, I'll be able to really compare the 890m to DT. Now that's going to be a fun test!

 

[MXM0]

Since these chips are produced on the first BSP node, 20A, I expect that the power requirements will change radically. The required power should be less, for the same perf, and, as a side effect, the heat less, which leaves more room for thermals, and more room for higher power profiles.

TBH the high power profiles are always just for people who want to tinker -- they don't yield enough extra perf to be worthwhile... at least not on previous Ryzen and Raptor Lake chips. Running at 100' C is just looking for trouble.

 

[KyaraM]

I was referring to this. PL4 is 400w, which to my understanding is the same as Iccmax. Not 400A.

The graph says something completely different. But whatever.

 

[emv]

So simpler questions: Is this for the TSMC version or Intel version of the CPU process. certainly 20A with backside power and GAA must be different than TSMC N3 for power?

 

[Lavadog321]

You are missing that it's your PSU's job to convert the current and amperage from your power plug into the values your CPU can withstand. Or do you think your CPU gets the full 110V or whatever it is in the US from the wall plug?

I (A) = P (W) / V (V)

This is the formula. In easy terms, Ampere is Watts divided by Volts. If your CPU pulls 295W with 1.5V, then the Amperage is just short of 200A, at ca. 196,7A. At 1V, it would be 295A. It feels like the limit is simply set higher so that it won't conflict with the rest of the stats, maybe allow for loss or whatever. I'm not an electrical engineer, though, so others here can likely explain more comprehensively. But the short version is, your PSU converts V and A received from the network into what your components can take.

THANK YOU! I appreciate that you took the time to give me a great answer.

 

[thestryker]

So simpler questions: Is this for the TSMC version or Intel version of the CPU process. certainly 20A with backside power and GAA must be different than TSMC N3 for power?

While Intel hasn't come out and confirmed anything yet all of the ARL compute tiles should be on 20A.

 

[emv]

While Intel hasn't come out and confirmed anything yet all of the ARL compute tiles should be on 20A

So you haven't seen the info showing that Intel moved most SKUs to TSMC N3? Check the teardowns on samples or just ask Intel. It might be important to you. its also big part of Intels restructuring

 

[thestryker]

So you haven't seen the info showing that Intel moved most SKUs to TSMC N3? Check the teardowns on samples or just ask Intel. It might be important to you. its also big part of Intels restructuring

There is zero evidence supporting N3 being used for the compute tile in ARL. If you have some feel free to share it.

 

[emv]

There is zero evidence supporting N3 being used for the compute tile in ARL. If you have some feel free to share it.

Oops sorry, ... I'm new to this site..... Oh wait Tom's Hardware is the one who publicly reported it. LOL.

Sounds like that is NOT the info you wanted to see. I will let you wait for the teardowns and Intel announcements on the roadmap change... I think they are scheduled soon ......

Ignore my comments ..... 20A is on track for Arrow Lake in 2024 and Intel is ramping the technology now. Yup.... thats it .... I was confused ....

 

[thestryker]

Oops sorry, ... I'm new to this site..... Oh wait Tom's Hardware is the one who publicly reported it. LOL.

Sounds like that is NOT the info you wanted to see. I will let you wait for the teardowns and Intel announcements on the roadmap change... I think they are scheduled soon ......

Ignore my comments ..... 20A is on track for Arrow Lake in 2024 and Intel is ramping the technology now. Yup.... thats it .... I was confused ....

Feel free to show this supposed reporting by Tom's as you seem like you're just trolling.

There was some leak out of China that was mistranslated that was only talking about certain mobile SKUs. Otherwise there's no evidence at all to support your claims.

 

[emv]

Feel free to show this supposed reporting by Tom's as you seem like you're just trolling.

There was some leak out of China that was mistranslated that was only talking about certain mobile SKUs. Otherwise there's no evidence at all to support your claims.

Your response is noted, thanks