Intel "Panther Lake" Targets Substantial AI Performance Leap in 2025

[Easo]

We could go on all day, and tomorrow too, discussing what is real AI and what isn't. I'd say - just remember Arthur C. Clarke's third law. Any sufficiently advanced technology is indistinguishable from magic.

Not in this case, it's literally "just" math. I really blame the hype on media and partially the general lack of IT education for the masses.

 

[A&P211]

Indeed much like crypto, VR, AR, and curved phone screens were the next big thing.

We are lightyears away from AI. Its just the next move to generate money.

As long has Nvidia stock keeps increasing, I"m more than happy. It funds my early retirement. And my dad said the stock market are for rich people.

 

[Dr. Dro]

I hope the AI fad will have died out by 2025. I couldn't care less about it. As for AMD vs. Intel part of the question, here's a proposal: ARM. Break the mold a little.

Panther Lake (top tier, KS rendition of the Core 9 Ultra) seems like an interesting upgrade from where I'm currently standing. But after around 9 months of ownership, I don't think I have even come close to taxing my i9-13900KS yet.

 

[lexluthermiester]

Nobody cares about your "AI" performance.

True!

We care about the fact that your architecture is ancient, its IPC is shit, and your chips are clocked way outside their comfort zone to compensate.

Not true..

BTW, I did not vote on the poll as the option "None of the above" was not offered!

 

[TumbleGeorge]

I hope the AI fad will have died out by 2025

Some of the greats once said that "hope dies last". Yes, the AI fad may pass, but it will remain routine. The first and possibly secondary trainings of gpt-5, which model Altman mentioned would be a milestone, are probably underway now.

 

[Noyand]

Nobody cares about your "AI" performance. We care about the fact that your architecture is ancient, its IPC is shit, and your chips are clocked way outside their comfort zone to compensate.

You can complain about the power consumption, but not the IPC. if RPL IPC is shit, then Zen 4 IPC is even shittier. RPL isn't a derivative of coffee lake

I hope the AI fad will have died out by 2025. I couldn't care less about it. As for AMD vs. Intel part of the question, here's a proposal: ARM. Break the mold a little.

Panther Lake (top tier, KS rendition of the Core 9 Ultra) seems like an interesting upgrade from where I'm currently standing. But after around 9 months of ownership, I don't think I have even come close to taxing my i9-13900KS yet.

The marketing around "A.I" might become less prevalent, but I doubt that we'll get back to the old chip design. Neural engine and the likes are all about efficiency, and this is becoming a nice thing to have since there's more and more ML based functionality in various software. It will become like a video decoder: you technically don't need one, your CPU can decode a video just fine, but it's not efficient at all at doing so.

 

[Assimilator]

You can complain about the power consumption, but not the IPC. if RPL IPC is shit, then Zen 4 IPC is even shittier. RPL isn't a derivative of coffee lake
View attachment 332003

A single synthetic benchmark doesn't invalidate my argument. Try harder.

 

[lexluthermiester]

A single synthetic benchmark doesn't invalidate my argument. Try harder.

No, but it does cast a very big shadow of doubt on it. Also, Cinebench R20 is not a synthetic benchmark.

 

[phints]

It's not at all clear to me how "AI performance" (what that can actually be used for I have no idea) matters to an end user.

Regarding the poll it means nothing, Intel chips are a moving target regarding fabs and features, we don't actually know what's going to ship and when.

Whichever Intel CPU will offer 3-5x the performance per watt gaming from their joke of a 14th gen, on a new process like Intel 4 or Intel 20A will be interesting, since then it might actually compete with AMD and Apple.

 

[lexluthermiester]

It's not at all clear to me how "AI performance" (what that can actually be used for I have no idea) matters to an end user.

There could be some uses, but it's being VERY over hyped.

 

[efikkan]

No, but it does cast a very big shadow of doubt on it. Also, Cinebench R20 is not a synthetic benchmark.

To all three of you; Cinebench is certainly a real benchmark, a benchmark of the rendering engine of Cinema 4D, a very niche professional tool which most in here haven't even heard of.
So Cinebench should be considered as just that, one specific narrow benchmark, and never be used (alone) to extrapolate generic performance.

It can be used to approximate IPC, but then in a mix with ~20 or so other benchmarks, preferably a mix of workload types, and obviously workloads that don't run into other significant bottlenecks at the set clock speed, plus the CPUs must be completely locked at a clock speed far below any throttling. Another avenue is to use synthetic benchmarks for this purpose (a mix which is not just computationally intensive int and float operations, but also stresses the CPU front-end), but in either case there is always debate over which balance is the most fair.

Ultimately, the closest thing we have is a good broad range of CPU loads (not GPU loads like games), take the average of those within a couple of standard deviations to eliminate the outliers.

 

[Noyand]

A single synthetic benchmark doesn't invalidate my argument. Try harder.

Try harder to what ? Keeping you grounded in your criticism ? You act as if AMD still got a massive clock deficit in ST, when the 13700k and 7700x are both rated at 5.4GHZ in ST, yet the I7 still manage to beat the Ryzen 7 in gaming, Photoshop, and After effects with the "shit IPC". AMD even boost beyond 5.4Ghz frequency by default. Intel IPC is only bad if you compare it to Apple, but nobody in X86 land is close to match the M3 IPC.

 

[lexluthermiester]

Intel IPC is only bad if you compare it to Apple, but nobody in X86 land is close to match the M3 IPC.

That's a load of moose muffins. M3 is good but it does not make X86/X64 look bad. Don't make mountains out of mole-hills.

 

[Noyand]

That's a load of moose muffins. M3 is good but it does not make X86/X64 look bad. Don't make mountains out of mole-hills.

I mean the M3 is super limited when it comes to frequency, so it even out, but that's literally the only CPU who can make RPL IPC look bad if you were to limit it to 4Ghz. If you were strictly only looking at IPC. Sorry if it looked like that I said that x86 is generally bad, this isn't what I meant to say. Otherwise, I don't understand where RPL IPC being shit can possibly come from. Some people seem to think that AMD still got a full Ghz deficit

 

[lexluthermiester]

I mean the M3 is super limited when it comes to frequency, so it even out, but that's literally the only CPU who can make RPL IPC look bad if you were to limit it to 4Ghz. If you were strictly only looking at IPC. Sorry if it looked like that I said that x86 is generally bad, this isn't what I meant to say. Otherwise, I don't understand where RPL IPC being shit can possibly come from. Some people seem to think that AMD still got a full Ghz deficit

Here's the thing. Depending on the arch, at matching clock frequency(4ghz for example) on a per core basis, Apple's M3 is a bit ahead of Zen3, but not Zen4, a bit ahead of AlderLake but breaks even with RaptorLake. That statement is very dependent on the task in question, but those are the general base line comparisons. Credit where it's due, Apple has made a great SOC platform. That does not make it in anyway better than X86/X64. Just on par and competitive.

(No I don't have citations, I can't remember where I saw the numbers, if you really need to see it, Google)

 

[qcmadness]

Try harder to what ? Keeping you grounded in your criticism ? You act as if AMD still got a massive clock deficit in ST, when the 13700k and 7700x are both rated at 5.4GHZ in ST, yet the I7 still manage to beat the Ryzen 7 in gaming, Photoshop, and After effects with the "shit IPC". AMD even boost beyond 5.4Ghz frequency by default. Intel IPC is only bad if you compare it to Apple, but nobody in X86 land is close to match the M3 IPC.
View attachment 332047View attachment 332048
View attachment 332049View attachment 332050

At the cost of double transistor budget and triple power consumption?

 

[Jism]

I find the whole word AI hidden inside CPU's just another marketing argument for PC vendors towards people who don't know what type of system to get.

With the old pat in lead, you get the old dog tricks again. "But Intel has AI in it's chips!"

 

[Wirko]

I mean the M3 is super limited when it comes to frequency, so it even out, but that's literally the only CPU who can make RPL IPC look bad if you were to limit it to 4Ghz. If you were strictly only looking at IPC. Sorry if it looked like that I said that x86 is generally bad, this isn't what I meant to say. Otherwise, I don't understand where RPL IPC being shit can possibly come from. Some people seem to think that AMD still got a full Ghz deficit

You understand how much sense it makes, or doesn't make, to compare instructions per clock cycle between x86 and Arm architectures?

 

[Noyand]

You understand how much sense it makes, or doesn't make, to compare instructions per clock cycle between x86 and Arm architectures?

Flawed premise leads to flawed outcomes, I guess. Maybe you can point me to the x86 architecture, who got a massive IPC lead over RPL, so that I can finally understand on which grounds assimilator arguments stands. Right now, all I could find didn't go in the direction of his argument. So I talked about the only thing that seemed to make sense, no matter how flawed it factually is. I know of an x86 arch that's way more efficient, but not one who's IPC is vastly superior to RPL. MTL does have an IPC regression, but that arch was also supposed to launch two years ago instead of RPL, I'm not even surprised that those CPUs are nothing to write home about.

Intel Meteor Lake P-cores Show IPC Regression Over Raptor Lake?

Intel Core Ultra "Meteor Lake" mobile processor may be the the company's most efficient, but isn't a generation ahead of the 13th Gen Core "Raptor Lake" mobile processors in terms of performance. This isn't just because it has an overall lower CPU core count in its H-segment of SKUs, but also...

www.techpowerup.com

At the cost of double transistor budget and triple power consumption?

yhea, I'm not denying that RPL isn't efficient, that's a fact. But doesn't mean that they have a bad IPC compared to the concurrence. They are just massively less efficient. RPL at 125w is as fast as zen 4 at 65w. You can absolutely say that RPL efficiency is shit, but not the IPC. I will stand on that hill unless proven wrong.

 

[Assimilator]

Flawed premise leads to flawed outcomes, I guess. Maybe you can point me to the x86 architecture, who got a massive IPC lead over RPL, so that I can finally understand on which grounds assimilator arguments stands. Right now, all I could find didn't go in the direction of his argument. So I talked about the only thing that seemed to make sense, no matter how flawed it factually is. I know of an x86 arch that's way more efficient, but not one who's IPC is vastly superior to RPL. MTL does have an IPC regression, but that arch was also supposed to launch two years ago instead of RPL, I'm not even surprised that those CPUs are nothing to write home about.

Intel Meteor Lake P-cores Show IPC Regression Over Raptor Lake?

Intel Core Ultra "Meteor Lake" mobile processor may be the the company's most efficient, but isn't a generation ahead of the 13th Gen Core "Raptor Lake" mobile processors in terms of performance. This isn't just because it has an overall lower CPU core count in its H-segment of SKUs, but also...

www.techpowerup.com

yhea, I'm not denying that RPL isn't efficient, that's a fact. But doesn't mean that they have a bad IPC compared to the concurrence. They are just massively less efficient. RPL at 125w is as fast as zen 4 at 65w. You can absolutely say that RPL efficiency is shit, but not the IPC. I will stand on that hill unless proven wrong.
View attachment 332122

To clarify my throwaway statement, its IPC is shit at sane power draw. I didn't believe I would have to qualify that, but oh well.

 

[Noyand]

To clarify my throwaway statement, its IPC is shit at sane power draw. I didn't believe I would have to qualify that, but oh well.

You were talking about the arch efficiency then. We could have avoided that whole argument . The thing is that I've always seen IPC being judged in a vacuum, like how zen 2 had the best IPC when it launched, but some people still complained about the arch not being able to clock as high as 10th gen and capitalize on their strong IPC. (There's always that set of people who want the fastest thing around, efficiency be damned)

I prefer to talk about efficiency, since if you compare zen 3 to zen 4, the IPC uplift goes from meh to awesome depending on what you do, but zen 4 being able to clock that high compared to zen 3 and still being very efficient is what made it a win. Even if intel were to somehow manage to a 25% IPC uplift compared to RPL, it would be meaningless if they pulled that off with an arch that's "only" 5% more efficient, and still scaling poorly once you try to up the frequency.

AMD Ryzen 7 7700X review (Page 9)

We have 8 cores and sixteen threads available with ther Zen4 based Ryzen 7 7700X review. We'll analyze, test, and benchmark the processor. It might become a sweet spot processor given its great perfo...

www.guru3d.com

RPL is a funny arch, if you look at TPU review, in ST it's actually efficient compared to zen4, but things start to crumble once you get into MT performance. Meanwhile, AMD efficiency in MT always looked like magic. There's something that AMD is doing right that Intel cannot match at the moment.

 

[trparky]

With the old pat in lead, you get the old dog tricks again. "But Intel has AI in it's chips!"

Ah yes, just another marketing term for the tech-illiterate to latch onto. "It has AI in it, it must be better than AMD. Right?"

 

[Noyand]

Ah yes, just another marketing term for the tech-illiterate to latch onto. "It has AI in it, it must be better than AMD. Right?"

From what I understand so far, in 2024, every CPU will have AI acceleration. AMD isn't exactly being quieter about this . The difference is that Lisa mostly talks in conferences, when Pat talks everywhere.

 

[Wirko]

Flawed premise leads to flawed outcomes, I guess. Maybe you can point me to the x86 architecture, who got a massive IPC lead over RPL, so that I can finally understand on which grounds assimilator arguments stands. Right now, all I could find didn't go in the direction of his argument. So I talked about the only thing that seemed to make sense, no matter how flawed it factually is. I know of an x86 arch that's way more efficient, but not one who's IPC is vastly superior to RPL. MTL does have an IPC regression, but that arch was also supposed to launch two years ago instead of RPL, I'm not even surprised that those CPUs are nothing to write home about.

It's all fine, I'm not objecting to you doing IPC comparisons of x86 vs anoter x86. But you also compared x86 to Apple's M3. I-P-C can't be meaningfully compared because the "I", instructions, can't be compared one-to-one. All that CISC vs. RISC stuff, I'm sure you know it. In case you're really interested in details, it's not easy to find good resources, and this is the best I can find:

How does the ARM architecture differ from x86?

Is the x86 Architecture specially designed to work with a keyboard while ARM expects to be mobile? What are the key differences between the two?

stackoverflow.com

 

[lexluthermiester]

It's all fine, I'm not objecting to you doing IPC comparisons of x86 vs anoter x86. But you also compared x86 to Apple's M3. I-P-C can't be meaningfully compared because the "I", instructions, can't be compared one-to-one. All that CISC vs. RISC stuff, I'm sure you know it. In case you're really interested in details, it's not easy to find good resources, and this is the best I can find:

How does the ARM architecture differ from x86?

Is the x86 Architecture specially designed to work with a keyboard while ARM expects to be mobile? What are the key differences between the two?

stackoverflow.com

While true, comparing tasks of a similar nature can easily be compared. Additionally, there are a number of benchmark utilities that are compiled for both platforms and many OSes.