Source: Intel plans to beat Apple in 2025

[Colstan]

So, the last time I did one of these, it stirred some interest. As we all know, Apple's future CPU designs are difficult to get a handle on. We know some vague hints about them from various sources, but nothing really concrete. However, the PC guys leak all over the place. Very messy, they are. "Moore's Law is Dead" gave us some details on what Jim Keller was working on while he was at Intel. Here's the video:

My quick summary of the pertinent details:

1. Royal Core is the architecture that Keller was working on to evolve past the current Core architecture.
2. The aim of the project was to beat Apple and ARM in the efficiency game for the foreseeable future.
3. Royal Core was originally planned to be released with Lunar Lake in 2024, but much of that effort has been put off to Nova Lake, in 2025.
4. Biggest update since the original Core architecture, double the IPC of Golden Cove.
5. Lunar Lake plans to be 30% more performant than Meteor Lake, Nova Lake extends this with Panther Cove.
6. SMT4, DDR5 7400, Machine Learning Accelerator planned for 2026.
7. Some of the features planned for Lunar Lake, etc. were "moonshot" ideas, which has gotten Intel into trouble in the past, so Pat Gelsinger has tried to reign in some of those plans from the previous administration.
8. Despite what is said in public, Intel considers Apple to be a substantial threat, not just competition with AMD.

Of the technical details, the most interesting aspect is that Intel is quote "strongly considering" 4-way Hyperthreading for future processors.

Also, I'd just like to point out that Tom has very high level sources in the industry (excluding Apple). He was quite accurate about Big Navi before RDNA2 was announced. Later he revealed that source was AMD Vice President Rick Bergman, a former colleague of @cmaier here at our forums. I'm sure most of you know his credentials. (And for those who don't, he was on the AMD team responsible for x86-64.) So, whoever inside Intel is leaking these details to him knows what they are talking about.

Despite all of the noise around Alder Lake, Intel doesn't actually think they will catch Apple for another four years. So, what do you folks think of Intel planning to beat Apple in 2025?

 

[cmaier]

So, the last time I did one of these, it stirred some interest. As we all know, Apple's future CPU designs are difficult to get a handle on. We know some vague hints about them from various sources, but nothing really concrete. However, the PC guys leak all over the place. Very messy, they are. "Moore's Law is Dead" gave us some details on what Jim Keller was working on while he was at Intel. Here's the video:

My quick summary of the pertinent details:

1. Royal Core is the architecture that Keller was working on to evolve past the current Core architecture.
2. The aim of the project was to beat Apple and ARM in the efficiency game for the foreseeable future.
3. Royal Core was originally planned to be released with Lunar Lake in 2024, but much of that effort has been put off to Nova Lake, in 2025.
4. Biggest update since the original Core architecture, double the IPC of Golden Cove.
5. Lunar Lake plans to be 30% more performant than Meteor Lake, Nova Lake extends this with Panther Cove.
6. SMT4, DDR5 7400, Machine Learning Accelerator planned for 2026.
7. Some of the features planned for Lunar Lake, etc. were "moonshot" ideas, which has gotten Intel into trouble in the past, so Pat Gelsinger has tried to reign in some of those plans from the previous administration.
8. Despite what is said in public, Intel considers Apple to be a substantial threat, not just competition with AMD.

Of the technical details, the most interesting aspect is that Intel is quote "strongly considering" 4-way Hyperthreading for future processors.

Also, I'd just like to point out that Tom has very high level sources in the industry (excluding Apple). He was quite accurate about Big Navi before RDNA2 was announced. Later he revealed that source was AMD Vice President Rick Bergman, a former colleague of @cmaier here at our forums. I'm sure most of you know his credentials. (And for those who don't, he was on the AMD team responsible for x86-64.) So, whoever inside Intel is leaking these details to him knows what they are talking about.

Despite all of the noise around Alder Lake, Intel doesn't actually think they will catch Apple for another four years. So, what do you folks think of Intel planning to beat Apple in 2025?

Jim Keller was also a former colleague of mine, as long as we are listing my former colleagues. . (Rick Bergman should count twice, since I also worked with him at Exponential Technology )

 

[dimme]

Competition is good, it spurs innovation.

 

[jdb8167]

So, the last time I did one of these, it stirred some interest. As we all know, Apple's future CPU designs are difficult to get a handle on. We know some vague hints about them from various sources, but nothing really concrete. However, the PC guys leak all over the place. Very messy, they are. "Moore's Law is Dead" gave us some details on what Jim Keller was working on while he was at Intel. Here's the video:

My quick summary of the pertinent details:

1. Royal Core is the architecture that Keller was working on to evolve past the current Core architecture.
2. The aim of the project was to beat Apple and ARM in the efficiency game for the foreseeable future.
3. Royal Core was originally planned to be released with Lunar Lake in 2024, but much of that effort has been put off to Nova Lake, in 2025.
4. Biggest update since the original Core architecture, double the IPC of Golden Cove.
5. Lunar Lake plans to be 30% more performant than Meteor Lake, Nova Lake extends this with Panther Cove.
6. SMT4, DDR5 7400, Machine Learning Accelerator planned for 2026.
7. Some of the features planned for Lunar Lake, etc. were "moonshot" ideas, which has gotten Intel into trouble in the past, so Pat Gelsinger has tried to reign in some of those plans from the previous administration.
8. Despite what is said in public, Intel considers Apple to be a substantial threat, not just competition with AMD.

Of the technical details, the most interesting aspect is that Intel is quote "strongly considering" 4-way Hyperthreading for future processors.

Also, I'd just like to point out that Tom has very high level sources in the industry (excluding Apple). He was quite accurate about Big Navi before RDNA2 was announced. Later he revealed that source was AMD Vice President Rick Bergman, a former colleague of @cmaier here at our forums. I'm sure most of you know his credentials. (And for those who don't, he was on the AMD team responsible for x86-64.) So, whoever inside Intel is leaking these details to him knows what they are talking about.

Despite all of the noise around Alder Lake, Intel doesn't actually think they will catch Apple for another four years. So, what do you folks think of Intel planning to beat Apple in 2025?

Can't say much about how accurate he is but wow that is kind of painful to watch. Every other sentence is I leaked this and I leaked that.

As for Intel catching Apple in 2025, it didn't sound like he had any clue about what Apple might be working on. So it was a pretty one-sided exposition about what Intel was (maybe) working on. Apple isn't going to sit still. By 2025 they will be on a 2nm node. What can Apple do with 30-40 billion transistors given what they've been able to accomplish with 16 billion. Until Intel shows some ability to deliver, I will remain skeptical. I'm waiting on confirmed information about Alder Lake especially in the 7-15W parts. If the leaked TDP/PL2 numbers are real, they aren't close to being competitive.

 

[cmaier]

Can't say much about how accurate he is but wow that is kind of painful to watch. Every other sentence is I leaked this and I leaked that.

As for Intel catching Apple in 2025, it didn't sound like he had any clue about what Apple might be working on. So it was a pretty one-sided exposition about what Intel was (maybe) working on. Apple isn't going to sit still. By 2025 they will be on a 2nm node. What can Apple do with 30-40 billion transistors given what they've been able to accomplish with 16 billion. Until Intel shows some ability to deliver, I will remain skeptical. I'm waiting on confirmed information about Alder Lake especially in the 7-15W parts. If the leaked TDP/PL2 numbers are real, they aren't close to being competitive.

Well, I’d also point out that Intel has *never* had a design advantage. Their designers have never been the best. When they only had to compete in the x86 market they beat AMD because of their fab superiority, not their design acumen. (To be fair, they beat Citrix, National, and some others in the x86 market both because of fab superiority and design skills). Apple has accumulated many of the best chip designers I know. So if Intel can only match TSMC (either because they fab on TSMC or they get their act together), they’ll still be out-designed by Apple. Not to mention the x86 tax doesn’t go away - four way hyperthreading or not.

 

[JMacHack]

Intel planned to have 10nm working in 2015 too.

Jim Keller was also a former colleague of mine, as long as we are listing my former colleagues. . (Rick Bergman should count twice, since I also worked with him at Exponential Technology )

I haven’t watched the video, but RGT has usually been light on technical information in his other videos. If there’s anything technical do you have any opinions on this? To my mind, this doesn’t seem feasible.

 

[jz0309]

Great plans ... Intel has a lot to actually deliver and only time will tell if they actually do.
Meanwhile, they are designed out more and more, Google apparently planning to equip their Chromebooks with their own chips in 2023, and on the server as well. I can't get to those articles right now but will post links later.

 

[jz0309]

Cloudfare on Icelake

Cloudflare says Intel is not inside its next-gen servers – Ice Lake melted its energy budget

64-core AMD Epycs win again as upgrade delivers performance boost without slurping more 'leccy

www.theregister.com

Google Chromebook CPU

Google developing own CPUs for Chromebook laptops

US software giant ramps up hiring blitz for semiconductor and hardware ambitions

asia.nikkei.com

 

[cmaier]

Intel planned to have 10nm working in 2015 too.


I haven’t watched the video, but RGT has usually been light on technical information in his other videos. If there’s anything technical do you have any opinions on this? To my mind, this doesn’t seem feasible.

The video doesn’t have a lot of technical information. Just “PowerPoint specs.” I wouldn’t get too excited.

 

[JMacHack]

The video doesn’t have a lot of technical information. Just “PowerPoint specs.” I wouldn’t get too excited.

Don’t worry, I’m very skeptical of Intels claims and plans. I was just curious about your take.

Google apparently planning to equip their Chromebooks with their own chips in 2023, and on the server as well. I can't get to those articles right now but will post links later.

I’m skeptical of this as well, Google doesn’t have the best track record when it comes to doing projects long term.

Apple set the stage for the M1 by poaching Intel engineers and buying PA Semi. And the buildup to the M1 began in 2010, and real work on Apple Silicon didn’t start until 2015 reports say. I’m not aware of any acquisitions that Google has made in a similar vein or any projects that they have been building up to.

I’m sure that Google could build a processor. I’m confident they could build a good one. But without a similar foundation I have doubts they could outdo ARM SOC efforts by Snapdragon, NVidia, or Qualcomm.

And even if they did, could they continue to build top quality silicon? Or would they let it fall by the wayside like their other projects.

In any case, we live in interesting times.

 

[Kpjoslee]

Well, I’d also point out that Intel has *never* had a design advantage. Their designers have never been the best. When they only had to compete in the x86 market they beat AMD because of their fab superiority, not their design acumen. (To be fair, they beat Citrix, National, and some others in the x86 market both because of fab superiority and design skills). Apple has accumulated many of the best chip designers I know. So if Intel can only match TSMC (either because they fab on TSMC or they get their act together), they’ll still be out-designed by Apple. Not to mention the x86 tax doesn’t go away - four way hyperthreading or not.

Nah, Intel was clearly doing better during Conroe-Sandy Bridge years over AMD's Bulldozer which almost bankrupted the company. I doubt it was just a fab advantage during that time.

 

[jz0309]

Don’t worry, I’m very skeptical of Intels claims and plans. I was just curious about your take.



I’m skeptical of this as well, Google doesn’t have the best track record when it comes to doing projects long term.

Apple set the stage for the M1 by poaching Intel engineers and buying PA Semi. And the buildup to the M1 began in 2010, and real work on Apple Silicon didn’t start until 2015 reports say. I’m not aware of any acquisitions that Google has made in a similar vein or any projects that they have been building up to.

I’m sure that Google could build a processor. I’m confident they could build a good one. But without a similar foundation I have doubts they could outdo ARM SOC efforts by Snapdragon, NVidia, or Qualcomm.

And even if they did, could they continue to build top quality silicon? Or would they let it fall by the wayside like their other projects.

In any case, we live in interesting times.

Well, the latest Pixel has their own SOC, can’t remember the name, so … and I totally agree, very exciting times indeed with the multitude of chip developments going on

 

[cmaier]

Nah, Intel was clearly doing better during Conroe-Sandy Bridge years over AMD's Bulldozer which almost bankrupted the company. I doubt it was just a fab advantage during that time.

I’ve previously spoken on bulldozer.

Former AMD engineer talks about Bulldozer

X-bit Labs reports an ex-AMD engineer spoke out about some of the reasons why Bulldozer disappoints: The reason why performance of the long-awaited Bulldozer was below expectations is not only because...

www.guru3d.com

Page 2 - Ex-AMD Engineer explains Bulldozer fiasco

Page 2 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.

forums.anandtech.com

Ex-AMD Engineer Reveals Why Bulldozer Design Is Inferior To Intels Sandy Bridge [Archive] - XtremeSystems Forums

AMDs recently launched eight-core Bulldozer processors didnt live up to pre-launch expectations, as it barely managed to outperform Intels current offering of quad-core Core i7 CPUs. The likely reason behind, a former AMD engineer revealed, is over-relying on automated design tools that design...

www.xtremesystems.org

Designers were fine. Architecture was not great, probably. The problem was largely design methodology, mixed with a subpar architecture.

 

[satcomer]

Intel already lost to AMD for Gamer and self building computers for gaming/video working! Then Intel soots down Steve Jobs plan for iPhone so Apple turn to buy and arm contract and small arm company to build a low power phone chip! Then that chip become so good Apple designs a computer chip M1 and dumps Intel!

I know why Apple design APFS for HFS+ replacement for formatting, the M1 chip!

 

[cmaier]

Intel already lost to AMD for Gamer and self building computers for gaming/video working! Then Intel soots down Steve Jobs plan for iPhone so Apple turn to buy and arm contract and small arm company to build a low power phone chip! Then that chip become so good Apple designs a computer chip M1 and dumps Intel!

I know why Apple design APFS for HFS+ replacement for formatting, the M1 chip!

What does APFS have to do with M1?

 

[ADGrant]

Apple set the stage for the M1 by poaching Intel engineers and buying PA Semi. And the buildup to the M1 began in 2010, and real work on Apple Silicon didn’t start until 2015 reports say.

I guess that depends on what you mean by real work or Apple Silicon. The A7 SoC first shipped in 2013. That was Apples first ARM64 design and the first ARM64 SoC in a smart phone.

 

[Joelist]

Let's remember that the time when Intel WAS actually doing good design work was when Intel Israel gave them Banias, Dothan, Nehalem and yes the roots for Conroe. Before these was the horror of NetBurst (which among other things had the Pentium 4 "Death Star" - nickname was for temps not performance). Well, a lot of that Intel Israel group has since moved on, many to.....Apple.

 

[satcomer]

What does APFS have to do with M1?

It means that Apple had to make a new format for the M1! that what's if have do with the M1!

 

[Joelist]

I guess that depends on what you mean by real work or Apple Silicon. The A7 SoC first shipped in 2013. That was Apples first ARM64 design and the first ARM64 SoC in a smart phone.

Apple Silicon is not an ARM design. It uses the ARM ISA but the all important microarchitecture is 100% Apple. Apple dumped ARM reference designs starting with the A6 (Swift).

 

[cmaier]

It means that Apple had to make a new format for the M1! that what's if have do with the M1!

No they didn’t. HFS works fine on ARM.

 

[ChrisA]

It means that Apple had to make a new format for the M1! that what's if have do with the M1!

I think someone is very confused about what is a file system. The above makes no sense.

 

[Gnattu]

No they didn’t. HFS works fine on ARM.

Actually iOS used HFS+ for a long time until APFS is invented.

 

[cmaier]

Actually iOS used HFS+ for a long time until APFS is invented.

Exactly. That’s how i know it works fine on Arm.

 

[Michael Scrip]

"Intel plans to beat Apple in 2025"

Notify us when new Intel chips are in a Macintosh running MacOS...

 

[ChrisA]

I the old days every computer manufacturer had their own instruction set architecture (ISA) IBM had System 360, DEC had the PDP series and then the VAX, CDC and Univac and Data General, Sun Micro systems had SPARC so on and so on. They were all very different from each other. Then at some point they all started using the same Intel chips.

I think we are about see the industry go back to the old pattern. Maybe not all the way but Apple and Google and some others will all use different processors.