Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Intel Says 10nm Chip Development is On Track
- Thread starter MacRumors
- Start date
- Sort by reaction score
PowerPC is going to 7nm w/ 24 cores in 2023! 
https://www.nextplatform.com/2016/04/07/ibm-unfolds-power-chip-roadmap-past-2020/
https://www.nextplatform.com/2016/04/07/ibm-unfolds-power-chip-roadmap-past-2020/
There are already 48-core MIPS processors.PowerPC is going to 7nm w/ 24 cores in 2023!
https://www.nextplatform.com/2016/04/07/ibm-unfolds-power-chip-roadmap-past-2020/
You can say what you want about "complexity", but in terms of single threaded performance, things like SPEC benchmarks, etc, almost every chip maker in the world has caught up to Intel, who has seen less than a 30% increase the past 8 years. As many have said, this may just mean that Intel hit the "end" first, and everyone else is finally getting there now, but the fact remains that there are no longer any tangible performance benefits to Intel's complexity - they are simply inefficient. Simpler chip designs will have "first mover advantage" to more advanced technology - they will be quicker to optimize. Heck - look at Zen - they pulled an entirely new design together in 2 years with a relatively small team - and they pay the x86 tax.
I disagree. I think built-in optimizations and specializations are the primary game today. If it was pure single-core calculations per second, then we haven't progressed much at all in the past decade. Intel had a Pentium 4 running at 3.8ghz with hyperthreading in 2004 - well over a decade ago. In terms of raw work, individual cores are certainly faster today, no doubt, but not that much faster. Not nearly same difference as comparing the improvements of the previous decade. Almost all the improvements over the past decade have been in multi-threading, power efficiency, and specific optimizations.
One example most people can see easily is HEVC 4k decoding. My 4th-gen Intel Core i7 CPU is very powerful even by today's standards. But it struggles decoding HEVC 4k movies - dropped frames all over barely keeping up. A 5th-gen Intel, any grade even much slower / lower benchmark than my CPU, will easily decode HEVC 4k movies without breaking a sweat because it has hardware specifically optimized for that single task built-in. So something seemingly simple - decoding a video - depends almost entirely on built-in optimization that really wouldn't be reflected in a pure work benchmark.
Apple's "tour de force" is more about supply chain management. They're very good at assembling disperate hardware into a neat and tidy package, and then optimizing their software for it. And they're certainly getting very good at SoC hardware optimization as well, but they're no Intel yet. Many of the optimizations are Intel proprietary - like the virtualization tech I mentioned. ARM licensees can't do that yet for patent reasons.
I'm not saying Apple can't catch up - they can and probably will. But I don't think it's right around the corner, and I disagree with the assessment that Apple's ARM-based chips are ready for Macbook-level duty despite what the benchmarks purport to show.
CISC (Intel) vs RISC (ARM) has been going on for years (like 30) with some of the early discussions that eventually CISC will drown in its own complexity. Maybe CISC has reached its end of life. Move over CISC, time for ARM.
On track is sort of a funny way to phrase it. If it had been on track then we wouldn't be on 14nm+++++++++++++ already.
Oh boy, this is gonna be a long post!
Benchmarks are designed taking in account the chip architecture, it's not the same a truck than a sports car, both run but each on have a different purpose, the same happens with CPU architectures.
ARM was long time ago supported by Intel (remember the StrongARM 64 bits chips?) and Intel collaborated with the chip development , specially into the 64 bits area and desktop like registers.
I think is a smart move for Apple to move to in home CPU design and use. The difference between iOS and MacOS is how the software is adapted to the CPU and chipset, and using their own silicon they, Apple, have more flexibility. x86-64 architecture still has a lot of baggage from the DOS days, and even that Intel has being trying to get rid of it, it's something complex due Microsoft users ( almost 95% Intel customers in CPU area ) still using old software, so must be an agreement between CPU markers and Software and OS markers, they are linked together is this business.
Apple move to in home silicon will help to get rid of these constrains and make the OS more fluent and helping software developers to create smooth apps.
Intel Business is not only CPU or Ram chips, the make telephone switches (micro cpus), ABS control chips and lots of different products. Intel most valuable customers in CPU area are servers users like amazon, universities, Google, etc, etc. So no a big loss for them.
Sorry for the long post....and I'm pretty sure I forgot something more to write down, as usual.
Benchmarks are designed taking in account the chip architecture, it's not the same a truck than a sports car, both run but each on have a different purpose, the same happens with CPU architectures.
ARM was long time ago supported by Intel (remember the StrongARM 64 bits chips?) and Intel collaborated with the chip development , specially into the 64 bits area and desktop like registers.
I think is a smart move for Apple to move to in home CPU design and use. The difference between iOS and MacOS is how the software is adapted to the CPU and chipset, and using their own silicon they, Apple, have more flexibility. x86-64 architecture still has a lot of baggage from the DOS days, and even that Intel has being trying to get rid of it, it's something complex due Microsoft users ( almost 95% Intel customers in CPU area ) still using old software, so must be an agreement between CPU markers and Software and OS markers, they are linked together is this business.
Apple move to in home silicon will help to get rid of these constrains and make the OS more fluent and helping software developers to create smooth apps.
Intel Business is not only CPU or Ram chips, the make telephone switches (micro cpus), ABS control chips and lots of different products. Intel most valuable customers in CPU area are servers users like amazon, universities, Google, etc, etc. So no a big loss for them.
Sorry for the long post....and I'm pretty sure I forgot something more to write down, as usual.
Last edited:
They already have pulled it off, with two separate operations systems. MacOS + iOS merged would continue to behave like two separate operating systems, if the experience is truly optimized for both desktop and touch input like it is now.
Apple has already announced they are working on easy ports of iOS apps to MacOS without having to maintain separate codebases, which I think is the right approach, rather than trying to merge the two.
This doesn't have much to do with merging MacOS + iOS, but instead with switching Macs to ARM. If Apple ever does this, they will simply release a version of MacOS built for ARM, like they did with x86 when Macs switched from PowerPC to Intel. We'll likely also see the return of Universal binaries that run on two different architectures.
But regarding Intel and the frequency of Mac refreshes, it's mainly Apple to blame for the delays with products like Mac mini and Mac Pro.
You know why it didn't happen? because there is not true substitute for CISC right now, maybe in the future, but not now.
Total B.S. - There have been multiple chip iterations. Apple could have been updating each time to a new chip. They also could have done subtle updates, adding Thunderbolt 3, new GPUS, etc. I understand they built themselves into a corner, but didn't they realize that 2-3 years ago? They could have prototyped out a new box for the Mac Pro back then. Other smaller companies are able to do this. The cost is a drop in the bucket to Apple. Hell, they mostly copy the motherboard design from Intel anyway.
Tim just doesn't care. He's busy surfing the web for the latest cool outfits, listening to U2 and playing candy crush.
Tim just doesn't care. He's busy surfing the web for the latest cool outfits, listening to U2 and playing candy crush.
Intel is quite "funny" lately. Delayed 10nm, rising prices of 14nm.
Ryzens are sold for half the price, but the performance is still quite amazing. This is what happens when you think you have no competition. Go AMD!
Ryzens are sold for half the price, but the performance is still quite amazing. This is what happens when you think you have no competition. Go AMD!
Just remember that quantum computers still need a classic computer to process the algorithm for reading the Quantum CPU output, and that 2 main problems still exists, the sensibility of the quantum bits to external influences (a simple fart can kill the system) and the decoherence problem when you add qbits to the CPU.
Yes, I'm into quantum physics too...
I love these debates. The problem... is that all the people that can actually contribute great information... can't.... because they are under NDAs. So we are left with a lot of semininformed opinión.
No, they already pushed people into running Windows on their Macs. They cannot take that away now.
It's better than nothing, but you can look around the internet and put together bits of information a recreate the idea.
But in short CISC has a problem going down to 10nm or less, so they are very few options left. Radical change the architecture, not possible right now due Microsoft and other software developers or, keep trying with the 10 nm and solve all the problems, but it will take some time...
For the long range the info is already out there. Quantum CPU+Classical CPU (hybrid system).
To be fair, you can't directly compare minimum feature sizes between intel and TSMC since they define the term slightly differently.
If they switch to Arm for desktops we are back to wondering which is faster and metrics to show who is. And what if Intel gets faster and macs are slower? If they switch to Arm they better have the sense to keep Intel running in a secret lab forever. At this point if no one else has faster Intel chips it hardly matters if Apple doesn’t.
Most cars can outspeed a semi-truck and yet the semi has a way bigger engine. The load is different. You are comparing apples to oranges.