“Do you want more power-efficiency or CPU horsepower with your next PC chip?” Agam Shah reports for IDG News Service. “The merits of each have propelled an interesting debate among those who want more performance or longer battery life in their PCs. But raw CPU performance drives the ongoing rivalry between AMD and Intel.”
“Intel wants to continue to deliver 10 to 20 percent performance improvements with chips based on the upcoming 10-nanometer and 7-nanometer manufacturing processes, company CEO Brian Krzanich said this week,” Shah reports.
“Some critics may deem those numbers to be incremental, which is why AMD’s upcoming Zen PC chip is getting attention. The chip will deliver 40 percent improvement in instructions-per-cycle, a metric related to CPU performance. Chip makers haven’t historically seen the big performance boost coming in Zen, AMD’s Chief Technology Officer Mark Papermaster has said,” Shah reports. “But Intel has different goals, with a larger focus on increasing battery life in laptops and keeping PCs affordable.”
Read more in the full article here.
MacDailyNews Take: Different chips for different needs. We look forward to seeing how Intel’s first 10-nm chips (Cannonlake) perform when they are released (scheduled for second half of 2017).
Cook won’t allow it. Better for Apple to allow incremental improvement in performance rather than a great leap.
uh 10 nm will be incremental improvement… the AMD is touting a %40 increase. Apple doesn’t use AMD.
That’s what I wrote.
Probably best to write it a little clearer then Frank as it could be read in about three different ways tbh. There has never really been much impetus to use AMD previously on a cost performance basis, but if the promises come to fruition and seriously out perform the glacial Intel designs just maybe…
What happened to higher Ghz? Where’s the 4 Ghz chip at? By now, there should be a 10 Ghz, 20Ghz, 50 Ggz chip. There has been NO INCREASE even up to 4 Ghz !!!! Well then, a computer with slots built in. Chips that have 8 cores, AND THERE OWN SEPERATE memory access. These 8 core chips, and memory, should come 4 on a card (each card has four 8 core chips. And you should have at least four slots per computer ! Anyone heard of the transputer? Both Atari, and Commodore where working on this concept. Of course, they had to deal with the tech they had at the time. BUT the transputer, needs to be given another shot, since there’s NO INCREASE in the speed of these processors. O.K.. there’s some overclocking…… big deal, so what?
The laws of physics have to be accounted for.
There is a huge cost in power for running a chip fast, and there is an inflection point where pushing much more power only yields minimal speed increases. Intel famously tried the “GHz over everything” idea that you are proposing with the NetBurst architecture. It sucked. They hit high speeds, but the chips didn’t actually have good performance. Further, they drew crazy amounts of power and required huge amounts of cooling to dissipate all that heat. Thankfully, they abandoned that idea starting with the Core series of chips, and we have much more powerful computers as a result, even though the GHz speeds may be much less than what the NetBurst Pentiums had.
As for the Transputer stuff… that idea has been replaced with networked computers, blade servers and the like. There is no point in building in the very complicated and expensive extra ports and timing hardware needed to be able to utilize an extra X number of CPUs. Remember that each additional chip you want to add has more than 1,150 pins which means you want computers to have literally tens of thousands of extra traces on the PCBs, all routed by hand, all connected to components, just for the off chance someone might plug in a card some day.
That’s a really good idea in theory, but in my opinion, terrible in practice.
Increasing the maximum potential speed of an automobile won’t make a difference if there’s only two lanes for traffic, one going in either direction, especially if the road is constantly winding.
It’s best to have multiple lanes of traffic and carpool at slower speeds.
What about their own A10 or A11; wouldn’t have to bother with Intel anymore, as they are so yesteryear!!
Truth is nobody knows except Apple Inc. They have surprised the market before with the 64 bit mobile SOC, so why not another submarine project that has been working to deliver real desktop class power!!
Apparently, I prioritize sharing useful information over having to deal with the blow-back of ignorance. Here I go again:
1) Apple’s A-Series chips are too slow at this point to replace Intel’s latest chips.
2) Apple’s A-Series chips are RISC based CPUs. Intel’s chips are CISC based. That means the apps and OS written for the Intel chips have a massive overhead that is dependent upon CISC based code. None of that code is provided by RISC chips. That makes any transition a massive operation that may be made impossible if Intel refuses to license their proprietary x86 code. – – Clearly this is far too complicated for many minds to comprehend. If that’s the case, then go learn in order to understand. Don’t throw ignorance back at me about this problem, OK? I don’t want to hear your problems.
3) All apps and OS X would have to be REWRITTEN to work with the A-Series chips. That is entirely possible with OS X thanks to the brilliant XNU kernel, which is son-of the flexible Mach kernel.
4) Yes, there could be a translation layer, but go see point #2 above about proprietary Intel x86 code. The translation layer would obviously SLOW DOWN OS X.
5) Apple has NOT built virtualization into their A-Series chips. But ARM has been working on providing virtualization. Again, see #2 above about proprietary Intel X86 code.
That’s enough lecturing for today.
DC: it’s not that Apple Computer is a stranger to Massive migratory moves … PowerPC to Intel, migration to OSX UNIX … as for the proprietary Intel X86 code, everything comes at a price or can be encompassed in a deal (i.e. Licensing for X86 code by allowing Intel components into the iPhone/iPad lineup). In other words, never say never with the master of Innovation!
Never say never is fair. But this meme is being tossed around flippantly with no consideration of the hardware or software factors involved. If you go over previous threads on the subject here at MDN you’ll find the same. IOW: Consider the facts of the matter, not the mindless meme mongering.
DC, you’re right about a lot of stuff. But I think the comment above yours was describing a “skunk works” type project to produce a chip that could run circles around the Intel chips similar to what apple did with the mobile chipsets.
Remember that when they bought PA Semi, they had a PowerPC/RISC variant that was faster than the fastest core chips in development at the time. I don’t see any reason why that team couldn’t have been working on both the mobile designs and desktop designs all these years.
As for the overhead issue, if the risc system is fast enough, a second Rosetta could easily mitigate the transition issues. And developers would build universal applications again. Just the exact reverse of the Intel transition. It all comes down to what is in the labs, and who would be able to fabricate the processor. I detailed the other day how I thought t might work:
They would all be “A” series chips, but with different power levels, and capabilities:
A10/10x mobile (iOS devices, Apple TV, and maybe the MacBook)
A10/A10x desktop (13″ MacBook Pro, 21″ iMac, Mac mini)
A10/A10x professional (15″ MacBook Pro, 27″ iMac, Mac Pro)
This rampant speculation, but I don’t think it’s out of the realm of possibility since Intel is hitting the wall again. And the RISC design does have a lot more benefits, and it wouldn’t surprise me if some type of secret project was in the works. I mean nobody had any clue about the A series processors before the A4 just showed up in the original iPad.
If/when Apple and ARM kick Intel’s ass with a faster generation of chips, that will be a lovely day. All the other factors remain. But if I was Apple, that would be a serious project.
Meanwhile, as someone pointed out in the thread, the state of technology is such that we have new barriers in chip design to overcome. Moore’s Law is over, for now anyway. There again is an incentive to escape CISC chips, especially Intel’s.
Totally right. There’s only so much die shrinking can accomplish. And the cisc design is running into a lot of barriers.
It will be a truly happy day when we have risc chips back inside every product. Much more efficient.
No, not much more efficient. They are CISC and RISC are simply different instruction sets. There is nothing inherently more efficient about one over the other. Just because you’re accustomed to Intel’s past history of designing for performance to meet the needs of desktop customers does not mean that ARM (RISC) is fundamentally more efficient. The a9X in the iPad Pro is roughly equivalent to the Core M processor in the 2015 MacBook, and power usage is about the same as well. But as you well know Apple jumped on a newer chipset for 2016, so the MacBook is once again faster than the 12″ iPad. With as good or better battery life as well.
The real story is that ARM/RISC simply doesn’t have as much pro-level high performance software on the market. The way Apple is mismanaging its app stores, ARM/RISC will always be too locked down and too limited to gain software developer support.
Apple’s current Macs wisely chose to move to CISC / Intel architecture so it could run the most powerful software on the market. Cook needs to do better to embrace performance computing if it wants to have a slice of the action for native software. As it is, multiplatform users still have to rely on Parallels or Fusion or Bootcamp+Windows or Crossover in order to run the critical Windows programs they rely on. Moving the Mac to run an ARM chip would kill that ability, and kill Mac sales for good.
Ha ha ha. I love this phrase: “Don’t throw ignorance back at me about this problem, OK?”