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 Shares Details on Upcoming 'Ice Lake' Chips to Follow Coffee Lake and Cannon Lake
- Thread starter MacRumors
- Start date
- Sort by reaction score
I think it's more to do with the fact that Apple has really gone deep with Thunderbolt and until next year when Intel open up the specification to other chip makers, Thunderbolt on an AMD platform is just crazy expensive.
You hit the nail right on it's head. People in this thread proclaiming AMD Ryzen is much better compared to Intel don't seem to have any clue on what they are talking about.
AMD Ryzen provides a great price/threads/cores scenario. For some use cases that is great and all. But for Apple? Not so much. Currently they have no mobile parts. So there is nothing available that is sensible for any MacBook, MacBook Air or MacBook Pro. So the wast majority of Apple's line-up doesn't have anything AMD Ryzen they could implement.
Not to mention how Apple is greatly optimised towards Intel and their architectural design. macOS uses specific Intel optimisation in order to get things like PowerNap and various other features to work. I'm not saying it's impossible to do the same with AMD, but it would require a lot of re-work of the kernel and the drivers. All that work will cost time and money and all that while AMD Ryzen still offers inferior single-thread performance and IPC which is still the most relevant thing for consistent performance for most users.
AMD Ryzen is more like Qualcomm-Snapdragon in the ARM world. They can't compete on the core level, so instead they optimise towards tossing additional cores into the mix in order to achieve better performance. That's not necessarily a bad thing, but as of today there are so much software that is not easily run in parallel. This is not about "lazy developers", this is about coding and how some tasks just doesn't do well when run in parallel making super-fast and super-efficient single-threads relevant for the foreseeable future.
Especially games that relies on so much real-time input and output is downright impossible to run efficiently in parallel. Sure, you have portion of the code that could benefit from parallelization. But all the real-time rendering, input and outputs etc can't be easily spread along multiple threads without causing a huge spike in added latency which is horrible for gaming. Other tasks like encoding etc where latency isn't relevant at all you can easily strip the workload over as many cores/threads as you want without any major drawbacks.
And that's only talking about Ryzen and the whole more cores vs less efficient cores etc.. But Ryzen has also proven to feature slower NVMe throughput, it lacks Thunderbolt etc.. And it comes with half the AVX performance so all macOS applications using modern AVX instructions will essentially get their performance halved.
And in terms of the "Pro" market there is no company that has Ryzen on their QVL lists yet. VMWare for instance do not recommend Ryzen as the virtualisation performance is doing much worse on Ryzen compared to Intel. And until things like these gets solved and stuff starts to get certified and optimised by the major companies in the professional market Apple should not really consider Ryzen unless they want to annihilate this entire market.
So again, to everyone proclaiming "AMD Ryzen is so much better". Do you really know what you are talking about?
I'm personally waiting for the Camp Crystal Lake chips due on Oct. 13.
I hear that will slay all the others.
I hear that will slay all the others.
Depends on what you're using it for. In Cubase [or Logic] I run out of CPU. Ive never even stressed the 16GB ram. Running plugins in realtime needs CPU. Only the very biggest Orchestral Sets use much ram.
Throttle Lake.
Thinner Lake.
Rose Gold Lake.
Solder Lake.
[doublepost=1502878690][/doublepost]
And that's why we need both. In a thicker, non-throttled form-factor. And nobody will be running worthwhile orchestral sets until eLicensers and iLoks go away; and those need USB ports. And the better orchestral sets need ethernet to connect to slave computers. There's a pattern here.
Thinner Lake.
Rose Gold Lake.
Solder Lake.
[doublepost=1502878690][/doublepost]
And that's why we need both. In a thicker, non-throttled form-factor. And nobody will be running worthwhile orchestral sets until eLicensers and iLoks go away; and those need USB ports. And the better orchestral sets need ethernet to connect to slave computers. There's a pattern here.
Kaby Lake...go to http://browser.geekbench.com and view Mac benchmarks to see exactly which CPU are in which model iMac, then head over to http://ark.intel.com to view details.
Perhaps Apple should buy AMD then and that way could invest time, money and resources into giving Intel a run for their money.
[doublepost=1502885476][/doublepost]
Hey Intel, how about Timber Lake?
Or if the Chinese counterfeiters produced on then they could call it Fake Lake
AMD is betamax. Intel is VHS.
AMD may have better tech, like betamax, but Intel will win the war (like VHS). Sorry, AMD, but Rizen will not be your saviour, nor anyone elses.
They don't have enough incentives. If you at most get four times the speed, it's not worth the trouble in many instances. There is also often overhead when you go multithreaded.
If you get 100 times, someone is going to make a game or application that uses all cores and it will be so superior that the rest will have to do it as well. New frameoworks and language would emerge that would make the programming easier.
Look at GPUs, they have thousands of cores, game developers and AI engineers use APIs to use them.
Last edited:
Thank you! Apple's own tech specs websites hardly tell me what generation Intel chip I'd be getting. Am going to upgrade my late-2012 iMac to the new one.
I don't understand the 'confusion' being reported by other websites. Long before now it was rumored that the 14nm++ Coffee Lake platform for mobile designs would be sold alongside the 10nm Kaby Lake platform for desktop designs because of the difficulty Intel was having to produce enough viable CPUs at 10nm for mobile solutions (which are more complex when accounting for iGPUs). The plan has always been to merge desktop and mobile platforms with Ice Lake at 10nm+.
The only head-scratcher is the use of Coffee Lake for desktops.
I believe the only reason Intel released this information is because of external pressure from its partners to prove it is not stagnating technologically in the mobile sector.
The only head-scratcher is the use of Coffee Lake for desktops.
I believe the only reason Intel released this information is because of external pressure from its partners to prove it is not stagnating technologically in the mobile sector.
Yep , But it would be AWESOME if Apple used Thread Ripper !!
AMD infinity fabric is a game changer !
The PCIe 4.0 spec was only just finalized, so I expect we'll see the first enthusiast/workstation/server products with 4.0 support next year. I imagine volume support in mainstream chipsets won't happen until at least 2019, perhaps with Ice Lake and Zen 2.
Intel still has much faster processors in the same power band to any equivalent AMD chip. They are only whipped on price.
So MacBook Pro won't get the Canon Lake 10 nm chip next year based on the article...it will stay on Coffee Lake 14 nm which does not support LPDDR4 RAM (this is what I understand from the article, correct me if I'm wrong)
meaning MacBook Pro won't see the 32G of RAM until IceLake! (Cause we know Apple won't use the regular DDR4 for battery life issues....)
Intel will always be superior to AMD imo. Sure AMD has done some great things and offer a lot more bang for the buck. The problem is that Intel really made a giant leap (similar to Nvidia's GPUs) around the time of the introduction of the Core 2 Duo and later on i7 CPUs that they really didn't have an incentive to keep pushing the envelope. They'll eventually sort themselves out and start delivering new chips on schedule. Meanwhile Apple should continue their own ARM development. The A10 Fusion and A10X are extremely impressive SoCs. Frankly they put a lot of Intel's mobile chips to shame especially those Core Ms. I could see a future regular MacBook using an ARM SoC instead of Intel's solutions in the next decade.
Incentive alone doesn't magically make difficulties of multithreaded code go away.
That would be interesting considering AMD's Ryzen was likely the result of hiring Apple's head of chip designers for the A-series.
I agree the above is the most likely scenario, but there was that one analyst who predicted Apple would release a 32GB RAM MBP in Q4 2017 using desktop RAM. Such a model might appeal to some customers who are willing to trade some battery life for the extra RAM.