As brought up by Gassee, a company called Ampere Computing already produces powerful ARM chipsets. This company produces chips with similar performance to Intel Xeon at half of the consumed power -- 201 watts versus the 400 watts needed by Xeon.
More processing power for less energy. That is exactly why ARM Macs are happening. A13 is already outperforming Intel on notebooks, so it's probably only a matter of time to at least comparable processing power on the high-end.'
The writing is on the wall. And the floor. And the ceiling. And...
Apple can accomplish the same thing transitioning to AMD processors beginning with Zen 2 architecture and still maintain x86 compatibility.
Nah. If you need x86 you can always buy a card for the Mac Pro. The MacBook Pro would need space for a slot though. It can be done. No showstoppers.
Anyone who works in or develops HPC software cringes at this. Sure it's likely you can make an ARM chip with the performance of high-end Xeons, but the world of HPC software is a million years away from XCode app development where you can flip a switch to build for a new architecture. There are so many bespoke libraries (e.g. Intel MKL) and years of optimization that have gone into getting HPC code to run fast on AVX Xeons.
Apple is a bit-player in HPC with the Mac Pro because of their 'war' with nVIDIA ( cutting off access to the compute power of their massively parallel GPUs) - it would just sideline them even more if they went ARM.
I’ve been discussing this for some time and have yet to see anyone actually generate LINPACK or HPCG benchmark results let alone easy to use software. You can blame it on Apple or NVIDIA but HPC hasn’t been on Apple’s radar for years.
Can anyone tell me why this would possible be a good thing?
Getting out from under Intel's lack of progress and vision, having complete control over yet another part in their supply chain. AMD is doing ok, but is still catching up, and it'll be 'the other one' that may encounter similar issues Intel has in the long run. Apple wants total control over CPU and GPU and any other component. Intel and AMD chips define what a computer is today, and an A12 is already fast enough for a lot of consumers. Want a more energy efficient device? Intel and AMD have no answers. With ARM apple can glue everything shut in a tiny box and be done with it. A Mac Mini the size of an Apple TV is not unlikely and would be a great testcase.
As brought up by Gassee, a company called Ampere Computing already produces powerful ARM chipsets. This company produces chips with similar performance to Intel Xeon at half of the consumed power -- 201 watts versus the 400 watts needed by Xeon.
More processing power for less energy. That is exactly why ARM Macs are happening. A13 is already outperforming Intel on notebooks, so it's probably only a matter of time to at least comparable processing power on the high-end.'
The writing is on the wall. And the floor. And the ceiling. And...
Apple can accomplish the same thing transitioning to AMD processors beginning with Zen 2 architecture and still maintain x86 compatibility.
Have you emailed Cook about this?
Seriously tho, if AMD offered the same benefits I’m sure they’d have done it. If ARM Macs come out clearly they believe they can do it better.
Very interesting, nice info. Is Gassee former Apple?
JLG is indeed ex-Apple, in the late 80s and early 90s, but he doesnotknowitall, knowitall. He doesn’t have any real sources inside Apple nor its supply chain, and is just shooting the breeze here.
I thought so, but thanks for the confirmation. I have seen him at an Apple session in Amsterdam. I know its someone with software expertise, one of the best I think.
JLG has pretty storied history vis a vis Apple. He was the guy that guided the features of Apple Macs post Steve Jobs starting in 1985. The Mac II, IIci, IIfx, all that Mac hardware of that era, etc, were his babies. He left Apple to form Be, Inc and push super threaded BeOS in 90s, which was the OS Apple was going to buy before deciding on NeXTSTEP. They thought they had Apple by the huevos and was pushing for a higher buyout, but Apple bought NeXT instead.
After that, he tried to push BeOS as an alternative PC operating system. Obviously that failed as you have to have MS Office to be successful as a PC operating system, or be free like Linux or Unix is. Tried their hand at being an Internet Appliance operating system after that, obviously failed. After that, Palm bought them out and BeOS tech was going to be in the next gen PalmOS, but they could never pull legacy PalmOS apps along, and was never able to get any OEMs to license PalmOS Cobalt. I’m guessing Rubenstein didn’t like it because Palm didn’t use it for webOS. It died inside Palm, or maybe I should say lies dormant in the Chinese company that bought Palm, Access.
Palm was basically the poster child of making the mistake of following pundit-class advice. They did everything that people were advising Apple to do: allow Palm clones, license the software, split up the company to be a separate hardware and software companies, find a buyer, who knows what else.
As brought up by Gassee, a company called Ampere Computing already produces powerful ARM chipsets. This company produces chips with similar performance to Intel Xeon at half of the consumed power -- 201 watts versus the 400 watts needed by Xeon.
More processing power for less energy. That is exactly why ARM Macs are happening. A13 is already outperforming Intel on notebooks, so it's probably only a matter of time to at least comparable processing power on the high-end.'
The writing is on the wall. And the floor. And the ceiling. And...
Apple can accomplish the same thing transitioning to AMD processors beginning with Zen 2 architecture and still maintain x86 compatibility.
Have you emailed Cook about this?
Seriously tho, if AMD offered the same benefits I’m sure they’d have done it. If ARM Macs come out clearly they believe they can do it better.
Re-read my tweet, I specifically said starting with the Zen 2 architecture which was officially released the last 3 months.
The writing is on the wall for Intel with their failed cellular efforts, flawed CPU's and a schedule for power usage and performance that is never meant. Apple can no longer count on them, plus bringing it in house will make everything work together better.
As brought up by Gassee, a company called Ampere Computing already produces powerful ARM chipsets. This company produces chips with similar performance to Intel Xeon at half of the consumed power -- 201 watts versus the 400 watts needed by Xeon.
More processing power for less energy. That is exactly why ARM Macs are happening. A13 is already outperforming Intel on notebooks, so it's probably only a matter of time to at least comparable processing power on the high-end.'
The writing is on the wall. And the floor. And the ceiling. And...
Apple can accomplish the same thing transitioning to AMD processors beginning with Zen 2 architecture and still maintain x86 compatibility.
Have you emailed Cook about this?
Seriously tho, if AMD offered the same benefits I’m sure they’d have done it. If ARM Macs come out clearly they believe they can do it better.
Re-read my tweet, I specifically said starting with the Zen 2 architecture which was officially released the last 3 months.
Are you under the impression that 1) Apple could wait until now to make plans, and 2) high-level Apple engineers under NDA wouldn’t have known about new AMD architecture plans before you? I’m not. I think they likely began this process years ago, and I think they still decided they can do it better than what AMD could offer behind closed doors. Not to mention not being beholden to yet another chip maker’s schedule, which has been the problem for decades.
I've always been an Intel fan. They always have just worked. I'd actually like to see a move away from Intel to ARM. I think Intel bungled their sizing of their chips too much in a competitive area, and that this change is justified. However, there is a secondary issue. Not all apps are available on Mac. Not all of these apps are on the Mac and no, don't try to say that there are "apps of the same quality" available because in some verticals, they just are not there. I have to run a virtual machine to run windows on my mac. Intel x64/x86 emulation is going to be horrible on an ARM mac. emulating one cpu instruction set on another cpu has always been horrible.
Mac Pro is dead anyways for serious enterprise customers who would not have bothered waiting 6 years for a new Mac Pro, they are not coming back. Any AMD CPU (or any Threadripper/Epyc based workstation) above 32 cores is faster than what Apple or Intel for that matter can offer. It honestly took me by surprise and people start to realise.
To gain those customers back Apple would have to offer similar or better performance than that, and currently it's infeasible for Apple to do so as they have neither the packaging technology, nor the interconnect, no system-wide coherency strategy (look up navi 2x and ZEN3) and no experience with seriously high power consumption designs (above 10 watt or what your Apple TV can emit).
If Apple can offer an advantage, it could have been maybe the Afterburner card, but it seems it's not enough to justify less than half the CPU cores and laughably small PCIe bandwidth.
As brought up by Gassee, a company called Ampere Computing already produces powerful ARM chipsets. This company produces chips with similar performance to Intel Xeon at half of the consumed power -- 201 watts versus the 400 watts needed by Xeon.
More processing power for less energy. That is exactly why ARM Macs are happening. A13 is already outperforming Intel on notebooks, so it's probably only a matter of time to at least comparable processing power on the high-end.'
The writing is on the wall. And the floor. And the ceiling. And...
Apple can accomplish the same thing transitioning to AMD processors beginning with Zen 2 architecture and still maintain x86 compatibility.
They can't. Not even close. Moving from Intel would be a lateral move, at best, as Apple would still be at the mercy of a different vendor for designing their CPUs.
Anyone who works in or develops HPC software cringes at this. Sure it's likely you can make an ARM chip with the performance of high-end Xeons, but the world of HPC software is a million years away from XCode app development where you can flip a switch to build for a new architecture. There are so many bespoke libraries (e.g. Intel MKL) and years of optimization that have gone into getting HPC code to run fast on AVX Xeons.
Apple is a bit-player in HPC with the Mac Pro because of their 'war' with nVIDIA ( cutting off access to the compute power of their massively parallel GPUs) - it would just sideline them even more if they went ARM.
Didn't someone in Apple once say...
"Dare to be different!"
I'm comfortable with this move provided there is ample time for the transition. Some key app suppliers work at a pace that even snails would find slow. If Apple brings the major app players with them then it will be fine.
Sure... a lot of the usual suspects will winge and moan but... Apple needs to be clear about how long they will support Intel Mac's. Five years is an absolute mimimum IMHO.
I've been messing with a 64-core/128-thread AMD Threadripper and it absolutely screams compared to a 28-core MP--frequently over 3x faster on threaded workloads and much less expensive.
I love Apple's CPU development but it seems like it doesn't make any sense for them to switch anything to arm anymore. It seems like they are giving a lot more power to the ipad pro with mouse support. I love the iPad Pro and it's replaced my laptop already.
I'm unclear about what the benefit would be to switch the other mac lines to ARM. It seems like it'd be a lot more issue than benefit to switch all the software, windows compatibility, etc. Plus with Foveros 3D from Intel and AMDs 3d stacking these chips are becoming a lot more power efficient.
Can anyone tell me why this would possible be a good thing?
I think there are several reasons why a MacBook Pro w. ARM makes sense: (1) tighter control over hardware/OS integration - an ARM chip could be designed or tweaked to help macOS; Intel isn't rolling out a custom version of x86 for Apple. (2) cost - one or more ARM chips will cost Apple a lot less than buying a single x86 chip from Intel. (3) an ARM laptop is going to be less energy draining than an equivalent x86 laptop, allowing Apple to reduce battery size (cost saving) or increase use time (better customer experience).
The only question in my mind is how Apple will transition its customers from x86 base software packages to ARM-based ones. I imagine much of the open source apps as well as currently supported commercial software will be recompiled/tweaked for the new processor, but what about the existing x86 stuff that isn't upgradeable? In past transitions, Apple included an emulator (e.g. Rosetta) - but that only works if the new processor has clearly more power than the one it replaces - otherwise the user experience for emulated software will suck.
Anyone who works in or develops HPC software cringes at this. Sure it's likely you can make an ARM chip with the performance of high-end Xeons, but the world of HPC software is a million years away from XCode app development where you can flip a switch to build for a new architecture. There are so many bespoke libraries (e.g. Intel MKL) and years of optimization that have gone into getting HPC code to run fast on AVX Xeons.
Apple is a bit-player in HPC with the Mac Pro because of their 'war' with nVIDIA ( cutting off access to the compute power of their massively parallel GPUs) - it would just sideline them even more if they went ARM.
GC (Grand Central) is Apples answer, as parallelism is the answer to supercomputer power. Works fine with XCode.
Grand Central works as well as the Backplane and interprocessing designs of the CPU architecture. In short, it's not a panacea. Slap two TR or EPYC CPUs in today's Mac Pro and you've got 128 Cores/ 256 threads that nothing ARM designs with it's energy footprint limitations will ever touch. Zen 3 arrives this Fall. Sorry, but no, Apple won't invest heavy resources for a small market product. They'll decide between continuing with Intel, or finally move forward on AMD Zen.
RDNA 2.0 is about to come out and it lays to waste the GPGPUs custom designed by AMD already. The rational mind says Zen 3/RDNA 2.0/Afterburner 2.0 in the next Mac Pro.
Why should we listen to guy that just suddenly realized the obvious?
Suddenly?
I first discussed “ARM-ing” the Mac in 2018 (here and here), and went at it again in July 2019 (here and here). Last year, I concluded that a move to the ARM processor created two serious challenges for the Macintosh line, two forks in the product line.
What about his article is full of crap? The part where he says " ARM-ing the Mac is easier said than done" or where he lays out all the reasons why this is a difficult undertaking?
Gasse was head of hardware when I worked at Apple. As I recall, he was fired for being a pompous jerk. In any event, Apple is NOT moving from Intel to ARM on a high end product they just recently started shipping. There is no equivalent ARM CPU that runs these high end workstation applications. Furthermore, given the relatively low volume of Mac Pro's sold, there is simply no financial justification to switch. My guess is that this is simply Jean-Louis being Jean-Louis...
If the Mac Pro becomes the likely candidate for the ARM chips then the apparent upgradeability of the machine would be limited. How many versions of ARM chips will Apple create (8 core, 12 core, 16 core or 28 core)? I do not think this would be economically feasible for the pro-market because Apple will overcharge for these processors as they overcharge for their RAM and other components. This would be a mistake for Apple because they have already priced this machine out of reach for the low-end professionals. The more Apple wants to control its components the more it will hurt the customer because of the greed of Tim Cook's Apple. I am sorry, I have been a long time Apple customer but see the end of the road for me buying any more Apple products because of Apple's apparent unyielding price structure! Macs no more for me!
Comments
With ARM apple can glue everything shut in a tiny box and be done with it. A Mac Mini the size of an Apple TV is not unlikely and would be a great testcase.
Seriously tho, if AMD offered the same benefits I’m sure they’d have done it. If ARM Macs come out clearly they believe they can do it better.
After that, he tried to push BeOS as an alternative PC operating system. Obviously that failed as you have to have MS Office to be successful as a PC operating system, or be free like Linux or Unix is. Tried their hand at being an Internet Appliance operating system after that, obviously failed. After that, Palm bought them out and BeOS tech was going to be in the next gen PalmOS, but they could never pull legacy PalmOS apps along, and was never able to get any OEMs to license PalmOS Cobalt. I’m guessing Rubenstein didn’t like it because Palm didn’t use it for webOS. It died inside Palm, or maybe I should say lies dormant in the Chinese company that bought Palm, Access.
We’ll see tho.
To gain those customers back Apple would have to offer similar or better performance than that, and currently it's infeasible for Apple to do so as they have neither the packaging technology, nor the interconnect, no system-wide coherency strategy (look up navi 2x and ZEN3) and no experience with seriously high power consumption designs (above 10 watt or what your Apple TV can emit).
If Apple can offer an advantage, it could have been maybe the Afterburner card, but it seems it's not enough to justify less than half the CPU cores and laughably small PCIe bandwidth.
The only question in my mind is how Apple will transition its customers from x86 base software packages to ARM-based ones. I imagine much of the open source apps as well as currently supported commercial software will be recompiled/tweaked for the new processor, but what about the existing x86 stuff that isn't upgradeable? In past transitions, Apple included an emulator (e.g. Rosetta) - but that only works if the new processor has clearly more power than the one it replaces - otherwise the user experience for emulated software will suck.