How do we know ARM won’t be like PowerPC?

[Quu]

PowerPC processors were a RISC-based processor that everyone claimed was an advanced design, superior to Intel, that could outperform Intel even at lower clock speeds. But as we all know, within a few years on the market, x86 chips kinda... kicked their ***.

Now Apple is switching once again to a RISC architecture that they claim is superior, at least on paper.

What went wrong with PowerPC? How do we know it won’t happen with ARM?

PowerPC was better but Intel had several advantages not intrinsic to their CISC designs.

1. Better fabrication. They could increase transistor density faster than their competitors including IBM. Intel had a two generation advantage for most of the last 15 years until very recently.

2. Better designs. Intel's architecture was just better. Making these advanced microchips is a 3 to 5 year design endeavour and so you may get outclassed at the last hurdle by a competitor that simply out designed you. It's about placing the right bets at the right time and it's exceedingly difficult, eventually you have to pull the trigger on a design even if you know it has some design flaws to meet your time obligations.

Neither of these things are true anymore. Intel is now two generations behind TSMC in fabrication for high performance chips and one generation behind for low-energy chips. Apple will be using TSMC to manufacture all their processors so they can leverage TSMC's advantage here as AMD has recently done.

Due to Intel's fabrication problems they're unable to execute on designs. They already had several generations of chips ready to go but were unable to manufacture them. Their problems with getting 10nm to work outside of a laboratory are well documented and Intel has had to scale back their release of 10nm parts significantly due to this.

So getting back to the premise of the topic with the above context out of the way. The reason PowerPC failed wasn't because RISC is a bad idea. It failed because IBM couldn't execute on new designs based around it. When you are in the chip game you're only as good as your latest chip because there is fierce competition.

The question is will Apple be able to maintain their high level of execution? I think they have the money to make sure they never fall behind and by that I mean they can hire whoever they want to design their chips, they can purchase almost any company that has unique technology that would give them an edge and they can pay for the latest fabrication technology (currently that is 5nm offered by TSMC and if you've not noticed Apple are the only company in the world shipping 5nm processors today in the latest iPad and the just announced iPhones).

I think the future for Apple Silicon is incredibly bright and it really all comes down to money. They are simply outspending everyone on chip design and fabrication right now and that alone will guarantee they stay at the top, it's really that simple. When you can make chips with a 1.83x higher density to your competition with a 9 to 12 month lead.. it's just impossible for anyone to catch up.

As an investor I will say there is one major thing that could hurt them, hubris. If they think they're so far ahead they decide to take their foot off the accelerator that will result in the competition nipping at their heels. This is partly what happened to Intel they sandbagged the mainstream market with quad cores for just over a decade and it gave AMD the time to come back with a stellar architecture, the chiplet design AMD came up with caught Intel completely off guard with its scalability, high yield rates and performance.

That same thing could happen to Apple if they don't keep pushing, you can't stop on any front (design spending or gaining almost exclusive access to the latest fabrication nodes available). So yeah, for me execution isn't a problem, it's will. Hopefully the management doesn't change their ideas around perusing the best technology with no expense spared.

 

[casperes1996]

N

As an investor I will say there is one major thing that could hurt them, hubris.

’Ah, yes. Hubris. My old nemesis

 

[panjandrum]

Look how fast and smooth iPadOS is. This is what macOS should feel like. Long term Mac users have always been annoyed the macOS support for third party graphics chips hasn’t been as good as Windows/Linux.

It's crazy how poorly AMD's GPUs work to accelerate the MacOS UI and I'm absolutely confident that if Apple was using NVIDIA, or if Apple and AMD simply bothered to optimize the drivers properly, we would be way better off (because AMD does make some fine GPUs these days).

The R9 in my 2015 MacBookPro can't even display a proper frame-rate in Netflix if I have subtitles on. I have to reboot into Windows for a decent Netflix experience. What a joke. (Also, it's this kind of thing that makes many of us worry about the future of ARM Macs if we lose decent Windows performance - Being able to run Windows on our Macs at full speed, or even quickly through virtualization, allows many of us to work around problems inherent in the MacOS. It's not just about being able to run specific Windows games and other titles. I'm also able to run Netflix at 1080p in Windows 10 virtualized (Parallels) on my 2009 (flashed to 5,1) MacPro. If I wasn't able to do that I would be stuck at 720p because the asshats at Apple completely broke Netflix in Safari on older Macs, no matter how capable they are. But with Parallels, Win10 and Microsoft Edge I can Netflix at 1080p on my old beast.)

For a comparison on how poorly AMD GPUs accelerate the Mac UI compared to NVIDIA take a look at the UI section of the benchmarks I put up here and see how well the sole, lowly NVIDIA card I texted did against the much more powerful AMD cards (only the 580 was better at UI tasks - although the NVIDIA card was, of course, effectively useless for 3D):

So what’s the cheapest graphics card for Mojave on a 5,1?

Ugh, I always forget I can't just upload images to this forum. So here are the benchmarks quickly: Unigine Heaven Benchmark 4.0 - Extreme preset, plus the now ancient X-Bench Quartz, Open GL, and UI Tests: NVIDIA GeForce GT 710 - Zotak (Mojave): FPS: 4 Score: 102 Min FPS: 2.5 Max FPS: 8.9...

forums.macrumors.com

Hopefully the move to ARM will allow Apple to really make the MacOS experience as good as it should be (and could have been already, had Apple/AMD bothered to make it work properly).

 

[mr_roboto]

sort Of.

our Austin office was working on another project. When we went out of business, most of those folks stuck around and became EVSX, which was mostly a consulting thing to keep themselves going. Then they became intrinsity and then they got bought. The guy in the cubicle next to me, who was a key architect of our work, moved to Austin and became a critical part of all that.

Thanks for the Exponential story, and I didn't realize the connection between Exponential and Intrinsity.

And speaking of Intrinsity, for those who don't know, Intrinsity hardened ARM's Cortex-A8 and A9 for Samsung's process. Apple used Intrinsity's A8 and A9 in iPhone SoCs. Around that time, I worked on a SoC built in Samsung 45LP, and we also used Cortex-A9. I was told it was the same hard macro which Apple used in A5 (the iPhone 4S SoC), and that Apple had funded a lot of the work. I don't pretend to know how all the licensing and IP rights worked to make it possible for us (a small fabless semi company targeting a non-smartphone market) to license that core, but we could.

It was around that time that Apple bought Intrinsity. I remember speculation that it was to deny other companies access to Intrinsity's hardened versions of ARM's Cortex series, and while I'm sure there was some truth to that, in retrospect it seems clear the main reason was to acquihire an in-house physical design team for their internal CPU design projects.

So getting back to the premise of the topic with the above context out of the way. The reason PowerPC failed wasn't because RISC is a bad idea. It failed because IBM couldn't execute on new designs based around it. When you are in the chip game you're only as good as your latest chip because there is fierce competition.

I wouldn't say "couldn't execute". It's more like "did not want to". The original 1990s concept was that Apple, IBM, and Motorola (AIM) would all partner together and design chips and share process tech so the chips could be made by either IBM or Motorola. By the G4 timeframe, IBM had given up on the AIM alliance, partially or completely pulled all its employees from the Somerset joint design center, and gone back to its pre-alliance activity of focusing on two primary markets: AIX workstations and big iron.

It was Motorola who had serious problems executing (both design and manufacturing), which eventually led Jobs to seek a renewed partnership with IBM. That resulted in G5, which was a derivative of IBM's POWER4. IBM executed quite well on G5, for what it was, but wasn't willing to design truly new cores suitable for laptops, at least not for what Jobs was willing to pay.

So at that point, it was either go all-in with PA Semi's PowerPC design, or team up with Intel. With the benefit of hindsight, PA Semi's design team was pretty good, since it went on to become the core of Apple's CPU design team, but in 2004-2005 it must've looked really risky - could a small startup keep up with the juggernaut that was the PC industry?

 

[JMacHack]

PowerPC processors were a RISC-based processor that everyone claimed was an advanced design, superior to Intel, that could outperform Intel even at lower clock speeds. But as we all know, within a few years on the market, x86 chips kinda... kicked their ***.

Now Apple is switching once again to a RISC architecture that they claim is superior, at least on paper.

What went wrong with PowerPC? How do we know it won’t happen with ARM?

Granted, I was very young when this was happening, but from my understanding PowerPC actually did outperform x86 chips for awhile. The problem came when the two major players (IBM and Motorola) couldn't or wouldn't make PC-level chips. Intel, meanwhile, was rapidly improving their PC-level processors and aggressively improving their manufacturing process (tick-tock).

Now, the shoe's on the other foot. Intel's stagnated, and for many reasons hasn't been able to improve their processors as rapidly as before. To the point AMD pulled a cinderella story and started whooping their ass with Zen despite being much smaller.

How do we know it won't happen with ARM? We don't, we don't have a crystal ball. However, Apple has likely been looking at their processors vs. others and determined that their trajectory is higher. Maybe in another 10 years it'll be someone else kicking Apple Silicon around, who knows? But sticking with Intel when they've been stagnating because they MIGHT get back on track someday is a foolish idea.

 

[cmaier]

Granted, I was very young when this was happening, but from my understanding PowerPC actually did outperform x86 chips for awhile. The problem came when the two major players (IBM and Motorola) couldn't or wouldn't make PC-level chips. Intel, meanwhile, was rapidly improving their PC-level processors and aggressively improving their manufacturing process (tick-tock).

Now, the shoe's on the other foot. Intel's stagnated, and for many reasons hasn't been able to improve their processors as rapidly as before. To the point AMD pulled a cinderella story and started whooping their ass with Zen despite being much smaller.

How do we know it won't happen with ARM? We don't, we don't have a crystal ball. However, Apple has likely been looking at their processors vs. others and determined that their trajectory is higher. Maybe in another 10 years it'll be someone else kicking Apple Silicon around, who knows? But sticking with Intel when they've been stagnating because they MIGHT get back on track someday is a foolish idea.

One major difference is motivation. IBM made money from its own servers and didn’t much care about Apple. And Apple was low volume, so why should they have cared?

Apple has MANY more customers for its processors now, and Apple is designing the chips for itself, so it’a motivated.

 

[curmudgeonette]

So at that point, it was either go all-in with PA Semi's PowerPC design, or team up with Intel. With the benefit of hindsight, PA Semi's design team was pretty good, since it went on to become the core of Apple's CPU design team, but in 2004-2005 it must've looked really risky - could a small startup keep up with the juggernaut that was the PC industry?

One other data point: For essentially the entire PowerPC era, Apple was one generation of memory performance behind leading edge x86 systems:

While Apple was still on FPM/EDO DRAM, x86 had gone to SDRAM.
When Apple started using PC66 SDRAM, x86 was on PC100.
Then PC100 vs. PC133.
...
Then PC3200 vs. PC2-4266.

 

[MisterMe]

...

I wouldn't say "couldn't execute". It's more like "did not want to". The original 1990s concept was that Apple, IBM, and Motorola (AIM) would all partner together and design chips and share process tech so the chips could be made by either IBM or Motorola. By the G4 timeframe, IBM had given up on the AIM alliance, partially or completely pulled all its employees from the Somerset joint design center, and gone back to its pre-alliance activity of focusing on two primary markets: AIX workstations and big iron.

It was Motorola who had serious problems executing (both design and manufacturing), which eventually led Jobs to seek a renewed partnership with IBM. That resulted in G5, which was a derivative of IBM's POWER4. IBM executed quite well on G5, for what it was, but wasn't willing to design truly new cores suitable for laptops, at least not for what Jobs was willing to pay.

So at that point, it was either go all-in with PA Semi's PowerPC design, or team up with Intel. With the benefit of hindsight, PA Semi's design team was pretty good, since it went on to become the core of Apple's CPU design team, but in 2004-2005 it must've looked really risky - could a small startup keep up with the juggernaut that was the PC industry?

Thank you. After reading so much nonsense on this thread, it is refreshing to read something written by someone who knows what he is talking about. I have owned a number of PPC-based Macs including a number of beige Power Mac G3s, a PowerBook G3, and a number of Power Mac G5s. In fact my primary computer at home and my primary computer at work are Power Mac G5s running MacOS X 10.4.11 with Classic. I have two MacBook Pros. I stuck with the G5s on the desktop because they run two applications--one of them Classic--that I depend upon.

Enough about me. In 2003, Apple introduced its top-level desktop computer based on the PPC G5. in 2004, Apple introduced the G5-based iMac G5. However, the computer that we anxiously awaited was the PowerBook G5. There was never a PowerBook G5 because the G5 ran too hot. Steve Jobs asked IBM to produce cooler powerful microprocessors, but IBM rebuffed him. IBM was into workstations, servers, and mainframes. It didn't give a flip about personal computers.

However, there came a time when IBM did give a flip. That was after Steve Jobs announced Apple was switching from PPC to Intel. Intel has recently experienced a breakthrough that allowed it to significantly reduce the heat generated by its processors. IBM executives went to Jobs and asked him to reconsider. However, Jobs having switched was not switching back.

 

[throAU]

PowerPC processors were a RISC-based processor that everyone claimed was an advanced design, superior to Intel, that could outperform Intel even at lower clock speeds. But as we all know, within a few years on the market, x86 chips kinda... kicked their ***.

They were.

The reason PPC "fell behind" (for the markets Apple target, like laptops) is that IBM and Motorola, who were largely responsible for the PPC architecture were not interested in mobile as it was a tiny segment of the market. They continued (and still do) to build PPC for servers, which wouldn't fit in mobile computers like Apple wanted to build.

Now Apple is switching once again to a RISC architecture that they claim is superior, at least on paper.

What went wrong with PowerPC? How do we know it won’t happen with ARM?

It won't happen this time because APPLE are designing and building the processors this time around, they aren't at the mercy of what some other company wants to sell (now, they can build exactly what they want for the products they sell). Before, back in the PPC days, Apple were almost broke, responsible for a low volume of machines, and IBM/Motorola just didn't care. They were a blip on the balance sheet.

To be clear: there's nothing wrong with the PPC architecture. Its just that the companies making it were (and still are) targeting servers (which Apple don't make), not mobile.

 

[Quu]

I wouldn't say "couldn't execute". It's more like "did not want to".

I honestly don't believe that, it's just the history revisioning. "We could, we just didn't wanna" is loser talk to save face.

We shouldn't forget they did the Wii, XBOX and PS3 chips around this time frame too. IBM had a lot of consumer chips on the go but weren't able to deliver. Remember Steve Jobs promising us 3GHz G5's? - They told him they would be able to deliver that, never happened. That's a failure to execute not a decision not to provide it, he never would have said it if they hadn't promised it.

And they went all in on the custom Cell processor with Sony and Toshiba. They clearly had the will to do special projects. That chip was PowerPC based and could have superseded the G5 to be honest, sure the SPE design made it complicated to program for but it had potential in later iterations or even Apple could have joined that alliance and been involved in the development process to get a custom version with more PPE's which is exactly what Microsoft did when they wanted a processor for the XBOX 360 (IBM took the PPE design from Cell, tripled it up and gave it to Microsoft in a custom design).

IBM in my eyes for sure couldn't execute on a competitive product. It's also important to realise that the Conroe architecture from Intel was just way too good. It didn't just eclipse IBM's G5 and their PowerPC range entirely across all segments (including servers) it also decimated AMD and it has taken them just over a decade to recover.

 

[sanfrancisofont1984]

Intel had like 2 generation fab process lead when PowerPC to Intel transition happened. Now Intel is 1 generation behind everyone else. And the benchmark shows.

I wouldn't spend too much time just thinking about the ISA even though x86-64 is likely the No.1 in terms of ISA complexity.
Other ARM chips (e.g. Qualcomm) seem far behind AS so I would almost say ISA doesn't matter as long as someone knows what they are doing.
A few people use Talos II POWER9 workstation. I thought the price was kinda high but now I realized it is probably cheaper than a MP 7,1.

 

[thingstoponder]

Thanks for the Exponential story, and I didn't realize the connection between Exponential and Intrinsity.

And speaking of Intrinsity, for those who don't know, Intrinsity hardened ARM's Cortex-A8 and A9 for Samsung's process. Apple used Intrinsity's A8 and A9 in iPhone SoCs. Around that time, I worked on a SoC built in Samsung 45LP, and we also used Cortex-A9. I was told it was the same hard macro which Apple used in A5 (the iPhone 4S SoC), and that Apple had funded a lot of the work. I don't pretend to know how all the licensing and IP rights worked to make it possible for us (a small fabless semi company targeting a non-smartphone market) to license that core, but we could.

It was around that time that Apple bought Intrinsity. I remember speculation that it was to deny other companies access to Intrinsity's hardened versions of ARM's Cortex series, and while I'm sure there was some truth to that, in retrospect it seems clear the main reason was to acquihire an in-house physical design team for their internal CPU design projects.




I wouldn't say "couldn't execute". It's more like "did not want to". The original 1990s concept was that Apple, IBM, and Motorola (AIM) would all partner together and design chips and share process tech so the chips could be made by either IBM or Motorola. By the G4 timeframe, IBM had given up on the AIM alliance, partially or completely pulled all its employees from the Somerset joint design center, and gone back to its pre-alliance activity of focusing on two primary markets: AIX workstations and big iron.

It was Motorola who had serious problems executing (both design and manufacturing), which eventually led Jobs to seek a renewed partnership with IBM. That resulted in G5, which was a derivative of IBM's POWER4. IBM executed quite well on G5, for what it was, but wasn't willing to design truly new cores suitable for laptops, at least not for what Jobs was willing to pay.

So at that point, it was either go all-in with PA Semi's PowerPC design, or team up with Intel. With the benefit of hindsight, PA Semi's design team was pretty good, since it went on to become the core of Apple's CPU design team, but in 2004-2005 it must've looked really risky - could a small startup keep up with the juggernaut that was the PC industry?

Did PA Semi license PowerPC? And if so, was it relatively cheap for a startup? It seems weird compared to how strict Intel is with x86.

 

[mr_roboto]

I honestly don't believe that, it's just the history revisioning. "We could, we just didn't wanna" is loser talk to save face.

We shouldn't forget they did the Wii, XBOX and PS3 chips around this time frame too. IBM had a lot of consumer chips on the go but weren't able to deliver. Remember Steve Jobs promising us 3GHz G5's? - They told him they would be able to deliver that, never happened. That's a failure to execute not a decision not to provide it, he never would have said it if they hadn't promised it.

And they went all in on the custom Cell processor with Sony and Toshiba. They clearly had the will to do special projects.

But not with Apple. You have to understand the context. IBM's memories of getting burned must have been quite fresh.

In the early 1990s, when the AIM alliance formed, there were high hopes all around. IBM thought it was a chance to unseat Intel and get market share back from PC clones. Apple and IBM were collaborating on what both thought to be revolutionary next-generation software technology. RISC was going to leave x86 in the dust. Microsoft's gathering power was going to be crushed.

Virtually all of these plans went horribly wrong. The competition didn't roll over and die, instead Intel got busy matching and even beating RISC performance and Microsoft released Windows 95. There seemed to be constant infighting among the AIM partners. The big Apple-IBM software projects (Taligent and Kaleida) were slow motion disasters which consumed immense amounts of money, shipped very little that was ever of use, and are now spoken of in hushed tones, so epic was the scale of the failure. And, frankly, Apple just kept doing things which prioritized the Mac over the hoped-for PowerPC reference platform (remember PPCP, CHRP, PReP?), which pissed off IBM and Motorola quite a bit since they wanted to replicate the PC clone market with an open platform.

So, IBM management had good reasons to be skeptical about Apple's request. Deriving G5 from POWER4 was one thing, designing a new chip for laptops was another. They knew Apple was the only realistic buyer, and it's easy to imagine them not trusting Apple to move enough units. And you may recall that Apple's lead negotiator at the time was a certain S. Jobs, an abrasive man not known to be generous to suppliers or partners. Who had played a big role in how the AIM alliance crumbled.

That chip was PowerPC based and could have superseded the G5 to be honest, sure the SPE design made it complicated to program for but it had potential in later iterations or even Apple could have joined that alliance and been involved in the development process to get a custom version with more PPE's which is exactly what Microsoft did when they wanted a processor for the XBOX 360 (IBM took the PPE design from Cell, tripled it up and gave it to Microsoft in a custom design).

These were collaborations where IBM got offered enough money to get interested, with companies lacking Apple's baggage. And, most importantly, believable paths to high sales figures. Almost guaranteed paths, even.

Remember, this was all before Apple was Apple. At the time, they had one big product, the iPod. They'd managed to prevent the Mac from dying, and people loved what they were doing with OS X, but nobody thought it mattered much either. Microsoft was too dominant for Apple to win back much market share no matter how good OS X was. So even aside from all the bad memories Apple brought to the table, they weren't negotiating from a position of strength.

As to the other points you mention - nothing derived from the console PowerPC chips could have helped Apple. The SPE died after one console generation for a very good reason. The idea which made SPEs different from ordinary processor cores was exactly the thing which made them so difficult to write code for. Fixing that would have made them into things which weren't SPEs any more, so there wasn't any potential, it was just a dead end. (I can elaborate on that if you like.)

I also disagree that PPEs could've replaced the G5. They used substantially less power, but were also far, far slower. What Apple needed was performance similar to G5 at laptop power levels. PPE derivatives would not have gotten them there.

 

[mr_roboto]

Did PA Semi license PowerPC? And if so, was it relatively cheap for a startup? It seems weird compared to how strict Intel is with x86.

Yes, they did. I have no idea how cheap it was, or wasn't.

They weren't the first. @cmaier has been talking about Exponential, a startup which designed a very high speed PowerPC (the x704) in the 1990s.

 

[Quu]

But not with Apple. You have to understand the context. IBM's memories of getting burned must have been quite fresh.

In the early 1990s, when the AIM alliance formed, there were high hopes all around. IBM thought it was a chance to unseat Intel and get market share back from PC clones. Apple and IBM were collaborating on what both thought to be revolutionary next-generation software technology. RISC was going to leave x86 in the dust. Microsoft's gathering power was going to be crushed.

Virtually all of these plans went horribly wrong. The competition didn't roll over and die, instead Intel got busy matching and even beating RISC performance and Microsoft released Windows 95. There seemed to be constant infighting among the AIM partners. The big Apple-IBM software projects (Taligent and Kaleida) were slow motion disasters which consumed immense amounts of money, shipped very little that was ever of use, and are now spoken of in hushed tones, so epic was the scale of the failure. And, frankly, Apple just kept doing things which prioritized the Mac over the hoped-for PowerPC reference platform (remember PPCP, CHRP, PReP?), which pissed off IBM and Motorola quite a bit since they wanted to replicate the PC clone market with an open platform.

So, IBM management had good reasons to be skeptical about Apple's request. Deriving G5 from POWER4 was one thing, designing a new chip for laptops was another. They knew Apple was the only realistic buyer, and it's easy to imagine them not trusting Apple to move enough units. And you may recall that Apple's lead negotiator at the time was a certain S. Jobs, an abrasive man not known to be generous to suppliers or partners. Who had played a big role in how the AIM alliance crumbled.



These were collaborations where IBM got offered enough money to get interested, with companies lacking Apple's baggage. And, most importantly, believable paths to high sales figures. Almost guaranteed paths, even.

Remember, this was all before Apple was Apple. At the time, they had one big product, the iPod. They'd managed to prevent the Mac from dying, and people loved what they were doing with OS X, but nobody thought it mattered much either. Microsoft was too dominant for Apple to win back much market share no matter how good OS X was. So even aside from all the bad memories Apple brought to the table, they weren't negotiating from a position of strength.

As to the other points you mention - nothing derived from the console PowerPC chips could have helped Apple. The SPE died after one console generation for a very good reason. The idea which made SPEs different from ordinary processor cores was exactly the thing which made them so difficult to write code for. Fixing that would have made them into things which weren't SPEs any more, so there wasn't any potential, it was just a dead end. (I can elaborate on that if you like.)

I also disagree that PPEs could've replaced the G5. They used substantially less power, but were also far, far slower. What Apple needed was performance similar to G5 at laptop power levels. PPE derivatives would not have gotten them there.

There's no money in making console chips. It's the whole reason NVIDIA got out of them after the PS3. They said the margins were razer thin. There was no money in it for IBM, fact is IBM just couldn't execute.

I don't need you to elaborate I have all the same knowledge you have already. We're simply drawing different conclusions.

But the whole point of my post that you've glossed over is I was attempting to say there is nothing bad about RISC it was IBM that killed PowerPC. Just like Intel's inability to deliver is pushing Apple to their own silicon now. Earlier in this thread people are blaming RISC and saying CISC is superior. That's just not the case, methods are only as good as the implementations.

 

[Abazigal]

PowerPC processors were a RISC-based processor that everyone claimed was an advanced design, superior to Intel, that could outperform Intel even at lower clock speeds. But as we all know, within a few years on the market, x86 chips kinda... kicked their ***.

Now Apple is switching once again to a RISC architecture that they claim is superior, at least on paper.

What went wrong with PowerPC? How do we know it won’t happen with ARM?

We don’t. As with many things Apple does, adopting them sometimes involves a huge leap of faith. However, given the scale at which Apple currently operates at, I am confident that Apple will spare no expense at making ARM macs work.

 

[ian87w]

PowerPC processors were a RISC-based processor that everyone claimed was an advanced design, superior to Intel, that could outperform Intel even at lower clock speeds. But as we all know, within a few years on the market, x86 chips kinda... kicked their ***.

Now Apple is switching once again to a RISC architecture that they claim is superior, at least on paper.

What went wrong with PowerPC? How do we know it won’t happen with ARM?

Intel wasn't better either, not until the Core Duo chips. Before that, the Pentium 4 was power monsters.

 

[theluggage]

How do we know it won’t happen with ARM?

TLNR with PPC, Apple were reliant on IBM and Motorola to make chips suitable for Macs - with Apple Silicon (and with Apple now being the giant gorilla in the room) they're not even dependent on ARM/soon-to-be-NVIDIA.

...but the real long-term plan must be to end the days of application software being dependent on CPU architecture. If the majority of software is written in Swift/C++/Obj C/whatever and calls OS frameworks for graphics, SIMD processing, locks and semaphores etc. then future CPU switches or major tweaks to the ISA shouldn't affect developers unless they're working on kernel code, compilers or suchlike. Apple are already quite a long way down that road - it helps that they're not obliged to support running 25 year-old binaries.

Earlier in this thread people are blaming RISC and saying CISC is superior. That's just not the case,

RISC vs CISC is so last century.

Newsflash: RISC won. By 1995's Pentium Pro, x86 architecture had embraced RISC technology and consisted of an x86 instruction translator feeding a RISC-like core. That's a horrid kludge that only succeeds because of the industry's dependence on x86 compatibility - but it's that dependence that keeps Intel dominant and gives them the funds and economy of scale to "brute force" the performance problems, and ensures that the x86 versions of software packages get the most intensive development and optimisation.

ARM came to dominate the mobile sector - primarily because of low power consumption, but also because software designed for a desktop is pretty much unusable on a phone or tablet, so there's zero advantage to x86 compatibility.

Today, it's really just the x86-64 and -32 instruction sets vs. everything else, where "everything else" is mostly some variation of "post-RISC" heterogeneous computing that gets its real power from how many extra cores, vector units/shaders/codecs/neural engines etc. you can cram on the die. However much they improve the underlying technology, Intel are going to be limited by the extra die space and power taken up by that x86 instruction translator, that could be used for extra cores or accelerators. Meanwhile, the x86-dependent power base is being nibbled away by increasingly powerful phones and tablets on one side and, if Amazon and co. succeed with their ARM server developments (which mainly consists of trying to educate people that "yes, NGENIX and Node.Js have been on ARM for years"), its going to get nibbled from the other end.

Most of the wailing and gnashing of teeth over Apple Silicon has been caused by the loss of (efficient) x86 virtualisation/bootcamp - while all the evidence shows that even the A12Z should comfortably out-perform all the "non-pro" Macs. In 2020, although there will be winners and losers, gaining iOS compatibility should offset losing Windows overall - and the availability of a proper, high-performance ARM-powered laptop/desktop (that can virtualise x86 Linux) might actually galvanise the uptake of ARM in other areas.

If it fails, it won't be "because RISC" - it will be because Apple stuffs it up by over-pricing, shoddy QA, too much lock-down or failing to schmooze key software/hardware developers to support it.

 

[jeyf]

evaluate what your needs are.
If you going to gift your grand child i would buy a Windows, they will ❤❤love you more.

I would hold off as see where independant reviews are headed

 

[mr_roboto]

But the whole point of my post that you've glossed over is I was attempting to say there is nothing bad about RISC it was IBM that killed PowerPC. Just like Intel's inability to deliver is pushing Apple to their own silicon now. Earlier in this thread people are blaming RISC and saying CISC is superior. That's just not the case, methods are only as good as the implementations.

I wasn't glossing over anything, I was countering an idea which seems as unfounded as worries that CISC is better than RISC. IBM almost certainly had the ability to design the PowerPC Apple wanted, but didn't bite on whatever terms Apple offered. (You yourself brought up the fact that, around the same time, they built what Sony and Microsoft asked them to.)

When you think about it that way, it's clearly different from Intel's recent failures. You want to go into some weird aggro macho mode and claim that trying to understand this history with a little more nuance than "IBM BAD!!!!!" is "loser talk to save face"? Okay, sure, whatever.

 

[Joe The Dragon]

G5s were power hungry only be because it was a server grade chip.

Pentium 4s of the time were equally if not more power hungry.

and Athlon 64 was kick ass as well.

 

[Project Alice]

and Athlon 64 was kick ass as well.

Yes I was gonna say that actually. Back then, I was never an Intel fan in the first place. Even the 32bit Athlons were better then the Pentiums IMO.

 

[Quu]

I wasn't glossing over anything, I was countering an idea which seems as unfounded as worries that CISC is better than RISC. IBM almost certainly had the ability to design the PowerPC Apple wanted, but didn't bite on whatever terms Apple offered. (You yourself brought up the fact that, around the same time, they built what Sony and Microsoft asked them to.)

When you think about it that way, it's clearly different from Intel's recent failures. You want to go into some weird aggro macho mode and claim that trying to understand this history with a little more nuance than "IBM BAD!!!!!" is "loser talk to save face"? Okay, sure, whatever.

Like I said, there's no money in consoles for chip makers. IBM made almost nothing on any of the XBOX, PS3 or Game Cube & Wii processors. As I stated, NVIDIA confirmed this by saying they didn't even bother bidding because the amounts were so low.

And I'm not trying to be macho that's ridiculous, I'm just saying some of this is revisionist history. They couldn't execute on a 3GHz G5 which they told Steve Jobs they would deliver. We're not talking about new architectures, same architecture clocked higher. And they continued to develop PowerPC processors, look at the liquid cooler needed for the later edition G5's it's just ridiculous to claim anything other than they couldn't execute on a roadmap.

The mobile situation is something else entirely. Apple was foolish to lean on IBM for a G5 mobile processor but Apple never promised us a G5 in a laptop. They did promise us a 3GHz G5 because that's what IBM promised Apple they could deliver, they couldn't.

And as I said, Apple could have been a part of Cell and shaped its development, to have a seat at the table. Microsoft came to IBM after the Cell had already been developed and said give us that but with two more PPE cores. Apple could have done the same. Cell never caught on and got killed but if they had a better understanding from a software developers perspective the design could have been altered to be easier to work with.

 

[Deinocheirus]

How do we know it won’t happen with ARM?

The last 13 years of iPhone. That's how.

 

[cmaier]

and Athlon 64 was kick ass as well.

You‘re welcome.