Rosetta 2 creator leaves Apple to work on Lean full-time

[Xiao_Xi]

Cameron Zwarich, creator of Rosetta 2, is joining Lean FRO to improve Lean's code generator.

I am thrilled to welcome Cameron Zwarich to the Lean FRO! | Leonardo de Moura

I am thrilled to welcome Cameron Zwarich to the Lean FRO! As the brilliant creator of Rosetta 2 and an exceptional software developer with over 15 years of experience at Apple specializing in low-level systems software, Cameron will focus on enhancing Lean's code generator. I can’t wait to see...

www.linkedin.com

Cameron has commented on Hacker News:

I’m looking forward to working with everyone else at the Lean FRO and the wider Lean community to help make Lean even better.
My background is in mathematics and I’ve had an interest in interactive theorem provers since before I was ever a professional software engineer, so it’s a bit of a dream come true to be able to pursue this full-time.

By the way, Cameron has also answered some questions about Rosetta 2 on Hacker News.

I was the only person working on it for ~2 years, and I wrote the majority of the code in the first version that shipped. That said, I’m definitely glad that I eventually found someone else (and later a whole team) to work on it with me, and it wouldn’t have been as successful without that.
When people think of a binary translator, they usually just think of the ISA aspects, as opposed to the complicated interactions with the OS etc. that can consume just as much (or even more) engineering effort overall.

Rosetta 2 creator leaves Apple to work on Lean full-time | Hacker News

news.ycombinator.com

 

[tenthousandthings]

By the way, Cameron has also answered some questions about Rosetta 2 on Hacker News.

Rosetta 2 creator leaves Apple to work on Lean full-time | Hacker News

news.ycombinator.com

Interesting glimpse of the inside. He says he worked on it for about two years by himself before “someone else” started to work with him, and then “later a whole team.”

Rosetta 2 was announced June 2020, so I think it’s safe to say the absolute latest he could have started work on it was June 2018, and it was probably even earlier, like 2017.

This just confirms what common sense tells us, that the initial decision to move forward with Apple silicon was probably made in 2017.

While I think the primary motive behind that decision was about performance and efficiency in MacBook Air and Pro, I also think things like Apple’s experience with the 2013 Mac Pro, and being forced to implement a kludge for the 5K Retina display in the 2015 iMac, made the decision easier.

 

[MayaUser]

Big companies always make decisions and think with 4-5 years ahead

 

[senttoschool]

Rosetta 2 was announced June 2020, so I think it’s safe to say the absolute latest he could have started work on it was June 2018, and it was probably even earlier, like 2017.

This just confirms what common sense tells us, that the initial decision to move forward with Apple silicon was probably made in 2017.

Why do you say it was June 2018 or 2017 based on his "I was been working on it alone for 2 years" comment?

He could have easily been working on it alone in 2015, and got more help in 2017?

I think the decision to move forward with Apple Silicon was likely made earlier than 2017. Starting in 2017 would give them about 3 years to design the base, Pro, Max, Ultra chips, build Rosetta 2, transition app developers, prepare macOS, design the new MacBook Pros around new chips, etc.

There is no way they could have done all that in 3 years in my opinion. It takes them 3-4 years just to design one generation of M chip. The first generation was likely even longer.

I'm thinking 2015, unless there is a reliable source that suggests otherwise.

 

[Homy]

The first Apple Silicon, A4, was released in 2010 in iPad and iPhone 4.

 

[Zest28]

Interesting glimpse of the inside. He says he worked on it for about two years by himself before “someone else” started to work with him, and then “later a whole team.”

Rosetta 2 was announced June 2020, so I think it’s safe to say the absolute latest he could have started work on it was June 2018, and it was probably even earlier, like 2017.

This just confirms what common sense tells us, that the initial decision to move forward with Apple silicon was probably made in 2017.

While I think the primary motive behind that decision was about performance and efficiency in MacBook Air and Pro, I also think things like Apple’s experience with the 2013 Mac Pro, and being forced to implement a kludge for the 5K Retina display in the 2015 iMac, made the decision easier.

What you mean “moving forward with Apple silicon” in 2017? Apple Silicon already existed with iPhone and the iPad for a very long time.

And the decision most likely came from the fact that iPads were faster than the 13” Intel MBA and 13” Intel MBP and stayed cool at the same time. Might as well put those iPad chips in the MBA and MBP like what some of us were saying back then (including me).

The first “Apple Silicon” Mac was the Mac Mini based with the A12Z chip from the iPad. So that is also how you see that Apple simply took the iPad chip and M1 is basically what the A14X would have been.

 

[Xiao_Xi]

I'm thinking 2015, unless there is a reliable source that suggests otherwise.

Cameron has commented on what projects he was involved in while at Apple.

I have been so lucky at Apple to have been in the right place at the right time for several projects: relatively early iPhone team, original iPad team, involved in the GCC -> Clang transition, involved in the 64-bit ARM transition, involved in early Apple Watch development, first engineer working full-time on the Apple silicon transition for the Mac, etc.

 

[tenthousandthings]

What you mean “moving forward with Apple silicon” in 2017? Apple Silicon already existed with iPhone and the iPad for a very long time.

And the decision most likely came from the fact that iPads were faster than the 13” Intel MBA and 13” Intel MBP and stayed cool at the same time. Might as well put those iPad chips in the MBA and MBP like what some of us were saying back then (including me).

The first “Apple Silicon” Mac was the Mac Mini based with the A12Z chip from the iPad. So that is also how you see that Apple simply took the iPad chip and M1 is basically what the A14X would have been.

I meant “move forward with Apple silicon” in Macs, but you know that. The fact the M-series built on a decade of experience with the A-series is a given.

The fact the M1 replaced A14X is also a given, since that’s what happened in the iPad Pro.

But your argument fails to account for the M1/M2 Pro/Max/Ultra. That is a whole new class of Apple Silicon (capital S), beyond the world of A14/M1 and A15/M2. Moving forward with that is the decision we’re talking about here. 2017 is four years ahead of that launch, in October 2021.

 

[tenthousandthings]

Why do you say it was June 2018 or 2017 based on his "I was been working on it alone for 2 years" comment?

He could have easily been working on it alone in 2015, and got more help in 2017?

I think the decision to move forward with Apple Silicon was likely made earlier than 2017. Starting in 2017 would give them about 3 years to design the base, Pro, Max, Ultra chips, build Rosetta 2, transition app developers, prepare macOS, design the new MacBook Pros around new chips, etc.

There is no way they could have done all that in 3 years in my opinion. It takes them 3-4 years just to design one generation of M chip. The first generation was likely even longer.

I'm thinking 2015, unless there is a reliable source that suggests otherwise.

Sure, it could be earlier, and 2015 is plausible, I just meant 2017 is the latest it could have been, and note that 2017 is four years ahead of the M1 Pro/Max, not three.

@Zest28 comment above highlights the iPad Pro silicon as the forerunner of M1, A12Z is just A12X with one additional GPU core, and it’s probably safe to assume the early development of Rosetta 2 was on A12X (launched October 2018, with prototypes in early 2017). So that fits.

The thing about the A14/M1 and M1 Pro/Max/Ultra family itself is that it was on a new, first-generation 5nm process. That means a new fab was built for it, with new equipment, I don’t know when that opened but there is probably a press release about it. They usually hold a ceremony. TSMC has said that it takes about two years for their customers to go from design to volume production on a new process node, let’s add a year to build up volume, so about three years from design to launch, like you say.

 

[Lounge vibes 05]

I think the decision to move forward with Apple Silicon was likely made earlier than 2017. Starting in 2017 would give them about 3 years to design the base, Pro, Max, Ultra chips, build Rosetta 2, transition app developers, prepare macOS, design the new MacBook Pros around new chips, etc.

There is no way they could have done all that in 3 years in my opinion. It takes them 3-4 years just to design one generation of M chip. The first generation was likely even longer.

I'm thinking 2015, unless there is a reliable source that suggests otherwise.

Former Intel Engineer Claims Buggy Skylake Chips Hastened Apple's Switch to Custom Silicon

At this week's WWDC, Apple confirmed its plan to switch from Intel to custom processors for its Macs over a two-year transition period. Apple...

www.macrumors.com

So around the same time as the 2016 MacBook pros, which makes a lot of sense.
Although there were rumors about Apple switching the Mac to their own custom processors going all the way back to around the launch of the original iPad.
As for how long it took to develop the M1 family, obviously outside of Apple no one really knows, but it probably didn’t take that long given that the M1 is in concept basically an A14X, and Apple had been making “A-X” chips since 2012.

 

[SkippyThorson]

Cameron Zwarich, creator of Rosetta 2, is joining Lean FRO to improve Lean's code generator.

I am thrilled to welcome Cameron Zwarich to the Lean FRO! | Leonardo de Moura

I am thrilled to welcome Cameron Zwarich to the Lean FRO! As the brilliant creator of Rosetta 2 and an exceptional software developer with over 15 years of experience at Apple specializing in low-level systems software, Cameron will focus on enhancing Lean's code generator. I can’t wait to see...

www.linkedin.com

Cameron has commented on Hacker News:


By the way, Cameron has also answered some questions about Rosetta 2 on Hacker News.

Rosetta 2 creator leaves Apple to work on Lean full-time | Hacker News

news.ycombinator.com

I’m not entirely sure how this wasn’t a front page @MacRumors article. What an incredibly talented person; which seems like quite a loss for Apple.

This single employee is as significant in both the Intel and M-series transitions as much of the top brass. Crazy to think with everyone there, that they worked on Rosetta ‘solo’.