Running Fusion with Rosetta 2 on Apple ARM-based M1 Chipsets

[keqingsong]

July 5, 2023 Update 

Hello all, the current Insider Build of Fusion now has native Apple Silicon support, and as you read some of the latest comments in this thread, the results are looking great. Our goal is make it available to everyone by our next product update, which should be happening towards the end of the month. If you want to try it now, you can sign up to join our Insider Program, and get access to the Insider Build. Keep in mind that once you become a member, you are under NDA and cannot sharing information publicly, with the exception of this particular project since it is already public knowledge. 

Click this link to sign-up and join: https://www.autodesk.com/campaigns/fusion-360/insider-program

Thank you to those you have who've expressed interested and have been testing it! Please continue to let us know about your experiences. 
___

November 23, 2022 Update

As you know we have been working closely with Apple on native support for Fusion on Apple Silicon Chipsets (i.e. M1 & M2). We are pleased to announce that we expect to achieve full native support by Summer of 2023.

As we have explained in this thread, the delay is a result of the need to ensure 100% compatibility between components from over 100 3rd party vendors including Autodesk.

If you want to access this functionality as soon as possible, please consider joining the Insider Program. If you have any questions on the topic please contact @Rajkumar.ilanchelian.

October 28, 2021 Update

Hey all, thanks again for the passionate discussion here. Even though we haven't chimed into this thread as much as we wanted, we are reading every single response and are actively working on getting Fusion to be natively supported on the new Apple chipset. Here's what I know from talking to the development teams: 
We are actively working on getting native support. This is still going to take some time because Fusion uses a multitude of services to work the way it does (Autodesk-owned as well as 3rd party) many of which are also not natively supported on M1 chipsets yet. We are collaborating closely with those teams to taking the necessary steps to ensure that the services we use are also natively supported. There is a lot of passion internally to get this done as well, so we definitely feel you. Again, I can't not say when this will happen, but as soon as we have something more concrete to share, we'll be sure to update you all. 

April 29, 2021 Update

We've been actively working on resolving the issues mentioned below and are glad to report that these issues no longer exist when running Fusion on the M1 chipset. We are also working closely with Apple and are in the process of certifying Fusion as 100% compatible running on M1 chipsets via Rosetta 2. 

In terms of running Fusion natively on the M1 chipset without Rosetta 2, we are still working towards this goal but is going to take some time to reach. We are confident to say that running Fusion on the M1 chipset via Rosetta 2 should be indistinguishable from running it on an Intel-based chipset, if not faster. 

If there are specific issues you've experience and are not mentioned below (strike-through items), please chime in and respond to this thread so we are aware and can look into it ASAP. Thank you for your continued support! 

___

Apple's original press release 

We are delighted to see Fusion being featured in the most recent Apple ARM-based M1 Macbook Pro announcement. Although Fusion isn’t natively compatible on Apple’s new M1 chipset architecture yet, Rosetta 2 should enable you to run Fusion*. We will be sure to keep you posted on our progress towards support of Apple’s new line of chipsets.   

-  
   
* While much of Fusion 360 works as expected under Rosetta 2, we discovered that some Fusion 360 components were not yet compatible. If you run Fusion 360 using Rosetta 2, you may experience issues in these areas:  

·         Switching Team Hubs in the Data Panel  

·         Insert from McMaster-Carr  

·         Explore Generative Design and Electronics Cooling Simulation results  

·         ECAD Tool Libraries and Content Manager  

·         Local Simulation Solves utilizing NASTRAN  

If you rely on the impacted areas for your work, we recommend you to stay on Intel-based Macs until we have these issues sorted out.   

[bmxjeff]

This is the most entertaining conversation on the forums! I love it..

My last MacBook Pro was intel. It had much more trouble than my new Mac Studio with an M1 chip. The problem was the processor was too slow for what I'm using Fusion for. That's not Autodesk's fault. Solidworks requires resources too.... If you did switch softwares, and you're still using your Mac, then you are not using Fusion anyway, and Rhino is the right place for you.  

1. You bought an M1 chip for software that isn't ready. That's on you. : )

2. College Students typically don't pay for the software anyway... To call yourself a customer is a stretch.

3. If you graduate, you'll use whatever software the new company tells you too. I hope it's Creo with Windchill, then your new complaints will be justified. 

I would love an update on the progress too, but I'm programming parts for a 5 axis mill with over 1 hundred operations with many that are dependent on the previous ops. Fusion, on Rosetta, is doing really well. And my alternative is 20k and a new computer. So, I'm pretty psyched... 

If you by a computer for any engineering software, you must make sure it's compatible. And We all know what an emulator is. : )   

[maruska]

Yes. What I find frustrating is that Apple announced this in 2017, and it started with the shift away from OpenGL. This has been a long-time coming. Rosetta 2 won't be around forever. The original Rosetta (PowerPC emulation for x86) had just over a 4 year lifecycle - so we might be more than halfway through this one.

[albertson.chris]

I keep checking in to see if Fusion 360 runs on M1/M2.   I'm a bit shocked that it has been five years since Apple announced the switch.  I would have assumed Autodesk would have bought M1 based computers as soon as they were available and gone right to work.

[albertson.chris]

A followup would be nice.   

My current best guess is that Fusion 360 goes away when Apple drops support for Rosetta as it will never be a native app.

A monthly status update would keep people from looking around for other CAD solutions.

[jeff_strater]

@albertson.chris - that is not the case at all.  If you read the first post in this thread (from @keqingsong ), you will see that we are actively working on making Fusion be a native application on Mac.  Keqing also explains why this is taking longer that we would all want:  Fusion is built on a very large technology stack, each component of which has to first be made to be a native library.  It has to be done bottom up.  And, contrary to some beliefs, it is not just a case of flipping a compiler switch and rebuilding.  We have to guarantee, at each level, that the component is functioning correctly, and fix those issues as they arise (and, they do arise, the compilers are not 100% compatible in all ways).  Also, some components come from outside of Autodesk, which adds more delays.  And, it has to be 100% complete.  That means everything:  Simulation, Manufacturing, Drawings, Electronics, etc.

[copumpkin]

I think that last part is the most frustrating one for us. I’d guess that any individual user you poll doesn’t use half those workspaces, so I for example use CAM so care about Manufacture. If hypothetically some obscure third party dependency of the Electronics stuff is buggy on ARM, that doesn’t affect anything I do, but I don’t get what I need until it gets fixed. Bundling all the workspaces into one single app is cool and I like the design in principle, but when it holds up a release by literally several years there has to be another way, right? I’d take a preview download with half the workspaces greyed out if that’s what it took, but this monolithic “everything needs to be perfect before you see anything” feels contrary to how most software works in practice. I’ll happily take a buggy (short of data loss) “iterate early and often” release plan for something like this. We don’t expect it to be perfect and mostly do understand the challenges of software development so are not holding Autodesk to a crazy standard, but it’s rare to see anything taking years in computing so it’s hard not to see this as either a prioritization issue or a misplaced expectations issue.

[oyvindTMNSU]

A problem here is that there has not been given any timeline at all. It must be possible to share what the status is by today, at least in the insider forum! More than just "we are working on it". It MUST be a more detailed timeline to share!

[jeff_strater]

"it’s hard not to see this as either a prioritization issue or a misplaced expectations issue."  I can assure you it is not the first.  This is a priority for us, but it is not one that we can just throw more people at and make it happen faster.  Mythical man-month and all...

"If hypothetically some obscure third party dependency of the Electronics stuff is buggy on ARM, that doesn’t affect anything I do, but I don’t get what I need until it gets fixed".  Yes, that is correct.  It might not affect anything you do, but there are tens of thousands of folks that it would affect.  So, we just don't have that option, in all honesty.  For better or worse, "Bundling all the workspaces into one single app" is what Fusion is, and to some extent, why it is successful.  So, we have no choice, in the production environment, of, say, releasing without Electronics or Drawings.  Surely you see that, right?  Similarly, we cannot release "a buggy (short of data loss) “iterate early and often” release plan for something like this" in the production environment, either.  People rely on this software.  As it is, we violate that trust way too often, unintentionally.  To do it intentionally would not be wise, I think.

"I’d take a preview download with half the workspaces greyed out".  While that is not available today, I do encourage you to join our Insider program:   https://feedback.autodesk.com/key/Fusion360Insider .  At some point down the road, I fully expect to see a Native M1 Fusion appear first in that environment.

[albertson.chris]

Apple announced the move to ARM in June 2020. After two years there are
not even screenshots or a blog post boasting of "improved performance,
coming soon"

I've worked in software development for a few decades. I think if my boss
asked me to "Get a Mac version going ASAP" in a month, I could use Apple's
x-code to mock up a prototype of the user interface with stubbed-out
functionality. And then start filling in functions and I could show him
what it would look like in 6 or 8 weeks. But apparently, they don't have
even this (because it does not exist if there are no pictures.)

The obvious conclusion is they are not trying. I suspect that they simply
do not want to spend the development budget on this because the user base
is too small.

Maybe the management at Autodesk is smart, their most profitable plan is to
do and say nothing and let the product die when Apple removes Rosetta.

A single link could prove me wrong.

[jeff_strater]

"A problem here is that there has not been given any timeline at all"  We are legally prohibited from giving these kinds of timelines.  It has to do with "revenue recognition" laws that I don't pretend to understand.

[oyvindTMNSU]

Ok  but are we talking more year? Does anyone even know?

[jeff_strater]

the same laws apply.  We cannot say.  Sorry, that is just the way it is.

[copumpkin]

Not denying it but struggling to understand how other companies can pre-announce work in progress without breaking this law and Autodesk can’t. I’m guessing it’s more of a tax thing that internal lawyers have decided is the right strategy for the company?

Either way, I think the core of our discontent is that it’s hard to believe this is a priority when tons of other feature work has been launching highly nontrivial improvements in the intervening years while it feels like pulling teeth to get updates on progress of this port. As the other poster said, an occasional blog post saying “we’re working on it, can’t give you a date yet but it’s looking good and we’re excited to get it into your hands” is a far cry from us having to poke a defunct thread like this every few months to maybe get another trickle of info about it. There are dozens of other preview features available in the app, often with long blog posts about them, but the only updates we get here are if someone posts something inflammatory (which I don’t condone btw) and gets a reaction out of a staff member on this thread to set the record straight. It’s easy for nonsense to breed if the only people who know anything don’t tell us. Maybe the technical difficulties are indeed insurmountable and a multi-year endeavor but I’m pretty sure that communication about the whole thing could be vastly improved.

(On your other note, thanks, I’ll sign up for the insider program and cross my fingers)

[brandonXNQ6E]

Someone is gaslighting you. "Revenue Recognition" is an accounting principal, not a law. It only pertains to law when it comes to taxes, and that is just defining what tax year income is applied to.

The only "law" that could be violated with a timeline is with the SEC. If Autodesk knows it is not going to release the product and publishing a timeline falsely implies that they are and their stock value increases because of it.

As copumpkin said, a lot of companies publish timelines and release dates, and deal with delays (game publishers!). Autodesk is making a choice, but if it is law preventing it, then it does not bode well.

[albertson.chris]

Yes, you nailed it!

"Law" only applies if Autodesk is gaslighting and knows they are. Autodesk
says the law applies. Draw your own conclusion.

My opinion: I like to quote. Linus Trevolds (He is the guy behind the
Linux kernel). When someone suggests an idea or says what he is working on,
he stops them and says, "Show me the code." This cuts out all the BS.
Either they can point to it or they can't. 90% just go away after he
says that. The fact that Autodesk can't point to anything says
something. People who are excited about their work find ways to show it
off.

Gaslighting might be Autodesk's most profitable solution. It costs them
nothing and what do they do if they do release a native App? Very little
really, maybe a couple of hundred more subscriptions at most. Maybe
$110,000 per month more income. Is that worth the cost? Likely not.
They likely keep more Mac users from leaving by gaslighting than they would
gain by developing the new Mac app. A smart plan might be to stall for
time until those numbers flip.

I know the development cost would be high. My guess is that the
source code for Fusion360 is what we call "a dependency hell of the fourth
order" and it was NEVER designed to be portable. And now they realize that
a professional-level macos product REALLY needs to be written with Xcode
and be using Apple "core Services" and the road from "the fourth ring hell"
to a native app is EXPENSIVE. Hence, plan B: "Stall for time".

One thing about coding for Mac is that Apple likes to radically shake up
the foundation frequently. Those who build directly on this are doomed to
one of the other levels of hell. So to prevent this, Apple publishes some
stable APIs called "core Services". Smart developers use these. My
educated guess is that Autodesk's programmers built Fusion360 on stuff that
rests directly in the basic foundation, now they are paying the price.

BELIEVE ME, if you do it "right" the switch to Apple silicone is very
easy. I write stuff that runs on Intel and Arm, and it is not rocket
science *if you planned ahead from day one*.

It's been years since I read Dante, But I think in the 4th ring you are
trapped head down in a vat of boiling oil, or something like that, not a
fun place.

[macmanpb]

@jeff_strater 

Hello Jeff,
now it has been announced that support for MacOS catalina will be discontinued in March 2023. I can no longer install a newer OS on my Mac! What should users do who currently own such a Mac and who have been waiting for Fusion360 to run natively on an M1 processor for a long time. That means I now have to buy a current Mac with an M1 processor in order to still get Fusion360 support from March 2023, but with the disadvantage that Funsion360 can only be operated on it under Rosetta. That is sick! Sorry Jeff!

[davidZ8URT]

---

@jeff_strater wrote:

"A problem here is that there has not been given any timeline at all"  We are legally prohibited from giving these kinds of timelines.  It has to do with "revenue recognition" laws that I don't pretend to understand.

---

When a company starts to explain the lack of product schedules with "legal laws" you know it's screwed up and you will never see this piece of software. 
What Autodesk tries to do here is to keep their mac customers in the "dark" and pick up the yearly subscription so long as possible.

[maruska]

What? We got finally dark mode?

Oh - never mind. Ha

[brandonXNQ6E]

Even Adobe has managed to get their full suite of apps running natively on M1.  Considering how massive that is and how much of a dinosaur Adobe is, that is no small feat.

I think you nailed it as well.  Autodesk has been coasting on old code to keep it simple for themselves, and now it is biting them in the butt.  Unfortunately, I think Autodesk will just move on we (the Mac based customers) are going to be the ones to pay.  They are either going to drop support and we have until Rosetta 2 expires, wait an eternity for actual native support, or be offered some crippled web based alternative.  Considering Fusion 360 is the ONLY CAD program that is any good on Mac, we are screwed.

I saw this exact same scenario play out 3 years ago with MYOB (a Mac native accounting software) when Apple quit supporting 32-bit.  They told us for two years that they were working on a 64-bit app, that it was very important to them, yada yada yada.  Then updates stop and lots of silence.  Finally, after angry customers demanding anwsers, MYOB said "Sorry, but the code is too old and difficult to upgrade so this won't be happening.  However, we just released a new web based subscription program..."

[albertson.chris]

Apparently we Mac users need to be looking for an "off ramp".  We have time, a couple of years, I'd guess.   Seriously, the "law" they speak of applies only to intentional gaslighting, they say the law applies, so "go figure."

There might be really good solutions.

Does anyone remember Eagle CAD.  It was bought by Autodesk and then folded into Fusion360.  It too was old code and the company was getting greedy and started adding restrictions to their free version.  The company unsurprisingly lost market share and evenly had to sell out to Autodesk.   But before they sold, already users were moving to KiCAD and now that Eagle CAD is gone and absorbed into Fusion360, KiCAD is the leading product.  What caused this was the Physics lab CERN decided to donate some serious funding to KiCAD and this allowed the OpenSourse KiCAD to jump ahead of Eagle in terms of functionality and usability.

What we can hope for is that FreeCAD takes the opportunity to fill the hole that Autodesk seems to be leaving.   I have it on my Mac as I type and (1) it's free and (2) it seems to run well and (3) the source code is available so it can be built on "anything".

FreeCAD has most of the functionality of the free version of Fusion360 plus it does at least one thing better. The parametric functionality is implemented as a spreadsheet, where a dimension can be taken from a spreadsheet cell.  This allows you to use tabular data and quite complex calculations. For example to make stairs or gears

With Fusion360 it is very hard to document the interrelation of the computed dimensions, but a spreadsheet consolidates them to one place where you can see the big picture all at one and even make changes.

So far every Fusion360 project I save as .step opens well in FreeCAD, some are quite large with hundreds of parts.

CAM tool path generation in FreeCAD is primitive compared to Fusion but OK for "2.5 D" milling.  They are making slow progress with manufacturing.

My hope is for some organization like CERN to dump money into FreeCAD to enable the quantum leap the KiCAD made.

Mac Users should look at FeeCAD and start evaluating it.   If you have an M1 or M2 powered Mac, try building from source