Nuvia: don’t hold your breath

[Colstan]

Oh boy, time to bust out the monkeys! Microsoft Silicon is apparently going to be a thing. According to Mayank Parmar from Windows Latest:

I have also spotted some job listings that suggest the company is building its own Silicon-based ARM chips for client devices. Additionally, I understand that Microsoft is optimizing Windows 12 for Silicon-ARM architecture.

This suggests that Microsoft is building its own ARM-based chips, aiming to compete with Apple’s M chips lineup in terms of performance and efficiency… According to another job listing… the in-house ARM chips are part of the “Microsoft Silicon Team.”

Microsoft to take on Apple M MacBook with new ARM chips & Windows 12

Microsoft is working on new ARM chips with partners and its in-house “Microsoft Silicon team” to rival Apple’s M chips. I have also spotted some job listings that suggest the company is building its own Silicon-based ARM chips for client devices. Additionally, I understand that Microsoft is...

www.windowslatest.com

 

[Citysnaps]

Oh boy, time to bust out the monkeys! Microsoft Silicon is apparently going to be a thing. According to Mayank Parmar from Windows Latest:

Microsoft to take on Apple M MacBook with new ARM chips & Windows 12

Microsoft is working on new ARM chips with partners and its in-house “Microsoft Silicon team” to rival Apple’s M chips. I have also spotted some job listings that suggest the company is building its own Silicon-based ARM chips for client devices. Additionally, I understand that Microsoft is...

www.windowslatest.com

From reading the story it looks like they're looking for a Srouji equivalent and thus starting from scratch. Good luck on that!

 

[Cmaier]

From reading the story it looks like they're looking for a Srouji equivalent and thus starting from scratch. Good luck on that!

I’m available,. if you know anyone at Microsoft let them know.

 

[Citysnaps]

I’m available,. if you know anyone at Microsoft let them know.

I'm also available. And can wow them with my algorithm (written in both APL and Forth) for approximating the area of a square knowing the length of just one side.

 

[Citysnaps]

Cmaier... Getting back to (me) being serious, any thoughts on how long you think it would it take for Microsoft to pull off something equivalent to what Apple was able to accomplish developing their own SOC silicon?

 

[Cmaier]

Cmaier... Getting back to (me) being serious, any thoughts on how long you think it would it take for Microsoft to pull off something equivalent to what Apple was able to accomplish developing their own SOC silicon?

I have no idea who their team is, so I couldn’t guess. They won’t get anywhere unless the core of their team is a set of folks who have already worked together on high-end silicon. There aren’t too many places to get such folks.

 

[mr_roboto]

From reading the story it looks like they're looking for a Srouji equivalent and thus starting from scratch. Good luck on that!

They wouldn't be starting from complete zero. Thanks to Arm-based Surfaces and more notably Xbox, they've been commissioning SoCs for their devices for many years. That's not the same thing as designing your own, but having management structure in place which knows how to spec out SoCs and work closely with a partner that's doing the actual detailed engineering work should make it a lot easier for them to hire their own engineering team. They'll know how such projects work, and what kind of engineering talent they're looking for. They'll have contacts in the industry. They'll be making godfather offers to key managers and engineers who worked on their projects.

Might not sound like much, but this is essentially where Apple started from. The genesis of "Apple" silicon was arguably the several generations of iPod and iPhone chips designed for Apple by Samsung. These chips were nothing special, just bolted together off-the-shelf IP cores, but the point is that Apple started from where Microsoft is: they weren't even bolting the IP cores together themselves, just handing an external design house (which also happened to be the fab house) a spec. Then Apple started hiring and acquiring its own team, and eventually had enough to bring the design of bolt-together SoCs in-house. At the same time, they were working on longer term programs to replace off-the-shelf IP in key areas. So, once the CPU team they'd acquired from PA Semi produced a core better than Arm IP, they rolled it in. Several years later, they did the same with GPU cores.

If the current Microsoft C-suite is smart and disciplined enough to treat this as a very long term program requiring years of careful hiring, acquisition, team-building, and engineering process development, they might do something good. Eventually. But if the C-suite insists on world-beating results ahead of any reasonable schedule, it will almost certainly fail.

 

[Nycturne]

If the current Microsoft C-suite is smart and disciplined enough to treat this as a very long term program requiring years of careful hiring, acquisition, team-building, and engineering process development, they might do something good. Eventually. But if the C-suite insists on world-beating results ahead of any reasonable schedule, it will almost certainly fail.

*snerk*

 

[Colstan]

If the current Microsoft C-suite is smart and disciplined enough to treat this as a very long term program requiring years of careful hiring, acquisition, team-building, and engineering process development, they might do something good. Eventually. But if the C-suite insists on world-beating results ahead of any reasonable schedule, it will almost certainly fail.

I appreciate your serious answer, @mr_roboto, you always add a new angle to the discussion, so I thank you for that. I fully understand that you are speaking in the hypothetical.

But I'm now going to say what @Nycturne wouldn't. Microsoft is the definition of a participation trophy. They don't innovate, they simply extend, extinguish, and make huge profits off of their legacy divisions. They wouldn't know innovation if it bit them in the ass. They've become what IBM was about 15 or so years ago: an over-glorified utility company. Users pay the Microsoft tax much like they pay the electric bill. They're a hugely profitable utility company, but I don't see them as anything more.

Microsoft lost its way when Bill Gates retired. Much like their counterpart, Intel, the company is composed of small fiefdoms, constantly at war with each other. That can be an advantage, if there is a strong king to rule over them, which Gates was. When Microsoft lost that leadership, the fiefdoms were left to their own devices. The same thing happened to Intel, when Andy Grove retired. The resultant rot takes a long time to take hold, but it eventually does. A lot of innovation can come from companies with this structure, but only if they are unified under a strong ruler.

On the other hand, you have companies like AMD and Apple that are cooperative in nature. It takes longer to steer such a ship, but when it happens, the entire corporation can be directed toward a unified goal. Both companies faced imminent death, but turned things around, because nothing motivates like an existential threat. Tim Cook and Lisa Su aren't aggressive leaders like Gates and Grove, but they don't need to be, perhaps shouldn't be.

Those are my personal thoughts on the issue, anyroad, others may disagree.

 

[Hrafn]

I appreciate your serious answer, @mr_roboto, you always add a new angle to the discussion, so I thank you for that. I fully understand that you are speaking in the hypothetical.

But I'm now going to say what @Nycturne wouldn't. Microsoft is the definition of a participation trophy. They don't innovate, they simply extend, extinguish, and make huge profits off of their legacy divisions. They wouldn't know innovation if it bit them in the ass. They've become what IBM was about 15 or so years ago: an over-glorified utility company. Users pay the Microsoft tax much like they pay the electric bill. They're a hugely profitable utility company, but I don't see them as anything more.

Microsoft lost its way when Bill Gates retired. Much like their counterpart, Intel, the company is composed of small fiefdoms, constantly at war with each other. That can be an advantage, if there is a strong king to rule over them, which Gates was. When Microsoft lost that leadership, the fiefdoms were left to their own devices. The same thing happened to Intel, when Andy Grove retired. The resultant rot takes a long time to take hold, but it eventually does. A lot of innovation can come from companies with this structure, but only if they are unified under a strong ruler.

On the other hand, you have companies like AMD and Apple that are cooperative in nature. It takes longer to steer such a ship, but when it happens, the entire corporation can be directed toward a unified goal. Both companies faced imminent death, but turned things around, because nothing motivates like an existential threat. Tim Cook and Lisa Su aren't aggressive leaders like Gates and Grove, but they don't need to be, perhaps shouldn't be.

Those are my personal thoughts on the issue, anyroad, others may disagree.

I'm not convinced they ever "had a way". I can recall both Fear, Uncertainty and Doubt, and Embrace, Extend, Extinguish. For that matter, I remember "Welcome to the social"

I think the truest thing Gates ever said was in an interview with Jobs where he talked about wishing he had any of the creativity Jobs showed. Gates just wanted to make money. Balmer was way worse, don't get me wrong. "Developers, Developers, Developers!"

 

[Colstan]

Balmer was way worse, don't get me wrong. "Developers, Developers, Developers!"

As I said in my previous post:

Oh boy, time to bust out the monkeys!

 

[mr_roboto]

I appreciate your serious answer, @mr_roboto, you always add a new angle to the discussion, so I thank you for that. I fully understand that you are speaking in the hypothetical.

I actually think better of Microsoft's current senior management than you and @Nycturne seem to; Satya Nadella has proven to be an immense improvement over Steve Ballmer. Yes, that's not saying much! And yes, they're still Microsoft. But they have somebody who's at least competent running the show now.

I don't expect them to do great things here, but it won't shake the foundations of my worldview if they do.

 

[Cmaier]

I actually think better of Microsoft's current senior management than you and @Nycturne seem to; Satya Nadella has proven to be an immense improvement over Steve Ballmer. Yes, that's not saying much! And yes, they're still Microsoft. But they have somebody who's at least competent running the show now.

I don't expect them to do great things here, but it won't shake the foundations of my worldview if they do.

I would be quite shocked if they succeed. I don’t know where they will get engineers who know how to do anything other than throw verilog into Synopsys.

 

[Nycturne]

I actually think better of Microsoft's current senior management than you and @Nycturne seem to; Satya Nadella has proven to be an immense improvement over Steve Ballmer. Yes, that's not saying much! And yes, they're still Microsoft. But they have somebody who's at least competent running the show now.

I don't expect them to do great things here, but it won't shake the foundations of my worldview if they do.

Here's the thing, I have a higher opinion of Ballmer than some here, but I got a bit of view into his management style and saw why he wasn't being effective in the role. And it wasn't because of Ballmer's lack of vision. He may have been a cheerleader, but he was smart enough to know that persona is not for making decisions with when it comes to the future of the company.

Some of those reasons Ballmer wasn't effective are still present in the culture of the company. Namely the fiefdoms issue, and the whole idea of bottom-up management. This allows a giant company to act in so many different ways that a company run by a controlling perfectionist like Jobs would never be able to do, but it also means the company itself can be its own worst enemy, and you have VPs that will protect their own vision against other groups and even senior leadership at times where maybe they really shouldn't.

But I'm now going to say what @Nycturne wouldn't. Microsoft is the definition of a participation trophy. They don't innovate, they simply extend, extinguish, and make huge profits off of their legacy divisions. They wouldn't know innovation if it bit them in the ass. They've become what IBM was about 15 or so years ago: an over-glorified utility company. Users pay the Microsoft tax much like they pay the electric bill. They're a hugely profitable utility company, but I don't see them as anything more.

Eh, it's more that I don't see Microsoft being able to be a patient gardener. Maybe I'm still thinking of the "good old days", but there was a reputation for introducing something new, and then quietly dropping it in the dumpster when it failed to be a big business. Microsoft was (too) early to wearables, smartphones, and more. I agree that a good silicon team would take time to ramp up, but that's just generally not the company MO that I remember.

 

[Citysnaps]

I don’t know where they will get engineers who know how to do anything other than throw verilog into Synopsys.

A few times that was a point of questioned legitimacy when potential customers (and potential acquirers) learned that all of our chips were schematic-designed using VI, simulated with our home-grown simulator, and laid out by hand using UCB's Magic with our own library of cells. The good news was our competitors (Analog Devices, Harris Semiconductor, and Xilinx/Altera) could never match our level of performance or functionality.

What was really funny was a couple of major manufacturers of programmable DSP chips thought they could match our level of performance. Not even close.

 

[Cmaier]

A few times that was a point of questioned legitimacy when potential customers (and potential acquirers) learned that all of our chips were schematic-designed using VI, simulated with our home-grown simulator, and laid out by hand using UCB's Magic with our own library of cells. The good news was our competitors (Analog Devices, Harris Semiconductor, and Xilinx/Altera) could never match our level of performance or functionality.

What was really funny was a couple of major manufacturers of programmable DSP chips thought they could match our level of performance. Not even close.

I designed a lot of chips in vi. The original opteron even had a day-of-tape out bug fix that I implemented by hand-editing a .def file to move metal around in vi at the last minute.

I‘m sure large parts of apple’s chips are designed using synthesis, but most of the work is the parts that aren’t. And you need teams that understand the right methodology, and how to actually do the work using that methodology. Most of those folks work at Apple, AMD, and Intel. Maybe there are a couple startups I don’t know about. But you aren’t going to beat apple with a bunch of qualcomm, nvidia, and Arm engineers who decide to jump ship (just for example).

 

[Citysnaps]

Maybe there are a couple startups I don’t know about.

I often wonder about that. We (5 engineers) worked out of our homes for the first two years in spare bedrooms, with used SPARCstation 1s. Hit it big time when we were finally able to rent a 1,500 sq ft dilapidated office space with the railroad tracks right behind us. Have a feeling that would be tough to pull off today.

 

[dada_dave]

I designed a lot of chips in vi. The original opteron even had a day-of-tape out bug fix that I implemented by hand-editing a .def file to move metal around in vi at the last minute.

I‘m sure large parts of apple’s chips are designed using synthesis, but most of the work is the parts that aren’t. And you need teams that understand the right methodology, and how to actually do the work using that methodology. Most of those folks work at Apple, AMD, and Intel. Maybe there are a couple startups I don’t know about. But you aren’t going to beat apple with a bunch of qualcomm, nvidia, and Arm engineers who decide to jump ship (just for example).

Nah all you by-hand designers will be replaced by AI deep learning algorithms which I’m sure don’t have any wacky corner cases nor need to be checked by experienced people

 

[Cmaier]

Nah all you by-hand designers will be replaced by AI deep learning algorithms which I’m sure don’t have any wacky corner cases nor need to be checked by experienced people

In my final years at AMD, I was in charge of design methodology. I wrote a lot of our tools. If I was still in the industry, I absolutely would be working on getting AI to replace me. I think it can be done, but only if the people working on the tool truly understand how design works, including all the physics that can get you in trouble.

Heck, there are days when I wake up and think “I should write that tool” even now.

 

[Citysnaps]

Heck, there are days when I wake up and think “I should write that tool” even now.

That's an awesome idea. Likely realizable. You've got the background that counts. I say go for it!