Intel has a toxic culture based on condescension and paranoia. They treat their contractors terribly. Given probably half of Intel runs on contractors, that was surprising to me. They haven't changed much in the last year.
Got a tip for us?
Let us know
Become a MacRumors Supporter for $50/year with no ads, ability to filter front page stories, and private forums.
Key M1 Mac Engineer Departs Apple for Intel
- Thread starter MacRumors
- Start date
- Sort by reaction score
That's not exactly how I read it. The CCG is a business segment, this is discussing client segments. Specifically it's referring to "all client segments", plural. So maybe all business segments related to PCs, but that seems a weird way to put it. Maybe all tiers of PC processor.
Anyway, the meat of my question wasn't about the definition of client, it was about the SoC bit. I'm not seeing much in the way of SoCs in Intel's lineup except in their Altera line. Digging a bit deeper, it looks like they're still producing Atoms they refer to as SoCs, so maybe those?
That wasn't a 'cocky' move that was far more a goose the stock price and profit margins higher move.
Intel needed to both buy lots more 14 -> 10nm (now Intel 7) equipment ( progressing past 14nm meant cranking up the multipatterning steps which drives a longer gestation process for the chips. Need more fab machines just to keep the same volume output of wafers per month.). At same time needed more EUV fab machines to get around the (old 7nm ) Intel 4 (and lower ) path finding problems and to grow productions. Those machines would be pure cost centers for a substantive amount of time. (and therefore would drive down profit margins ).
Pragmatically Intel could have afforded to do both. The stock price would have dropped (well earlier ... it sagged anyway so really only ducked the issue over the relative short term ) . Wall Street general doesn't care about long term strategies if it means a dividend cut in the short term.
It was also not 'cocky' because to some extent Intel didn't beat on its fab folks coming through. Somewhat reasonable because of stumbles on the transition to 14nm and the giant crater dunk on initial 10nm forays. No, Intel pushed EUV ASML equipment farther into the future because management wasn't 'cocky' sure it was going to work and they went with the "less risky' non-buy.
There is a "next gen EUV" tech that ASML hasn't shipped to anyone yet. For the current EUV tech from ASML .... Intel is always going to be behind. Never going to catch up in high volume ability. That horse left the barn in 2019-2020.
The next gen EUV is a technological inflection point. There is a chance for Intel to catch up there if don't screw up the pathfinding to solutions using that tech. TSMC and Samsung have way too much money that they'll get their hands on the tech too, so if Intel stumbles , then Intel will fall behind again. But the issue is that TSMC can't race away from Intel with stuff they don't have.
The ASML equipment doesn't matter as much as Intel getting a viable Foundry services business up and running. If there is a client that can "consume" next gen EUV wafers if the mainstream Intel products can't then they would have a way of walking forward until they do. Also could have a mix of different tuned fab solutions to buy. What miight not be a good match for Intel CPUs could be a decent enough match for something else. Pathfinding could find something for someone else first and the Intel products next.
I suspect that is somewhat backwards. Going down into the angstroms range is going to hard for everyone ( Intel , TSMC , Intel , etc. ). There is just going to have to be excellence just to get things to work well at all. Let along "way better than the others guys". And the designs will have to be very good for the somewhat more narrower areas they are targeted at. ( big issue is that Intel probably is going to have to give some of on the build everything for everybody path they have been on. )
Apple show up as a fab client ... again that would be highly dependent upon there being a viable and a reliable (with a vetted multi-year) track record for Foundry services. Apple won't be coming back because Intel has the sexiest marketing slides for their Foundry. Intel has talk about getting info Foundry services multiple times and botched a number of effort. They have a reputation hole to dig out of.
Intel can now find out just how many years behind they really are going forward.
Intel has been running their chips hot, and I wouldn't be surprised if KS was the byproduct of scraping the cream. If they can get a flagship Arc out in reasonable numbers they will be positioned to negotiate with nvidia to integrate Deep Link. That alone resolves would resolve any concern about other CPUs chipping away at their lead.
Hopefully this will not have an impact on Apple in the future
M1 is already done deal. Apple will find someone else who understands what he did and improve on it. E = MC² took a long time to figure out, and such discoveries far and few between. But once the secret is out, anyone can start using and deriving from it. This back and forth never ends. Competition in the industry is good for consumers, as each company races to make faster processors. It thwarts complacency. If a company feels like it's holding all the cards what's the hurry to change anything?
Hopefully it will make Apple find something that tops M1 and make it look like child's play.
What? The guy was offered a high level position at another company, and decided to advance in his career. You’re right, its not rocket science. There can only be so many leadership positions at one company. Srouji currently holds the top position on the silicon team, so if you aspire to have a title like his, then you need to go to another company.
I think there are MUCH safer and easier ways to “prop up AAPL stock” than embarking on a platform transition. Thats literally the opposite of what a stock obsessed CEO would do….lol. You do realize that not only investing in the mere prospect of a Mac architecture transition, and then actually pulling the trigger and DOING it, is probably one of the riskiest things you could do to a mature platform that is selling more than ever in its 30 year history.
M1 is proof of the EXACT OPPOSITE of what you’re accusing Tim Cook of. Thank god someone intelligent and level headed like Tim Cook is running the company, and not an irrational fool.
That's an increasingly archaic distinction; especially inside of the Intel client platform division. Maybe that persists in the server group, but the laptop solutions are primarily SoCs.
At this point Intel has put Thunderbolt inside of the "CPU" package. Ethernet.. on the CPU paciage. Wifi ... base unit ... on the CPU package.
Going to "CPU" packages with tiled/chiplet implementations that is likely to get even more integrated in 2-3 generations down the road.
If go back to Intel even 5-6 this wasn't the case. When Intel was selling lots more discrete chip bundles to customers that was a distinction. When there were viable 3rd party chipset vendors even more. But in 2020+ that is a "terminology" that is just lost in significant distinction.
AMD desktop Ryzen solutions are GPU less. Pragmatically all of Intel's client desktop and laptop solutions have iGPUs. iGPUs always present is in the SoC-zone.
Picking someone who was in charge of the Mac transition I suspect has little to do with ARM instruction set and more so about putting together better SoC package combos. Which set of technology implementation tiles does Intel integrate into a product groupings ?
x86_64 certainly could do with a dump of the constipation (more than complexity) there is present. Old dead ends from the 1990's . What do they toss and what do they keep would be more helping than doing an ARM "swap out" . That is in part what go 'sorted' in the Mac transition process. Some stuff got dumped on the transition.
For the client platform group .... most of what they sell is Laptops "CPU" . The whole high skew to super modularity on the desktop isn't the driving force for that division anymore. As CTO of division that is the 'ball' needs to be primarily focused on.
The server platfrom division may not weave GPUs into the "CPU" package , but there are probably specific accelerators that will be "tiled' into a broader range of server "CPU" packages in the generations 2-3 iterations out.
Servers tend to have more variable I/O components attached so harder to all the possible combos on a single 'chip' package, but it isn't going to be immune to the SoC effect of the CPU packages subsuming some stuff off the standard server logic board over time.
One of INtel's problems is that they try to build everything for everybody. That probably isn't tractable going forward. Not shrinking down to the overly narrow product line that Apple pursues, but Intel needs to get better at choosing where to apply more focus and which edge cases aren't worth doing solely internally.
It’s not like production of Apple Silicon will just suddenly stop. The architecture has already been designed and Apple will be able to keep improving it for many years to come. Plus there are many other members to the team who know what’s going on and learned from him.
Apple will keep chugging along pumping out updated versions of the chips and it will take Intel years to put out their own.
This is such a non issue.
Apple will keep chugging along pumping out updated versions of the chips and it will take Intel years to put out their own.
This is such a non issue.
Yeah I thought the same because they mostly use SoC for their integrated Altera line, but it looks like some of their Atom parts are marketed that way too:
Intel Atom® C Processors Series - Microserver Processors
Intel Atom® C processor family for microservers is designed to reduce energy demands, be configured to fit server specs, and built-in security.
www.intel.com
I think they’re forced to, though. It’s the nature of the beast, any market they decide NOT to compete in is a market someone else has an opportunity to make an effort, learn from the experience, iterate and eventually challenge Intel. Lucky for them, that last time that happened, the company that gained from the experience only makes chips for ONE company
They understand they don’t have to excel everywhere, they just have to be good enough for any challengers to not have a reasonable chance.
All these tech companies are always happy to poach each others top talent and most of that top talent will happily switch companies every few years to keep their value up.
Nothing unusual, nothing rotten here.
No proof of Intel being x years behind, no proof of AS being doomed and for sure no proof of any sinister plan to pump and dump stock ( at least not based on this move).
The reality is that Apple does have an advantage of using a process 1-3 steps ahead of the competition without having to carry 40 years of legacy crap.
It is also a reality that everything AS is sofar based on cores released (and designed for) cell phones 16 months ago.
-> it will get closer in the fields were Apple and Intel actually compete
--> Apple's lead in the low power markets will stay
---> Intel's lead in the "burn a 1000W to get the most power" markets will stay
Nothing unusual, nothing rotten here.
No proof of Intel being x years behind, no proof of AS being doomed and for sure no proof of any sinister plan to pump and dump stock ( at least not based on this move).
The reality is that Apple does have an advantage of using a process 1-3 steps ahead of the competition without having to carry 40 years of legacy crap.
It is also a reality that everything AS is sofar based on cores released (and designed for) cell phones 16 months ago.
-> it will get closer in the fields were Apple and Intel actually compete
--> Apple's lead in the low power markets will stay
---> Intel's lead in the "burn a 1000W to get the most power" markets will stay
Apple might just be hoping that the security of Intel’s chips gets as messed up as the T2’s, which can be compromised by a trivial USB gadget.
dude, I think you have an account in the wrong forum , what you search for is intel.com/noonecare.
My thoughts:
1. When the M1 Pro and Max came out, they’d been in development for 3 years prior. The M2 and variants are already done and the M3 is likely all but fully completed, and product roadmaps are in place to use those chips in coming years.
2. A company as large and well structured as Apple can mitigate any issues caused by losing a staff member such as him. And this wouldn’t have been a sudden “I quit” resignation, but more of a planned, well discussed transition handled by professionals, allowing both companies opportunity to adjust to the new situations.
1. When the M1 Pro and Max came out, they’d been in development for 3 years prior. The M2 and variants are already done and the M3 is likely all but fully completed, and product roadmaps are in place to use those chips in coming years.
2. A company as large and well structured as Apple can mitigate any issues caused by losing a staff member such as him. And this wouldn’t have been a sudden “I quit” resignation, but more of a planned, well discussed transition handled by professionals, allowing both companies opportunity to adjust to the new situations.
"In development" != "done"
(my guess)
A16 samples have most likely been shipped to Apple sometime last year, the base variant (for the iPhone) ready for full scale production whenever needed for the launch in September.
Base level M2 (more A16 cores) is fully designed and samples about to ship.
A team is working on finalising M2Pro/Max/??? which will be sampled later this year for 2023 launches.
Other teams are at different stages in the design of A17 and A18 (maybe even early stages of A19).
The x86-64 emulator has existed in Windows on ARM since late 2020. Unfortunately Apple won't allow Windows on M1 so hard to compare performance.
Are you certain about that? Last I heard Microsoft wasn't licensing Windows for ARM to individuals.
Or are you referring to loading Windows directly on M-series Mac computers and not using something like Parallels to run Windows in a macOS environment? But even then you have the issue about getting a license for Windows ARM if you are an individual.
Not saying there is one but if a noncompete clause covers anything it should cover this.
You must be one of those Apple fans this entire website was made for. I’m gonna head over to androidwindowsepicvirtnetx.com to hang out with more of us.