I did not expect ARM would take so long to catch up with even the Apple M1 or the same core designs in the Apple A14. The Cortex X3 in the current flagship chip Snapdragon 8 Gen 2 still _just_ fails to match the Apple A14 (let alone the A15 and A16), perhaps this little design boost is finally enough to match it.
If Microsoft can get Qualcomm to implement some of the same chip tricks Apple pulled off for x86 emulation performance they can use it for a potentially somewhat compelling laptop package. It's a tough ask though, as Qualcomm seems mostly to just use pretty vanilla ARM designs. And I don't expect anyone but Microsoft to be interested in taking that plunge.
This. Vendors skimp on cache size because of overall cost. But cache size is just as critical as CPU core when it comes to performance. Apple doesn't sell chips to anyone else, so they have no pressure to have small caches.
Sure, but it's worth noting that the Cortex X3 in the Snapdragon 8 Gen 2 is _almost_ at the level of the A14 in single core performance, just not quite there. Even if the performance bump is less than promised, they will still equal it.
The Snapdragon 8 Gen 2 is on TSMC 4nm compared to the A14 on TSMC 5nm.
Also, the Cortex X4 spec projections are based on the second gen TSMC 3nm process node that isn't in volume production yet. If the M3 is announced (as rumored) next week at WWDC on the first gen TSMC 3nm, we'll have some basis for comparison (at least as far as projected specs go).
None of the manufacturers want in. x86 is unencumbered by licensing issues, and RISC-V is moving through evaluation like a jackrabbit on speed. Apple has incentive to make ARM chips - they're deep in bed with Softbank (ARM licenseholders) and were already designing their own ARM cores. Nvidia has a few industrial-application HPC ARM chips they manufacture, but have been noticeably quiet about ARM after the failed acquisition attempt. Intel is cashing in on fabless manufacturing and also not designing their own cores. AMD has it's headphones on and doesn't see the RISC truck barrelling down Main Street.
So... I would honestly be surprised if anyone tried, at this point. Most of the potential stakeholders don't seem super interested.
To quickly lower expectations before reading the article:
> Every year with these Arm flagship chip announcements, the company also includes a wild design for a giant mega-chip that usually never gets built.
That being said, this 14-core figure is when doing a full-Arm SoC (with Arm internal bus). SoC vendors are still allowed (AFAIK [0]) to do their custom interconnect that goes higher than that
[0] there have been rumors seen here that Arm is pushing towards a model where SoC vendors must do all-Arm components. That still sounds weird to me, and I'm not witnessing any actual effect of this, so I still highly doubt it.
Given that there is a whole publicly traded company that does ARM SoC interconnect as their bread and butter (Arteris IP) and they and their stock price seem to be doing OK is an indication that it’s still allowed.
Still waiting for non handicapped (thinkpad x13s) ARM competitors. I would like to get performance + long battery + good screen + Linux but it seems to me that It will take so much time that I'm better off waiting for Asahi Linux to get mature and buy a second hand M1.
Finding this was helpful to me, as I can never keep track what is a fast (eg, Xeon equiv) type arm chip, and what's a low power embedded (Atom equiv) and what's a laptop / desktop chip.
It's a bit of an unfair statement but I do see where he's coming from. Apple make consumer phones and laptops, professional workstations, wearables, headphones, and TV/Home products. We've yet to see an M series chip for products in any of the following fields:
Cloud Computing, Computer Vision, Data Center Solutions, Edge Computing, High Performance Computing, IT Infrastructure, Network Connectivity, Robotics, Security, Broadcasting, Energy & Utilities, Financial Services, Government & Public Sector, Health & Life Sciences, Hospitality & Restaurants, Industrial, Manufacturing, Retail, Smart Cities, Transportation.
Intel, NVIDIA, AMD, Qualcomm, and Arm make devices or design IP for every single one of those sectors. Apples' portfolio is streamlined to just "lifestyle sectors"; stuff people directly interact with in their every day lives. How can you call them the number one chip maker when they don't compete in any of those markets, and even in the one they do they're only superior in performance per watt; not sheer performance benchmarks? Until we start seeing M2 chips in everything from server racks and satellites, I'm inclined to side with Gelsinger on this one.
> How can you call them the number one chip maker when they don't compete in any of those markets,
Because they produce north of 300 million SoCs every year. They may not end up the number one chipmaker every single year but they're in the top ten and depending on other companies' demand vs Apple's demand may end up number one.
If we're talking about manufacturing volume as the sole definition for "number one chip maker", then the idea that Apple might be first is frankly laughable. If we discounted pure foundries (TSMC, UMC, Global Foundries) they'd struggle to scrape top 5 at the best of times. Intel, Samsung, Qualcomm, Nvidia, AMD and the like work in the range of billions, or near enough. Currently Intel has 76.7% of the laptop chip market share alone; that's before you start considering their server, edge, and network markets. If we start including trailing edge chips, from the likes of Texas Instruments, Micron, Broadcom, etcetera, then Apple would be a small drop in the pond. Even if Apple ate up the entire consumer market, they'd still struggle to compete with the rest of the industry on volume.
I don't think people appreciate the obscene number of leading edge chips that go into the industrial sectors. I've seen more chips go into a single Industry 4.0 factory than I have to an entire states Micro Center inventory.
Number one in performance / watt? maybe? sales? Well no. According to the web[1] total chip sales in 2022 excluding arm amount to 374 million units. Server chips being 36.1 million units. Apple's biggest quarter in 2022 was 7 million macs in Q4[2].
For anyone curious, iPhones ship 234 million a year, Watch around 60 million, iPad another 20 million. Not sure about Apple TV but let's assume that's 0. That's over 300 million CPUs a year.
I doubt the total chip number for the industry though - Intel alone ships ~100M. But it's clear Apple is very likely the largest manufacturer of CPUs.
* Correction: MediaTek shipped 350M units in 2020 and Qualcomm shipped 319M (due to the Trump embargo on China). So Apple is up there competing with manufacturers that sell their chips to anyone who will buy, but it may not necessarily be the largest.
Yeah. How you count things is very subjective to where you draw the line. As for margins it's hard to say because Apple doesn't sell chips independently so their profit margin on a chip is withdrawn from however they want to distribute the profit margins for the entire product and Apple does not provide a breakout a sufficient level of granularity into the financials to understand that.
It's not surprising, anyone who sells iPhones is going to need a lot of chips. Gelsinger wasn't half wrong anyways - outside the consumer hardware market Apple is a no-show.
Indeed, to an almost comical extent. I worked at a Mac-only shop that spent 50+ engineering hours per-month debugging the Darwin runtime. Our product didn't ship to Mac either, so we were basically fanagling a Win32 + Linux server toolchain to work on a platform we didn't support.
Tech companies count as consumers, and many consume the Apple product even if it's to their detriment.
If AMD is a chip maker then so is Apple by the same metrics.
So, pick those metrics, they're certainly pushing large volumes, they're pushing some of the highest performance and some of the most efficient CPUs available.
So, they're #1 in power efficiency, #1 in fanless performance, #1 in Tablets, #1 in Fanless computers, #1 in Mini computers.
If you outsource chip creation, then you are chip designer, arent you?
>Fabless manufacturing is the design and sale of hardware devices and semiconductor chips while outsourcing their fabrication (or fab) to a specialized manufacturer called a semiconductor foundry. These foundries are typically, but not exclusively, located in the United States, China, and Taiwan
So if we're talking about high-end chips Intel and Samsung are both designers and manufactuerers? And Samsung seems to have mostly abandoned Exynos and is using Qualcomms SoC now. So that only leaves Intel..
>Today an iPhone 14 Pro embarrasses Arm's best with 63 percent higher single-core scores in Geekbench compared to an X3-equipped Snapdragon 8 Gen 2.
> but these launch events have a history of making performance claims that don't align with what actually arrives in consumers' hands.
Sigh. Both are factually wrong.
First point, Where the heck did that 63 percent came from? The iPhone 14 is about 38% faster. With much larger die size, and larger cache.
Second point implies ARM is lying. When ARM is the only company that gives precise reading of their benchmarks and ISO performance. How about asking that from any other manufacturers.
My guess is that for hardware you should ignore Ars and only read Anandtech or other site who goes into deep dive. Ars is fast becoming Engadget or just another blog.
I really wish Ars would let Ron Amadeo go, or PiP him. Their other writers are much better.
He’s woefully technically inept, his articles often have unchecked errors like this and he jumps to bombastic conclusions like the whole Samsung ROM size issue.
Almost certainly, when I see complaints about writing on Ars, it’s a Ron piece.
If Microsoft can get Qualcomm to implement some of the same chip tricks Apple pulled off for x86 emulation performance they can use it for a potentially somewhat compelling laptop package. It's a tough ask though, as Qualcomm seems mostly to just use pretty vanilla ARM designs. And I don't expect anyone but Microsoft to be interested in taking that plunge.