ARM Announces Next-Gen 64-Bit Artemis Mobile Chip On 10nm TSMC FinFET Process (hothardware.com) 51
MojoKid writes from a report via Hot Hardware: ARM has been working closely with TSMC for years now. Over the last six years or so especially, ARM and TSMC have collaborated to ensure that TSMC's cutting-edge process technologies work well with ARM's processor IP. However recently, ARM just announced the successful tape-out of a test chip featuring next-generation, 64-Bit ARM v8-A mobile processor cores, codenamed Artemis, manufactured using TSMC's upcoming 10nm FinFET process technology. The test chip features what ARM calls an Artemis cluster. It's essentially a quad-core processor with power management IP, a single-shader Mali graphics core, AMBA AXI interconnect, and test ROMs connected to a second cluster by an asynchronous bridge that features the memory subsystem, which is stacked with a Cortex M core that handles control logic, some timers, SRAM, and external IO. Compared to 16nm FinFET+, at nominal voltage, the 10nm test chip offered a 12% performance improvement in a similar power envelope. In super-overdrive mode (Vsod), the Artemis test chip offered similar performance, but at 30% lower power.SoCs for premium mobile devices with next-generation cores produced on the 10nm process node are expected to arrive later in the second half of this year.
Uhh .. (Score:5, Funny)
.. Bingo?
Re: (Score:3)
.. Bingo?
Yes, Squiffy?
Super-Overdrive Mode! (Score:3)
I think Intel needs to get some super-overdrive mode too if it wants to get down to 10nm as well... I think they're struggling with 14nm just now.
Does anybody know what's going on in the left of this picture [hothardware.com] from TFA?
Re: (Score:2)
Looks like "bring your kid to work day" and one of the little darlings had their watercolors with them.
Re: Super-Overdrive Mode! (Score:2, Insightful)
If it's true that TSMC can do volume at 10 nm then it really is a watershed. Intel is struggling to do volume at 10 nm http://www.theinquirer.net/inquirer/news/2447789/intel-denies-further-delays-to-10nm-chips-beyond-2017
It would be a watershed because Intel has been ahead in process node technology for decades. If it looses it's edge on the node then its a huge event.
I'm skeptical that TSMC can really crank out volume at 10 nm beating Intel which apparently can't. Volume is the difference between a science
Re: (Score:2, Interesting)
This happens every process node.
Just before intel starts shipping volume on a production ready process, someone else announces their process that's at the same pitch to give the impression of being 'almost caught up' when they aren't even close.
Don't think that 10nm TSMC is the same thing as 10nm Intel either. In node naming, nanometers are a marketing term.
Re: (Score:2, Informative)
It isnt true. TMSC can NOT do 10nm yet. The whole 10nm rating is a scam.
When they say 10nm, what they really mean is 12nm-14nm. Effectivley TMSC 10nm process is really the same size as Intels 14nm process.
http://wccftech.com/tsmc-10nm-volume-production-begins-2016-7nm/
Re: (Score:2)
And no one i know at Intel is sleeping well, still. They are challenged, and struggling, to keep innovating all the way to the 0.1nm process.
Re: (Score:2)
keep innovating all the way to the 0.1nm process.
Since the diameter of a single silicon atom is 0.2nm, I don't think they will reach 0.1nm.
Re: Super-Overdrive Mode! (Score:2)
You're still thinking in Silicon. Intel isn't.
Re: (Score:2)
You're still thinking in Silicon. Intel isn't.
Ok, then the diameter of a carbon atom is 0.17 nm, so 0.1 nm would still be impossible. Even a hydrogen atom has a van der waals diameter of 0.12 nm.
Re: Super-Overdrive Mode! (Score:2)
And you're also thinking literally.
Re: Super-Overdrive Mode! (Score:2)
My comment was intended to be figurative; that the industry is driven to smaller and smaller processes, and eventually must, in the instance of silicon, reach unavoidable limits. Intel is looking at new materials, as all others must be, in the drive to minimum process size. And management's answer to engineering's complaint that 'it can't be done yet' is 'try harder'.
No one I know at Intel is happy. They have, so far, missed the mobile market, are under enormous pressure in every other market, and their
Re: (Score:2)
"The whole 10nm rating is a scam."
It is a scam only in the sense that *no semiconductor manufacturer* has used the node name to represent any particular dimension of that node for years. 10nm just means comes after 16/14nm and before 7nm.
Re: Super-Overdrive Mode! (Score:5, Insightful)
If it's true that TSMC can do volume at 10 nm then it really is a watershed. Intel is struggling to do volume at 10 nm
We'll see. As far as I know, Intel doesn't announce test tapeouts since they're not in the business of selling design and process tech but rather actual processors. But if you look at slides from their 14nm release you see PRQ = product release qualification = volume production was late Q2 2014 [extremetech.com] and there's yield graphs back to May 2013 [wccftech.com]. Considering Intel was on a two-year tick-tock I would think it's natural if their first test was mid-2015 and Kaby Lake was announced in July because the yields were way too low.
A full 300mm wafer is 70695 mm^2, a Skylake quad is 122mm^2 and at 10nm you're probably looking at ~1000 CPUs/wafer. So with 0,1% yield you can issue the PR release saying you've made a 10nm chip, but if you don't have consistency in the process it's not worth launching because the net result costs more than a mature 14nm process. In any case parity is pretty good too, being a half-node behind is fighting with a cost handicap that's really tough to beat as AMD has noticed. When they launch Zen/Polaris it's the first time in forever they're on roughly even footing with Intel.
Re: (Score:3)
You honestly think Zen or Polaris will put AMD on an even footing with Intel? If so what makes you think that?
Well the entire paragraph was about process tech. Polaris and Zen is launching on GlobalFoundries' 14nm process, Intel has announced that for the next year or they'll stay on 14nm with Kaby Lake. So give or take a little marketing bullshit 14 ~= 14. Whether the design is any good, we'll see...
Comment removed (Score:4, Interesting)
Re: (Score:1)
TSMC delivered their last node ahead of schedule, so it's pretty certain that they will deliver 10nm on time - i.e. Autumn 2017 - ready for the Apple A11. The latest rumour says that Samsung will pip them to the post, though and Intel will only be a few months behind.
Intel are more agressive on tolerances, so their 10nm will be more dense than TSMC's. This comes at a cost though, because it's much harder to get good yields.
You should bear in mind Intel's processes are built for high end SoC's costing around
Re: (Score:2)
I think Intel needs to get some super-overdrive mode too if it wants to get down to 10nm as well...
But the dilithium crystals 'l never take it Captain!
Ummm, what? (Score:2)
Intel has been producing 14nm in volume for quite some time now. Starting with Broadwell their chips have been 14nm. So any system you see with a 5000 or 6000 CPU is 14nm. That started in 2014.
In terms of 10nm, well let's see what happens. TSMC has a history of over promising. 40nm, 28nm and 16nm have all been cases where they've talked a big game about availability and then it has been large delays before it finally started happening, and small quantities/yield problems in the beginning even still. That do
Re: (Score:3)
Awesome news for Web 2.0! (Score:2)
Mobile angular web apps might become usable now.
Re: (Score:2)
Are we getting more RAM to go with these suckers? (Score:2)
While there certainly are devices with more than 2GB, 1GB and 2GB are still the norm for budget and non-budget devices respectively (in the credible device space, that is. I know some PAYG phones are still 512M.) Are we going to see that come up to 2GB and 4GB any time soon?
Re: (Score:2)
Moto G4 Plus is $300 and comes with 4GB ram and 64GB of flash (plus SD card support!), the regular G4 is going to be ~$200 and 2GB of ram so I'd say they're coming and much sooner than this processor (released already in certain markets, coming to the US soon). The G4 also has a 5.5" 1080P screen so there's plenty of room to come in even cheaper by using a smaller and/or lower resolution screen.
Re: (Score:2)
There was no need to go to 64bit for additional ram anyways, ARMv7 supports a 40bit PAE which supports up to 1TB of ram.
The primary advantage of AArch64 is that it more than doubles the number of GP registers and the new SIMD instructions that leverage the larger registers.
Re: (Score:2)
Well there's the Cortex-A32 which provides a 32bit version of ARMv8.
I'm not sure the jump to 64bit will be instantaneously universal if budget Cortex-A7 phones migrate to the A32.
Re: (Score:2)
I might have to get a G4+ after the first price drop. I have a G 2nd right now and the lack of ram is galling
Re: What's with the ARM to x86 comparisons? (Score:2)
Intel was trying to get into smartphones and failed. They're still in the tablet space though, where ARM still dominates, so the comparison is still valid.
Re: (Score:2)
You definitely wouldn't use ARM for servers
Oh you would if you had proper fabric and skilled programmers. Currently there's lack of both.
Re: (Score:2)
The Aarch64 will certainly attract the required focus on proper fabric and skilled programmers. The result would take some years to show up, but it will.
Re: (Score:1)
Unfortunately Intel's 10nm and TSMC's 10nm are two different processes and it's comparing apples to potatoes. TSMC is typical a half of a generation behind the leading foundries, and they have a tendency of stating a smaller process generation than they are really capable of producing. All TSMC's elements for 16nmFF are larger than 16nm in size. The same goes for their 10nm, there are always some big bullshit marketing slides in the presentations they give us to explain away the size discrepency. It boils d