Intel's First 10nm Processor Lands In China

Intel's much-anticipated, and oft-delayed, 10nm processors have popped up in a Lenovo Ideapad 330, which is a low-end notebook that comes with 4GB of memory and a 500GB HDD, at Chinese retailer JD.com.

The dual-core Core i3-8121U sports the Cannon Lake architecture, but the integrated graphics are either disabled or not present. As a result, the notebook comes with RX540 graphics. We followed up with Intel, which confirmed that the chips are only on sale in China. The company won't share a specific date for possible U.S. distribution, but it shared the full specs of the new processor

Intel has long been known for its aggressive cadence of transitions to smaller process nodes, but the 10nm process has proven to be its biggest challenge in recent history. Intel first planned on delivering processors based on the smaller process in 2015, but now it has pushed high volume manufacturing back to 2019 due to yield issues that make the end products too expensive for mass production.


Intel Core i3-8121U
Intel Core i3-8130U
Architecture
Cannon Lake
Kaby Lake
Lithography
10nm
14nm+
Cores/Threads
2 / 4
2 / 4
Base/Boost Frequency (GHz)
2.2 / 3.2
2.2 / 3.4
Cache
4MB
4MB
TDP
15W
15W
Memory Support
DDR4-2400, LPDDR4/x-2400
DDR4-2400, LPDDR3-2133
Memory Channels
2
2
Memory Bandwidth
41.6 GB/s
34.1GB/s
PCIe
PCIe 3.0 x16
PCIe 3.0 x12
Tjunction
105C
100C
Package Size
45mm x 24mm
42mm x 24mm
Integrated Graphics
No
Intel UHD Graphics 620
AVX
AVX2
AVX2

In the interim, Intel has released the Core i3-8121U to market. This new chip comes bearing the Cannon Lake microarchitecture, but the details of the new design are still wrapped in secrecy. We expect that the new architecture will provide improved instructions per clock (IPC) that will boost overall performance, but we won't know the details until the processors land in the hands of a capable tester. Intel isn't disclosing pricing on its newest chip, and it has yet to make an appearance on its processor pricing list. 

The Core i3-8121U comes equipped with Turbo Boost 2.0, much like its closest cousin, the Kaby Lake Core i3-8130U. Both chips feature the same 15W TDP, but the Cannon Lake model comes with a 200 MHz lower Turbo Boost frequency, which could be offset by increased IPC. Last year, Intel revealed that its inaugural 10nm process would not provide more transistor-level performance than the highly-refined 14nm++ process, but the company does claim it will offer more performance than the 14nm+ used on the Kaby Lake models. We'll have to wait to see how that translates to chip-level performance.

The Core i3-8121U has a few new notable additions, such as 16 lanes of PCIe 3.0, which is an improvement over the Kaby Lake model's 12 lanes. Intel also included support for up to 32GB of LPDDR4, which yields higher memory bandwidth than the Kaby Lake model.

Intel increased the size of the package to 45 x 24mm, which means the chip isn't pin-compatible with the Kaby Lake BGA mounting. We also see a slightly increased Tjunction of 105C, which is the maximum temperature the chip can sustain before it begins throttling.

It has long been rumored that the leading 10nm Intel chips come with a disabled graphics unit, but we can't be sure that Intel built the capability into the die. In either case, the -8121U comes without integrated graphics, which means it also doesn't support other graphics-acceleration features. Intel recently updated its Architecture Instruction Set Extensions And Future Features (PDF) document, which states that the Cannon Lake processors support AVX-512. The Core i3-8121U doesn't support the feature, but we wouldn't expect to find AVX-512 in low-cost notebooks anyway.

Intel's eighth-generation chips now include Kaby Lake, Coffee Lake, and Cannon Lake processors built on the 14nm+, 14nm++, and 10nm processes. The extended eighth-generation branding is indicative of Intel's continuing struggle to field new microarchitectures as it tries to deploy new process technologies. Intel has made impressive performance gains over the course of the four iterations of the 14nm process by tightening design metrics, but the company has curiously decided to lock its new microarchitectures behind the smaller process nodes. It's puzzling that Intel hasn't simply brought a new microarchitecture to the existing process, thereby offering larger performance benefits than we see with the rewarmed Skylake architecture. 

For now, it seems that the Core i3-8121U will serve as a learning vehicle for Intel as it improves its 10nm process. The relatively unimpressive dual-core chip is also rumored to surface in the upcoming Crimson Canyon NUC next month.

Create a new thread in the UK News comments forum about this subject
This thread is closed for comments
2 comments
Comment from the forums
    Your comment
  • _lc_
    "Not so much anticipated" here, after we had to learn that they still ship their vulnerable chips. Once again, their performance will suffer after the patches are being applied in software. :-(
  • genz
    Honestly I gotta love Toms for how it hints whenever someone makes a huge blunder but also sends them lots of nice free stuff haha.

    14nm is 14nm+ is 14nm++. If Intel could have brought 6 core Sandy in 2010 for $600 why couldn't they bring it to consumer prior to Coffee Lake and Ryzen bringing 8+ cores, esp when they get a bit more chip real estate to work with every node? You think they couldn't work it out?

    Any actual new tech is already there in the 10nm prod line, it's just not refined enough for production. 14nm'+' and '++' are brand names, meant to make you believe that Kaby and Coffee were actually on the roadmap before Cannon was missing deadlines. Meant to justify enforced pin-incompatibility. Meant to explain away X299 and the reason Kaby and Sky were on the same socket. I'm not sayin they aren't different, but they are essentially 'tocks', all of them, and logically compatible with each other's mobo, Coffee included.

    Tick tock broke with Haswell because of the same thing (yield) and it's getting a lot worse at Intel as the nm go down. Cannon is almost 4 years late because 10nm is almost 4 years late.

    Intel's issues are already quite clear. For the last few years they have been forced to refine the best commercially viable process they have, and now it's so optimised they are going to struggle to make 10nm as dense and clockable as 14nm without cooking itself, let alone with good yield, so they have to start at the low end. Expect Haswell/Ivy heat issues here again until '10nm+' makes it cos I can guarantee they will ship it the moment it can turn a profit, shiny or not.

    And LC makes a very good point, Intel likely included those same vulnerabilities in the designs they were actually working on when Spectre went live, including Cannon Lake, necessitating a new stepping of every chip in production.