Samsung Scales 10nm FinFET Mountain, Products To Ship In 2017

News
By Paul Alcorn
published

Samsung announced that it is the first company in mass production of a system on a chip (SoC) built on a 10nm FinFET process. Samsung's 10nm process will have at least two iterations, with the 10LPE process serving as the lead vehicle.

The move to 10nm FinFET marks a more conventional approach to scaling, whereas GlobalFoundries is skipping 10nm in favor of the 7nm process. AMD is beholden to the GlobalFoundries roadmap, so it will obviously skip 10nm, but Intel plans to plow forward with its 10nm tri-gate in 2017. TSMC also has its own unique 10nm process that it plans to ramp later this year. Samsung has the distinction of being the first company in mass 10nm FinFET production, and it already has an aggressive plan to ramp to a more refined 10nm process.

Samsung indicated that the 10nm FinFET 10LPE process offers a "30% increase in area efficiency with 27% higher performance or 40% lower power consumption (emphasis added)." The word "or" is important, as it might indicate that you cannot have higher performance and lower power consumption at the same time. The company employs triple-patterning to allow bi-directional routing, which allows it to retain design and routing flexibility from prior nodes. The company also plans to build a healthy 10nm ecosystem that includes reference flow verification, IPs, and libraries. The company expects the leading edge 10LPE products to land in digital devices early next year, and (of course) availability will become more widespread as the year progresses.

“The industry’s first mass production of 10nm FinFET technology demonstrates our leadership in advanced process technology,” said Jong Shik Yoon, Executive Vice President, Head of Foundry Business at Samsung Electronics. “We will continue our efforts to innovate scaling technologies and provide differentiated total solutions to our customers.”

Not content to rest on its 10LPE laurels, the company also announced that it would produce a second-generation 10LPP process in mid-2017. The company narrowly described the newer process as providing the obligatory performance boost, but we expect more detail in the months to come.

Samsung already enjoys a considerable advantage over its rivals in 3D NAND production, largely due to its early start with the 3D architecture. Now the company has a leading position on 10nm FinFET as well, which might prove to be an advantage in the years to come. Of course, competition is pitched, and much of its success will depend on how quickly its competitors can ramp up production.

Samsung noted that production-level process design kits (PDK) and IP design kits are available now, so it's clearly off to the races with 10nm FinFET 10LPE.

Paul Alcorn
Paul Alcorn
Managing Editor: News and Emerging Tech

Paul Alcorn is the Managing Editor: News and Emerging Tech for Tom's Hardware US. He also writes news and reviews on CPUs, storage, and enterprise hardware.

See more CPUs News
14 Comments Comment from the forums
  • jkhoward
    Will they also explode?
    Reply
  • josejones
    Globalfoundries, TSMC, AMD and even Samsung may be sued for false advertising as their nm claimed sizes are completely false as a 14nm is closer to 20nm but they still name it 14nm and the recent claims by AMD of skipping 10nm and going to 7nm is false - it's more like 14nm. This is huge news and everybody should be made aware:

    Understanding how foundries name their process nodes: Intel’s 10nm versus TSMC 10nm and beyond

    "The answer lies in the fact that Intel’s naming standards and TSMC’s naming standards are drastically different. Intel’s processes use the same backbone as the advertised node (a 14nm process will use a 14nm backbone) while as all pure play foundries use a mixture of process technologies. TSMC’s 16nm FinFET tech for example uses a 20nm backbone (BEOL). So it is almost a certainty that they will be using a 14nm BEOL for their ’10nm finfet’ node.

    "Let me explain a bit further, the process names that foundries use have now become more or less marketing material and not accurate physical descriptions of the node (except maybe in the case of Intel). Simply marketing a node as 7nm FinFET or 10nm FinFET does not make it a true 7nm or 10nm node respectively. I think the most relevant benchmark in this case (in my attempt to explain the point) is that of the Transistor Gate Pitch. This is the measure which is usually a very good indicator of the “true” node that a foundry might be using."

    http://wccftech.com/intel-losing-process-lead-analysis-7nm-2022/

    http://semiengineering.com/to-7nm-and-beyond/

    So when Globalfoundries, TSMC, AMD and Samsung claim they have a 14nm it's more like 20nm and their 10nm is more like 14nm and their 7nm it's more like 10nm. Sorry. I wish it were true too but, it's not their 7nm is more like 10nm.

    ;
    Reply
  • PaulAlcorn
    18740794 said:
    Globalfoundries, TSMC, AMD and even Samsung may be sued for false advertising as their nm claimed sizes are completely false as a 14nm is closer to 20nm but they still name it 14nm and the recent claims by AMD of skipping 10nm and going to 7nm is false - it's more like 14nm. This is huge news and everybody should be made aware:

    Understanding how foundries name their process nodes: Intel’s 10nm versus TSMC 10nm and beyond

    "The answer lies in the fact that Intel’s naming standards and TSMC’s naming standards are drastically different. Intel’s processes use the same backbone as the advertised node (a 14nm process will use a 14nm backbone) while as all pure play foundries use a mixture of process technologies. TSMC’s 16nm FinFET tech for example uses a 20nm backbone (BEOL). So it is almost a certainty that they will be using a 14nm BEOL for their ’10nm finfet’ node.

    "Let me explain a bit further, the process names that foundries use have now become more or less marketing material and not accurate physical descriptions of the node (except maybe in the case of Intel). Simply marketing a node as 7nm FinFET or 10nm FinFET does not make it a true 7nm or 10nm node respectively. I think the most relevant benchmark in this case (in my attempt to explain the point) is that of the Transistor Gate Pitch. This is the measure which is usually a very good indicator of the “true” node that a foundry might be using."

    http://wccftech.com/intel-losing-process-lead-analysis-7nm-2022/

    http://semiengineering.com/to-7nm-and-beyond/

    So when Globalfoundries, TSMC, AMD and Samsung claim they have a 14nm it's more like 20nm and their 10nm is more like 14nm and their 7nm it's more like 10nm. Sorry. I wish it were true too but, it's not their 7nm is more like 10nm.

    ;

    I agree, there is a lot of 'slop' in the marketed measurements, but we usually have to wait until a report from a third-party analysis firm, such as TechInsights, before we can attain positive verification of the actual process used for the BEOL. Marketing horseplay at its finest, but unfortunately, most of the reports that detail the actual measurements are very expensive.
    Reply
  • Emanuel Elmo
    Unless that goes wireless and they improve there sensor and reliability as a brand maybe then I will think of switching from my logitech g900.

    Otherwise.... no thank you.
    Reply
  • IndignantSkeptic
    Does it really matter what the nm number is? That is not the number that is important. All that matters is that there is progress. Actually, if they are lying about the numbers, then maybe that is a good thing because it means that there is even further to go before the nm physics limit is hit.
    Reply
  • bit_user
    AMD will probably have at most 1 CPU product cycle before they fall behind on process, again. Yay. And, given GF's track record, Intel will probably ship product on 7 nm before they do.

    The word "or" is important, as it might indicate that you cannot have higher performance and lower power consumption at the same time.
    No, it's just the classic power/performance tradeoff. There's a power vs. performance curve, and they're telling you two points on it: "if you hold performance the same, you get this power savings" and "if you hold power the same, you get that performance". But that's not saying you can't pick any point in between (or beyond), depending on your power budget & thermal envelope.
    Reply
  • bit_user
    18740794 said:
    Globalfoundries, TSMC, AMD and even Samsung may be sued for false advertising as their nm claimed sizes are completely false as a 14nm is closer to 20nm but they still name it 14nm and the recent claims by AMD of skipping 10nm and going to 7nm is false - it's more like 14nm. This is huge news and everybody should be made aware:
    According to semiaccurate, Samsung's lead over Intel is real. Even taking into account Intel's second-gen 14 nm process:

    http://semiaccurate.com/2016/09/01/intel-finally-narrows-14nm-process-technology-gap-samsung/
    Reply
  • josejones
    Understanding how foundries name their process nodes: Intel’s 10nm versus TSMC 10nm beyond

    "The answer lies in the fact that Intel’s naming standards and TSMC’s naming standards are drastically different. Intel’s processes use the same backbone as the advertised node (a 14nm process will use a 14nm backbone) while as all pure play foundries use a mixture of process technologies. TSMC’s 16nm FinFET tech for example uses a 20nm backbone (BEOL). So it is almost a certainty that they will be using a 14nm BEOL for their ’10nm finfet’ node.

    Let me explain a bit further, the process names that foundries use have now become more or less marketing material and not accurate physical descriptions of the node (except maybe in the case of Intel). Simply marketing a node as 7nm FinFET or 10nm FinFET does not make it a true 7nm or 10nm node respectively. I think the most relevant benchmark in this case (in my attempt to explain the point) is that of the Transistor Gate Pitch. This is the measure which is usually a very good indicator of the “true” node that a foundry might be using."

    http://wccftech.com/intel-losing-process-lead-analysis-7nm-2022/

    ;
    Reply
  • bit_user
    18809638 said:
    Understanding how foundries name their process nodes: Intel’s 10nm versus TSMC 10nm beyond
    Repost.

    wccftech is also a weak source. They have a track record of leeching their content from other sites, rather than doing any original reporting. I wouldn't trust them, if I had any alternatives.
    Reply
  • josejones
    wccftech cites its sources - if you actually read the article you'd know that. It's not wccftech I worry about - i simply don't trust YOU and could not care less about your opinion. So, you can stop responding and wasting my time. It has already been well documented that you are wrong and your previous link was presented here fallaciously as they were talking about timing to get new products out for release. No matter how much you don't like it, no matter how much you can't handle the truth the fact still remains:

    Understanding how foundries name their process nodes: Intel’s 10nm versus TSMC 10nm beyond

    "The answer lies in the fact that Intel’s naming standards and TSMC’s naming standards are drastically different. Intel’s processes use the same backbone as the advertised node (a 14nm process will use a 14nm backbone) while as all pure play foundries use a mixture of process technologies. TSMC’s 16nm FinFET tech for example uses a 20nm backbone (BEOL). So it is almost a certainty that they will be using a 14nm BEOL for their ’10nm finfet’ node.

    Let me explain a bit further, the process names that foundries use have now become more or less marketing material and not accurate physical descriptions of the node (except maybe in the case of Intel). Simply marketing a node as 7nm FinFET or 10nm FinFET does not make it a true 7nm or 10nm node respectively. I think the most relevant benchmark in this case (in my attempt to explain the point) is that of the Transistor Gate Pitch. This is the measure which is usually a very good indicator of the “true” node that a foundry might be using."

    http://wccftech.com/intel-losing-process-lead-analysis-7nm-2022/

    ;

    Reply