Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/another-new-specialized-server-processor-by-marvel-on-tsmc-5nm.14729/
        )

    [addOns] => Array
        (
            [DL6/MLTP] => 13
            [Hampel/TimeZoneDebug] => 1000070
            [SV/ChangePostDate] => 2010200
            [SemiWiki/Newsletter] => 1000010
            [SemiWiki/WPMenu] => 1000010
            [SemiWiki/XPressExtend] => 1000010
            [ThemeHouse/XLink] => 1000970
            [ThemeHouse/XPress] => 1010570
            [XF] => 2021370
            [XFI] => 1050270
        )

    [wordpress] => /var/www/html
)

Another new specialized server processor by Marvel on TSMC 5nm

hist78

Well-known member
The Register: Marvell says it has samples of TSMC-made 5nm server DPUs – and 3nm is next.

The server related processor development is moving fast forward from several companies. Intel's server division is at a crossroads. They can either embrace the TSMC foundry ecosystem to speed up products rollout or wait for Intel's own manufacturing capabilities to catch up. IMHO, the second route can lead Intel to win the battle but lose the war.
 
Personally I think 3nm will be a HUGE process node for TSMC, probably the biggest in the history of the company. Intel does not have enough EUV systems for their 3nm equivalent node, Samsung is struggling with yield and customer confidence, and GAA may be farther out than people think. Intel, AMD, Nvidia, QCOM, Marvell, Mediatek, and all of the majors will use TSNC N3. TSMC may even take another FinFET step after N3 to keep the node going.

Customers may weigh in on this as well with the Intel vs AMD battles. TSMC has a huge fanbase that may prefer the Intel/TSMC chips versus the Intel/Intel chips.

It truly is going to be an epic battle of the titans at the 3nm node, absolutely.
 
Intel needs to become a foundry, not immediately but at a brisk pace. The demand for idm semiconductors declines every quarter. If it wasn't for habana and mobileye it would be worse.
 
Personally I think 3nm will be a HUGE process node for TSMC, probably the biggest in the history of the company. Intel does not have enough EUV systems for their 3nm equivalent node, Samsung is struggling with yield and customer confidence, and GAA may be farther out than people think.
Samsung just announced that their GAA 3 nm will go into "mass production" at EoY 2022.
  • 3GAP will follow in one year for mass production at EoY 2023

  • 2GAP will take another couple of years, mass production in 2025
Source: https://www.anandtech.com/show/16995/samsung-foundry-2nm-silicon-in-2025
 
Back
Top