Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/dear-chip-designers-if-youre-struggling-to-get-components-made-try-28nm-supply-set-to-overtake-demand-the-register.15484/
        )

    [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] => 2021770
            [XFI] => 1050270
        )

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

Dear chip designers, if you're struggling to get components made, try 28nm. Supply set to overtake demand [The Register]

UMC 28nm is an immersion node. Not an outsourcing shop level design effort. You will spend way more than $1m on getting physical IP redesigned, and other NREs. Not to mention state of eFLASH on 28nm (all proprietary offers,) is not bright, and SRAM on 28nm is too expensive per area.

Nothing below 40nm is really friendly to what car industry uses. You can't use 28nm for high voltage analog, and pretty much anything else analog. SRAM/flash issue. RF IP is increasingly important because of demand for WiFi in MCUs is also not there. High temperature silicon is also not there, which means just anything 28nm will fail car certification in a lot of countries (even doorknob LED blinker often needs the same level of certification as the engine ECU chip.)
 
UMC 28nm is an immersion node. Not an outsourcing shop level design effort. You will spend way more than $1m on getting physical IP redesigned, and other NREs. Not to mention state of eFLASH on 28nm (all proprietary offers,) is not bright, and SRAM on 28nm is too expensive per area.

Nothing below 40nm is really friendly to what car industry uses. You can't use 28nm for high voltage analog, and pretty much anything else analog. SRAM/flash issue. RF IP is increasingly important because of demand for WiFi in MCUs is also not there. High temperature silicon is also not there, which means just anything 28nm will fail car certification in a lot of countries (even doorknob LED blinker often needs the same level of certification as the engine ECU chip.)

Then why did TSMC spend so much time and money on making their processes auto friendly?


 
Then why did TSMC spend so much time and money on making their processes auto friendly?



As far as I can see, there is a lot of analog stuff they will try on their 150mm-200mm fabs. There is nothing concrete I see there, in comparison to very exact description of what they provide on logic nodes.

  • NVM technologies for next generation MCU and AI memory

That can possibly be MRAM, which would be the long-lived NVM for data logging.
 
Back
Top