Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/index.php?threads/xilinx-unveils-revolutionary-adaptable-computing-product-category.10239/
        )

    [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] => 2020970
            [XFI] => 1050270
        )

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

Xilinx Unveils Revolutionary Adaptable Computing Product Category

Jozo035

Member
First waves of 7nm devices starts showing. After 7nm platform announced by Qualcomm, 7nm CPU and GPU from AMD and several ASICs from mining hardware providers... Today we have finally something from Xilinx.

Xilinx Unveils Revolutionary Adaptable Computing Product Category

Plus interview with Victor Peng, CEO of Xilinx:

Xilinx Announces Project Everest: The 7nm FPGA SoC Hybrid

I am looking forward to dig into technical details, but from what i can see now, for deployment in datacenters it makes sense. FPGAs are hard to program and lot of HPC platforms ends up just as array of soft cores with not very ideal memory and interconnect hierarchy. If they make it directly as hard-wired cores with certain degree of configurability (extension to current DSP48) then it might be both easier to program and with better characteristics (performance, power draw...).

It looks like Xilinx will have another few years of monopoly in leading edge FPGAs since even if competitors can manage to release 10nm products, it will be no match to 7nm.
 

Daniel Nenni

Admin
Staff member
Xilinx’s ACAP portfolio will be initiated with TSMC’s 7nm manufacturing process, with the first tapeouts due in late 2018. Xilinx states that Project Everest has been a monumental internal effort, taking 4-5 years and 1500 engineers already, with over $1b in R&D costs. The final big chips are expected to weigh in at 50 billion transistors, with a mix of monolithic and interposer designs based on configurations.

So we are looking at Xilinx 7nm HVM in 2020? Why are they so behind? Is Xilinx that comfortable now that Intel owns Altera? From what I am told Intel 14nm FPGAs are faster and cheaper than Xilinx.
 
Top