Semiwiki on LinkedIn Semiwiki on Facebook Semiwiki on Twitter Register / Log in 
Array
(
    [content] => 
    [params] => Array
        (
            [0] => /forum/threads/wafer-costs-low-end-cpu-and-gpu-implications.13490/
        )

    [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
)

Wafer Costs - low end CPU and GPU implications?

X

Xebec

Well-known member
Neophyte question here --

I've seen the articles about increasing chip/wafer/R&D costs per node, (and have been joyfully following Moore's Law since at least the "0.8/1.0 micron" days). Are we at or nearing the point where it's going to be really hard to advance $80 CPUs / $150 GPUs because even small dies on newer nodes will be expensive to manufacture?

Or will we see price reduction in older nodes that make it feasible - i.e. $200+ CPU/GPUs get manufactured on 3nm in a few years, while < $200 ones get manufactured on a 'cost reduced' 5nm node*?

*I assume the cost would come down from depreciation, increasing yields, etc -- maybe other improvements
 
Xebec said:
Neophyte question here --

I've seen the articles about increasing chip/wafer/R&D costs per node, (and have been joyfully following Moore's Law since at least the "0.8/1.0 micron" days). Are we at or nearing the point where it's going to be really hard to advance $80 CPUs / $150 GPUs because even small dies on newer nodes will be expensive to manufacture?

Or will we see price reduction in older nodes that make it feasible - i.e. $200+ CPU/GPUs get manufactured on 3nm in a few years, while < $200 ones get manufactured on a 'cost reduced' 5nm node*?

*I assume the cost would come down from depreciation, increasing yields, etc -- maybe other improvements
Click to expand...
Very likely, the low end will stay on cheaper processes for longer. Those will be scaled down, downclocked versions of latest designs. The top tier has no alternative, but to go for latest, and greatest. Everything else will be drifting +- 1 node of where it makes more economic sense for the price point of the product.
 
I also suppose this is also where the chip stacking / MCM tech comes in too -- i.e. Zen 2/3 using a 7nm CPU die and a 12/14nm I/O die. iGPUs seem like the ultimate 'easy' here..

I'm curious which components could be broken out on GPUs to allow for lower cost while taking advantage of newer nodes. Perhaps some of the I/O and memory pieces..

Thanks!
 
You must log in or register to reply here.
Back
Top