GlobalFoundries Stops All 7nm Development

[Gubbi]

You mean GlobalFoundries?

Yeah, mea culpa

Cheers

 

[Alexko]

We don't know the 7nm yields of TSMC or Samsung. We know Intel struggles with it (they call it 10nm). So you have a massive investment in tooling, and an uncertain chance of succeeding.

Then there is the amount of customers. As the price goes up, the number of customers goes down. 10/14nm saw half the volumne of 28nm. Doing 7nm with quadruple patterning requires in excess of 80 mask sets for a full metal stack SOC. Even if your yield is adequate, few customer can afford the mask sets, we're talking Apple, Qualcomm, Nvidia and AMD.

By ditching 7nm, GF is choosing a path of significantly lower financial risk. Their 22FDX process has performance of a 14/16nm process but the cost of 28nm. They have chosen to move forward with their 12FDX, which will have good performance, good density and excellent power characteristics, - instead of betting on 7nm.

Cheers

So what's the trade-off associated with FDX? Just density? Maximum clocks?

 

[Entropy]

So what's the trade-off associated with FDX? Just density? Maximum clocks?

At present, both density and maximum clocks. It has benefits as well of course. Significant ones too, which is why there is growing interest outside max density/clock applications. It will be interesting to see how it evolves going forward. That remains to be seen, but it is a technology where GloFo is actually very well positioned, so it makes sense for them to focus on it.

 

[Alexko]

At present, both density and maximum clocks. It has benefits as well of course. Significant ones too, which is why there is growing interest outside max density/clock applications. It will be interesting to see how it evolves going forward. That remains to be seen, but it is a technology where GloFo is actually very well positioned, so it makes sense for them to focus on it.

So, is the GloFo partnership still relevant for AMD? For IO hubs, or some embedded/semi-custom designs, perhaps?

 

[3dilettante]

The 14/12nm AMD products will persist for some time, particularly the pro SKUs that emphasize stable product cycles and long-term availability.

The specialized processes GF is pursuing would benefit things such as the already mentioned IO or chipset silicon, or possibly future satellite chiplets on an interposer in the supposed dis-integrated future.
On the other hand, GF's approach to 2.5D and 3D has been more of a third-party based model, reflective of GF's more limited resources and expertise. TSMC's goal has been to be more of a one-stop shop for fanout and interposer integration, which could be an area AMD is concerned about given the wide-ranging logistical train it wrestled with getting the HBM GPUs developed and put together.