GlobalFoundries Partners With Google's Open-Source Silicon Effort To Provide 180nm Tech

[phoronix]

Phoronix: GlobalFoundries Partners With Google's Open-Source Silicon Effort To Provide 180nm Tech

Over the past two years Google has been spearheading an effort to make silicon design more open-source and allowing more projects to get started in chip fabrication. That got started with a partnership with SkyWater Technology and Google covering the costs for open-source projects to see their initial chips fabricated on a 130nm process. Google's Open-Source Silicon Design Initiative recently announced SkyWater 90mm manufacturing will get underway for future manufacturing runs. Today the news out of Google's open-source group is that GlobalFoundries has joined this initiative and is providing 180nm manufacturing access...

https://www.phoronix.com/news/Google-GloFo-GF180MCU

[cl333r]

I'm clueless about what this actually means... I mean.. why such ancient size.. why all the other effort, do they think people make new chips in mom's basements..?

[schmidtbag]

I also don't get why this size in particular. GloFo offers 12nm last I checked, so why not use that?

[Snaipersky]

Because it's extremely mature, cheap, and still in use by designs which would require expensive and time-consuming re-validation should they change, and for certain applications, there's little call for reducing power draw, increased clock speed, or a physically smaller chip

[andrei_me]

What uses this lithography, TV remote controllers?

[schmidtbag]

Because it's extremely mature, cheap, and still in use by designs which would require expensive and time-consuming re-validation should they change, and for certain applications, there's little call for reducing power draw, increased clock speed, or a physically smaller chip

Sure but then why not 90nm? From what I heard (don't remember where), halving the transistor size makes it relatively easy to switch to a new node. 90nm is also extremely mature and cheap, and you can fit proportionately more usable product per-wafer, which is more profitable. I can't imagine 180nm is that much cheaper, when they're both outdated tech.

[Snaipersky]

Sure but then why not 90nm? From what I heard (don't remember where), halving the transistor size makes it relatively easy to switch to a new node. 90nm is also extremely mature and cheap, and you can fit proportionately more usable product per-wafer, which is more profitable. I can't imagine 180nm is that much cheaper, when they're both outdated tech.

In certain fields, *any* change requires re-validation, and the ODM/OEM is likely uninterested in even sinking the cost of re-evaluating a known viable product. I suspect 180 has seen a dip in customers due to supply issues holding up other components, and GloFo wants to keep the line going until its less profitable than replacing with a new node.

[AndyChow]

Sure but then why not 90nm? From what I heard (don't remember where), halving the transistor size makes it relatively easy to switch to a new node. 90nm is also extremely mature and cheap, and you can fit proportionately more usable product per-wafer, which is more profitable. I can't imagine 180nm is that much cheaper, when they're both outdated tech.

We are still in a massive worldwide chip shortage. I'm going to go on a limb and say their 90nm process is booked up. They might have some occasional underused capacity on 180nm.

This are free chips made from available capacity. I think it costs virtually the same to have the chips being made 24/7 vs having a stop in production. Might even be more expensive to have a stop in production if it means extra recalibration.

It's beyond cool that they would allow various people to make functional prototypes or proof-of-concepts, for free. It's very generous of them and Google.

[tildearrow]

I'm clueless about what this actually means... I mean.. why such ancient size.. why all the other effort, do they think people make new chips in mom's basements..?

Well, at least it's a lot better than the 2cm "process" we have at home, and 1mm with best effort...

[tildearrow]

I also don't get why this size in particular. GloFo offers 12nm last I checked, so why not use that?

Cost of production.

12nm is cheap only when you produce it in the hundred-thousands.