The open question about 28nm APUs was where they would be fabbed. TSMC was in the running mainly because they had a working 28nm HKMG process. GloFo’s delays meant that the first generation 28nm APUs, Wichita and Krishna were summarily canceled, and for a bit, you could almost feel the love between AMD and the ex-AMD fab that is now called Global Foundries. That was then, this is now.
Between last fall’s knifing of the 28nm parts and the final decision on fabbing, TSMC has stumbled badly too, and then made the situation far worse with their ham-handed messaging. The more they opened their mouth, especially by proxy, the worse they looked. SemiAccurates sources have said that things were pretty much as bad as we heard for a very short period, but have gotten much better since. Unfortunately, TSMC’s proxies don’t know when to stop digging their own hole, especially with the financial community, and are currently exacerbating the issue.
Last SemiAccurate heard, the Kaveri and Cabini duo were looking to be made at TSMC, but the final decision was, despite rumors to the contrary, still being negotiated. That decision was due this spring, and since it is spring, AMD obliged during the Q1 2012 analyst call. They said that 28nm APUs would be starting production at Global Foundries this year, so that means product on the shelves in early 2013.
Given TSMC’s woes of late, yield issues mostly behind them, capacity still at issue, it makes sense to move both chips to GloFo. This wasn’t stated directly, but given Llano and Trinity’s progress, SemiAccurate believes this to be the case. All in all, musical fabs, messy production problems, worse messaging, and at long last, a definitive answer.S|A
Latest posts by Charlie Demerjian (see all)
- Globalfoundries 7nm process isn’t even close to the name - Sep 26, 2016
- ARM upgrades realtime offerings to v8-R and adds Cortex-R52 - Sep 21, 2016
- Everspin and Globalfoundries team up for embedded ST-MRAM - Sep 15, 2016
- Intel’s Xpoint is pretty much broken - Sep 12, 2016
- ARM adds 2048-bit vectors to v8A with SVE - Sep 7, 2016