Finally, Intel's Mr. Silicon admits Intel does not lead in silicon technology for SOC
"Itís fair to say Intel didnít have much of a focus four or five years ago on SoCs, but thatís changed,Ē said Mark Bohr",
plus more at
Intel's mobile/Atom 22nm SOC in market in late 2013 is 2 years behind Foundry/ARM 28 nm shipping now (foundry 28nm and Intel's SOC 22nm has about same transistor density making it a reasonable point of comparison).
Intel's silicon group has been the biggest hindrance to getting into mobile market.
Intel's move to finFET has also been a major setback to getting into mobile market. Intel finFET SOC have been talked about for 2 years but are still another 1.5 years away from chips on market. Intel's finFET SOC is late since large delay in moving all the design IP due to issues with transistor matching, analog, RF , and I/O are all not attractive on a bulk finFET SOC vs since at total chip system level intel's 22nm finFET has negligible advantage over foundry 28nm planar transistor.
Means if foundry product requires a full range of transistors (low leakage,RF, quality analog †etc.) for a mobile 22nm SOC, that is not offered even for internal intel product until end of 2013. †Mobile SOC intel foundry †customers thus at a disadvantage †compared to tsmc and other foundry.†
I can't think of any way that Intel's 22nm process integration is cost competitive against the foundry cost. Unless the finFET yields at the foundries are really bad ( a possibility since it would be their first generation)
Blog Doing Math in FPGAs Tom Burke 2 comments For a recent project, I explored doing "real" (that is, non-integer) math on a Spartan 3 FPGA. FPGAs, by their nature, do integer math. That is, there's no floating-point ...