Intel Details Silvermont Microarchitecture For Next-Gen Atoms
crookedvulture writes "Since their debut five years ago, Intel's low-power Atom microprocessors have relied on the same basic CPU core. That changes with the next generation, which will employ an all-new Silvermont microarchitecture built using a customized version of Intel's tri-gate, 22-nm fabrication process. Silvermont ditches the in-order design of previous Atoms in favor of an out-of-order approach based on a dual-core module equipped with 1MB of shared L2 cache. The design boasts improved power sharing between the CPU and integrated graphics, allowing the CPU cores to scale up to higher speeds depending on system load and platform thermals. Individual cores can be shut down completely to provide additional clock headroom or to conserve power. Intel claims Silvermont doubles the single-threaded performance of its Saltwell predecessor at the same power level, and that dual-core variants have lower peak power draw and higher performance than quad-core ARM SoCs. Silvermont also marks the Atom's adoption of the 'tick-tock' update cadence that guides the development of Intel's Core processors. The successor to Silvermont will be built on 14-nm process tech, and an updated microarchitecture is due after that."
Silvermont is a just core (CPU). It sits inside an SoC (system on chip), and your final power figures will still depend on the efficiency of the rest of the SoC (the GPU, the IO interfaces, the memory interfaces, any other dedicated hardware, etc.). And even then, the integration of technology is getting to the point where the SoC's power consumption is only a partially limiting factor in battery life. During lower power states and standby states, the comms units, the display, etc. can all consume way more power than the core.
If power consumption when lightly loaded is competitive with ARM, then Intel may have something. Peak power consumption isn't as important for devices where the cpu is never pegged, or only pegged for a tiny fraction of a percentage of total time the cpu is running.
I have one arm dev board with an exynos4 on it, that has a huge heatsink on top. Pull the heatsink, and you never get even close to speed/power consumption when running with heatsink at 100% cpu. I have yet to see a phone with a heatsink as big as the phone, so I suspect that these phones *never* see 100% cpu, or only see it for such a short period of time (before thermal throttling takes place), that peak power usage is meaningless for most devices using arm SOCs.
I hope Intel pulls it off. It would be nice if power consumption factored larger in their other offerings too.
Intel's Atom processors typically retail for $30-$80, with some being more, some less. OEM pricing is lower. That's just the CPU so it's not directly comparable to ARM-based SoCs which I hear cost about $15-$25. So Intel is substantially higher priced, but not ruinously so from an end-user's purchase standpoint. Certainly not $650.
The more interesting thing to watch will be how this impacts the broader computing market. Intel has managed to stay ahead of the competition buoyed by the enormous profits it generates from its Core CPUs, which typically sell for $100-$400. As CPUs get faster, the general population can get by with something lower down the product chain. I've already been recommending i3s to most of my customers for the last couple years. I'm very close to dropping the bar to high-end Atom or AMD CPUs. As more and more of Intel's sales shift towards these lower-end CPUs, their overall profit margin will start to dry up. It's going to be interesting to watch how they'll react to that.