Intel Confirms 10nm Ice Lake CPU Shipments Ramp In June, Roadmap To 7nm And Beyond

Intel Engineer
For Intel, the road to 10 nanometers has not been a smooth or easy one. That is why we are still waiting on volume shipments of 10nm CPU products, even though the original plan was to already be at this point a long time ago. So, where are things now? According to the latest update from Intel, volume 10nm CPU shipments will finally commence next month.

Intel will kick off its volume 10nm play in mobile, with its Ice Lake platform. The new chips will start shipping in June, bringing with them "approximately 3 times faster wireless speeds, 2 times faster video transcode speeds, 2 times faster graphics performance, and 2.5 to 3 times faster artificial intelligence performance" compared to past generation products.

intel Ice Lake

Ice Lake will kick off the 10nm party in volume form for mobile devices, though it will not be a one-and-done deal for that process node.

"As announced, Ice Lake-based devices from Intel OEM partners will be on shelves for the 2019 holiday season. Intel also plans to launch multiple 10nm products across the portfolio through 2019 and 2020, including additional CPUs for client and server, the Intel Agilex family of FPGAs, the Intel Nervana NNP-I (AI inference processor), a general-purpose GPU and the ;Snow Ridge; 5G-ready network system-on-chip (SOC)," Intel said.

The latest update is in line with comments Intel CEO Bob Swan recently made during a conference call with investors. At the time, he said confidence in 10nm was "improving," and that Intel remained on track to have volume client systems on store shelves in time for the holiday shopping season.

intel nodes

At a cursory glance, it would appear that Intel has fallen way behind AMD, which is getting ready to announce a new round of Ryzen processors built on a 7nm manufacturing process. AMD is firing on all cylinders for sure, and finds itself in a good place—it's been able to nab some additional CPU market share from Intel, as a results of Zen.

Simply comparing node sizes grossly oversimplifies the underlying technology, as it completely ignores things like gate pitch, transistor count, and so forth. That's not taking anything away from AMD, which is using a more advanced process node at the moment. However, Intel's 10nm is roughly on par with AMD's 7nm. Both will start shipping around the same time, though AMD will be hitting the desktop first.

Intel Roadmap

Looking further ahead, Intel also has plans of transitioning to its own 7nm node, starting in 2021. By then, Intel expects to have squeezed out at least three generation of 10nm products as it refines the node over time (10nm+ and 10nm++).

Likewise, Intel expects to refine its 7nm node over time, culminating in 7nm++ products in 2023. Of course, things are bound to change and get shuffled around when looking that far into the future, but as of this moment, that is Intel's roadmap. After that, it will be interesting to see if 5nm products arrive, or if some other technology emerges.

One thing to not is that Intel also plans on using 7nm GPUs around the same time as the first 7nm CPUs arrive.

"The lead 7nm product is expected to be an Intel Xe architecture-based, general-purpose GPU for data center AI and high-performance computing. It will embody a heterogeneous approach to product construction using advanced packaging technology. On the heels of Intel’s first discrete GPU coming in 2020, the 7nm general purpose GPU is expected to launch in 2021," Intel chief engineering officer Dr. Mrthy Renduchintala said.

Intel 7nm GPU

On other side (CPU or GPU), Intel is claiming that the first updates on its 7nm process technology will deliver 2 times scaling and a 20 percent increase in performance per watt, with a 4 times reduction in design rule complexity.

Those are bold claims to go along with an aggressive timeline, considering the trouble Intel has had with 10nm. However, this will mark the first commercial use of extreme ultraviolet (EUV) lithography, which Intel says will help drive scaling for multiple node generations.