Like many recent generation CPUs, Raptor Lake was designed at the company’s Israel Development Center (IDC). The chip is also being manufactured in Israel at the company’s Fab 28, as well as at other locations around the world. Final details on 13th-gen Core processors are expected to be unveiled at the Intel Innovation event in San Francisco at the end of September, but for now the company was eager to share that it will support speeds of up to 6 GHz (and the ability to be overclocked up to 8 GHz). Many of the more interesting announcements from the Israel Tech Tour, however, centered on other elements of a PC beyond the CPU. This is a continuation of the trend towards emphasizing performance and overall experience (vs. raw numbers) that we’ve seen over the last few years. On the wireless front, Intel announced that Intel Connectivity Suite version 2.0 will come with 13th-gen Core PCs. The first version included the ability to aggregate wired Ethernet and wireless Wi-Fi connections into a single, fatter data pipe, and now version 2.0 adds support for cellular connections as well.
This will let you combine the throughput of two Wi-Fi connections (2.4 GHz, 5 GHz and 6 GHz frequency options) as well as a 5G cellular connection for PCs equipped with the appropriate 5G modems. The result is the fastest wireless connection possible on a single PC (dependent on the speed of the individual connections, of course). You can’t use the aggregated connection for a single application, but you can put different apps on different connections to optimize the overall connectivity experience. The Intel software will also let you select and optimize different workloads – like Teams, Zoom, Webex video conference calls, streaming, and others – and have them directed to the best possible connection. In the event of one connection dropping, such as a Wi-Fi connection in a busy environment, the application will automatically jump to the other one to avoid having your video stream drop on a call.
Unfortunately the new application will be limited to new PCs with both 13th-gen Core PCs and Intel Wi-Fi components. It would be tremendously useful to have it work with different combinations of components from other suppliers as well. Nevertheless, it’s a good proof-of-concept for the ability to leverage the combination of multiple different wireless connection technologies on a single device. On the AI front, Intel discussed a dedicated neural processing unit (NPU) codenamed “Keem Bay” that will be coming to 13th-gen Core mobile CPUs, which will launch after the desktop parts.
Based on the Movidius VPU technology that Intel purchased several years ago, Keem Bay marks the first time that Intel has incorporated an AI accelerator directly into its PC-focused silicon. This is a big step towards a PC SoC design that along with conceptually similar advancements that AMD is bringing to its Ryzen CPUs, should lead to a whole new range of AI-accelerated PC software. As you probably know, smartphones have benefitted from AI accelerators for many years now, but PCs are still lacking them. With the integration of dedicated neural processing and AI acceleration engines into next generation CPUs, we should start to see interesting new applications coming to PCs as well. Initially, most of the work will likely be on things such as audio and video enhancements. At present, these tasks are handled primarily by the CPU or, in a few cases, by GPUs, but NPUs will be able to do execute them more efficiently and with a higher quality. Laptop PCs with NPUs should see significantly improved battery life for heavy videoconferencing app users. Computational photography, video editing/content creation and gaming are likely to benefit from the presence of NPUs.
Down the road we’re likely to see other innovative applications – many of which, I’m convinced, haven’t even been thought of yet. In the meantime, computational photography, video editing/content creation and gaming are also likely to benefit from the presence of NPUs. Another bit of good news that we learned from Intel is that algorithms written in popular AI frameworks such as PyTorch and TensorFlow, can be easily ported among different NPUs (from Intel or AMD). This is extremely important and could have been a serious roadblock for software developers interested in working on AI-enhanced applications for PCs. In addition, the overhead of running these algorithms in a software abstraction layer that sits on top of the NPU hardware are extremely modest (measured in tenths of a percent vs. writing directly to the hardware, which is a significantly harder and more time-consuming task).
Intel also discussed enhancements to onboard webcams by integrating image signal processing (ISP) technology into Raptor Lake. This can be used to generate better image quality, but also for new features like camera tracking, where the onboard camera follows you as you walk around the room, or presence detection, where the PC automatically wakes or goes to sleep depending on whether it senses a person is in front of the screen. Enhancements to Bluetooth capabilities on PCs, including adding support for Bluetooth LE Audio-based audio broadcasting, are also on tap for 13th-gen Core PCs. Performance improvements on new CPUs are what most people tend to focus on when they think about future PCs. And, to be clear, both Intel and AMD’s next generation CPUs will offer nice raw performance boosts. But these system-level upgrades brought on by new platforms are also the kind of enhancements that consumer and business PC users will primarily benefit from in the years to come. Bob O’Donnell is the founder and chief analyst of TECHnalysis Research, LLC a technology consulting firm that provides strategic consulting and market research services to the technology industry and professional financial community. You can follow him on Twitter @bobodtech.
