Intel’s Lion Cove contains significant departures from previous CPU microarchitectures, prioritizing performance efficiency by omitting hyperthreading and other performance-boosting innovations but still manages to deliver a double-digit IPC uplift.
Microsoft’s Copilot+ PC demands saw Qualcomm, Intel and AMD all launch new PC SoCs with advanced NPUs. Intel’s Gaudi 3 will compete with NVIDIA—although only NVIDIA’s previous generation H100. Advanced packaging is a significant industry trend, and AR is a driving force behind new packaging technologies from TSMC seen in the Apple R1, and an experimental SoC from Meta.
Intel’s Lunar Lake, the biggest overhaul of its architecture for PC processors since the addition of efficiency cores with Alder Lake, promises a big leap in power efficiency that may nullify AMD’s long-standing advantage in this metric.
Apple’s R1 sensor hub chip in the Vision Pro AR/VR headset is the first use that TechInsights has seen of TSMC’s InFOM advanced packaging technology. The R1 has a 5nm processor chip plus two LLW DRAMs linked by short very-high-density interconnect to speed up the data processing.
Intel’s Gaudi 3 AI accelerator may not feature the headline performance numbers of NVIDIA’s Blackwell, but is nonetheless a competitive offering on price, power, and inference efficiency.
Interest in Windows on Arm (WoA) is at a high point. Microsoft and Arm must take the opportunity to plug the remaining gaps and create a broader WoA ecosystem.
Processor companies are packing more neural processing capability into processors for IoT edge devices. Here’s a look back at several of these that we covered in May.
Meta’s Reality Labs discussed a prototype AR processor at the 2024 ISSCC, fabricated using hybrid bonding as 3D packaging. This improved performance for the trial tasks of hand tracking and image processing.
Synaptics’ new Astra platform aims to ease the integration of more AI compute capability in IoT edge devices with Arm-based SoCs featuring AI acceleration engines and an ecosystem of supporting software and development tools.
Arm has updated its neural processing technology, offering its EthosU85 as a follow-on to the prior EthosU65. It positions Arm as having a critical new technology for the edge.
AMD’s latest series of processors for the enterprise market, the Ryzen PRO 8000 series, offers an upgraded AI engine that improves AI acceleration for businesses and professional users.
Arm has bolstered its automotive offering with CPUs, an image-signal processor, and multicore cluster infrastructure blocks. Configuration options enable ASIL B or ASIL D safety.
AMD plans a second generation of Versal AI Edge devices that integrate more CPUs with a more efficient hardware accelerator to bring more AI inference capability to edge applications.
NVIDIA’s Blackwell pods bring exascale compute to a rack while pushing the power envelope to 120 kW. The DGX GB200 NVL72 pod offers up to 1.44 ExaFLOPS of FP4 performance.
Cadence has updated its Vision DSP line to handle radar data types and instructions. It’s also offering an accelerator for high-FFT workloads such as 4D imaging radar.