Merging Memory and Compute

Near-memory and in-memory compute are techniques for reducing computing power—especially for AI. But they mean different things to different companies. Understanding the differences is important for understanding how some AI chips work.
11Apr

Nvidia Hopper Leaps Ahead

The next-generation AI architecture powers the H100 card and DGX-H100 system. The 700W flagship card triples peak performance over Ampere while adding FP8 support for more-efficient training.
11Apr

Agilex Dials M for Memory

Intel announced its first Agilex to copackage DRAM, sporting a pair of HBM2e stacks that deliver up to 32GB. It’s also using the Intel 7 process to further cut power dissipation.
05Apr

PCI Express Gen6 Adopts PAM4

In its sixth generation, PCI Express underwent the biggest changes since its debut in 2003, adopting PAM4 signaling. The new modulation enables data rates to reach 64Gbps per lane.
07Feb

Mobileye EyeQ Ultra Aims for Level 4

The ADAS-processor leader is developing a single-chip solution for self-driving cars, along with its own lidar and radars. By 2025, Mobileye hopes its system will reduce the car buyer’s add-on cost to just $10,000.
01Feb

Ceva Tackles Unstructured Sparsity

NeuPro-M, a licensable deep-learning accelerator, offers up to 16x faster inferencing than the previous generation thanks to its new architecture, Winograd modifications, and unstructured-sparsity engines.
01Feb

Discrete GPUs Cut Power for Laptops

Low-cost laptops get a graphics boost from AMD and Nvidia, which now offer their leading GPU architectures in low-power products that increase gaming frame rates and enable hardware ray tracing.
01Feb

Ryzen 6000 Doubles GPU Performance

AMD’s new laptop processor, code-named Rembrandt, is the first PC processor with integrated hardware ray tracing. Its new RDNA2 GPU also doubles graphics performance relative to the previous generation.
25Jan