NVIDIA and Google are racing to dominate the AI chip market, with NVIDIA's Blackwell and Rubin architectures and Google's TPU v8 and Ironwood v7 leading the shift toward energy-efficient 'tokens per w…
NVIDIA unveiled the BlueField-4 DPU at Hot Chips 2026, a fourth-generation data center processor co-designed into the Vera Rubin platform, featuring ConnectX-9-class networking and Grace CPU-class cor…
A team from Nagoya University and the University of Tokyo used Claude Code to port CReSS, a 250,000-line Fortran typhoon simulator, to GPUs, achieving 162 validated GPU kernels and a 5.1x speedup on a…
NVIDIA released new architectural details and SPEC CPU 2026 benchmarks for its upcoming Vera CPU, revealing a 88-core Olympus architecture with spatial multithreading and up to 1.2TB/s memory bandwidt…
Microsoft and NVIDIA's new Spark PCs, featuring Grace CPU and Blackwell GPU, enable local frontier-model execution. The Spark Governance SDK introduces a governance layer for agent-native Windows RTX …
NVIDIA technologies power 81% of the world's 500 fastest supercomputers, with 376 systems using NVIDIA networking and 238 accelerated by NVIDIA GPUs, according to the latest TOP500 list released at IS…
Nvidia announced the RTX Spark, an Arm-based superchip combining a Grace CPU with a Blackwell RTX GPU, at GTC Taipei on May 31, 2026. The chip delivers up to 1 petaflop of AI compute in a slim 14mm la…
NVIDIA and Apple have resolved the hardware bottleneck for on-device AI, with NVIDIA's RTX Spark (Blackwell GPU, Grace CPU, 128GB unified memory) and Apple's M-series chips enabling 4B+ parameter mode…
Nvidia's DGX Spark, a compact personal AI supercomputer built around the GB10 Grace Blackwell Superchip, delivers up to 1 petaflop of FP4 AI performance and can run models up to 200 billion parameters…
NVIDIA announced the RTX Spark "superchip" at Computex 2026, combining a Blackwell RTX GPU with 6,144 CUDA cores and a 20-core Grace CPU for AI, creative, and gaming workloads in laptops. The company …
Researchers from UC Berkeley's UCCL project released mKernel, a library of persistent CUDA kernels that fuse intra-node NVLink communication, inter-node RDMA, and compute into a single kernel to addre…