Latest AI and tech news
Nokia today announced growing global momentum for AI-RAN, with operators across North America, Europe, Asia-Pacific and the Middle East advancing AI-RAN from early evaluation to lab and live field trials, accelerating the evolution toward AI-native 6...
Air pollution is a serious public health risk, contributing to an estimated 30,000 deaths in the U.K. alone last year. Data-driven insights can help — but computing air quality with traditional chemistry-based models is expensive, which limits how de...
A week of NVIDIA coverage measured three ways. The naive headline query finds only 30% of it....
Learn how to integrate a medical knowledge graph with Poro2 LLM via MCP to simplify medical text on AMD Instinct MI300X GPUs
GitHub user ch32-riscv-ug has developed wch-protocols, a knowledge base for the protocols behind flashing and debugging WCH’s CH32 RISC-V chips....
On a sweltering August evening in Silicon Valley, as the sun dropped and air conditioning loads spiked, Silicon Valley Power sent a signal to an AI factory to adjust its power consumption....
Ian Buck, vice president of hyperscale and high-performance computing at NVIDIA, Tuesday spoke on AI factory efficiency at the AI Infra Summit, the Santa Clara Convention Center event that has morphed into a Coachella of infrastructure tech....
Power is a defining constraint for AI factories. As AI workloads demand a full compute platform to serve them, each component of that platform must maximize output within the factory’s limited power budget. This makes performance per watt—rather than...
For operators of large-scale AI factories, maximizing continuous output is essential for productivity. In massive-scale AI training, every GPU in the cluster must synchronize gradients across thousands of collective operations per second. Similarly, ...
Together with partner NVIDIA, Salesforce today at Dreamforce Conference 2026 unveiled a new CRM domain-specific reasoning model for Agentforce dubbed Koa....
Federated learning (FL) projects often begin with a straightforward setup: one server, a few clients, and one dataset at each site. As those projects grow, the challenge shifts from running an algorithm to operating shared infrastructure. GPUs must b...
rh1tech has published FRANK-386, an i386 PC emulator that runs on the RP2350, the chip on the Raspberry Pi Pico 2. It is based on Tiny386 by Chunhui He, ported to the microcontroller by Mikhail Matveev and DnCraptor....
See how AMD brings local AI and AI-ready workstations to Autodesk University 2026 for demanding design, engineering, and AEC workflows.
gpasquero has published the source for Vintage Emulator Studio (VES) on GitHub. VES is an audio plug-in for macOS (Apple Silicon and Intel), Windows, and Linux, in AU and VST3 formats where supported, plus a standalone application. It wraps a curated...
In my last post, we designed a dual Roaring Bitmap inventory engine that mapped our GPU rack runtime availability status to a tiny payload. We set up Server-Sent Events (SSE), tested it on localhost, watched the latency plummet to sub-millisecond lev...
Koa is trained with 27 years of Salesforce CRM intelligence to enable agents to reason through the complex, multistep tasks required for enterprise work...
Palantir Foundry integrates NVIDIA cuOpt to optimize supply chain allocation with GPU-accelerated routing. Officially launched September 11, 2026.
Build high-performance BF16 GEMM kernels on Helios GPUs with HipKittens, from a naive baseline to optimized schedules.
Inside SB Energy — the SoftBank-backed, Nvidia-aligned infrastructure developer racing to turn AI’s $500 billion promise into gigawatts…...
Selecting the right serving engine for your embedding model can dramatically outperform hardware upgrades, yielding up to an 11x throughput increase on the same GPU.
How Domyn built sovereign AI models (Italia, Colosseum, Domyn Large/Small) using NVIDIA NeMo, Megatron-Core & Leonardo supercomputer. Technical insights on pruning, distillation, RL & EU AI Act compliance.
Reconstruct which models, data, policies and GPU runtime an AI avatar used—without assuming identical generative output or retaining every conversation.
Getting a model serving on day one is the easy part. The expensive half of running a model catalog is maintaining every model, precision, GPU, and inference engine combination as the stack underneath keeps moving.
How Speculators and Mooncake enabled multi-node DSpark training for Kimi K3.
Novita AI has open-sourced Chord, a high-performance W4A16 MoE CUDA kernel for Kimi K2.x serving shapes, with a Humming-compatible indexed path and grouped SM90 operators.
Rubin is the first platform co-designed across six products for the agentic era: Rubin GPU, Vera CPU, NVLink 6 Switch, ConnectX-9, BlueField-4, and Spectrum-6. Today we are publishing the first verified agentic inference results for Rubin, measured o...
A new project called zSST is a SystemVerilog implementation of the 3dfx Voodoo Graphics chipset, the SST-1. Paired with the z486 soft CPU and the rest of a PC’s hardware, it forms z486 XL: a DOS machine with Voodoo graphics running entirely in the pr...
Enterprises have a rhythm. A semiconductor company may investigate thousands of yield excursions, compare thousands of lots, or trace problems across products, tools and process steps. The exact questions change, but the patterns of analysis repeat....
Ken Shirriff has published another look inside Intel’s 8087 floating-point coprocessor on righto.com, this time tracing the microcode behind a single instruction: FSCALE. He is working with the Opcode Collective, a group reverse engineering the chip’...