AMD Launches Triton Inference Server with ONNX Runtime Backend
AMD released Triton Inference Server with ONNX Runtime backend for AMD GPUs, supporting dynamic batching and MIGraphX optimization.
Archive
Browse all published articles.
ARCHIVE
-
-
AMD's Gluon GEMM Tutorial Achieves Near-Peak Performance on MI355
AMD's Gluon GEMM tutorial on MI355 GPUs reaches 99% MFMA efficiency with FP16 kernels, achieving 1489 TFLOPS through iterative optimization.
-
AMD Ryzen AI Max+ Enables 100B+ Parameter LLM Inference
AMD Ryzen AI Max+ processors with 128 GB unified memory allow running 100B+ parameter models locally, eliminating the need for multiple GPUs or cloud services.
-
Alta Daily Uses Meta's Segment Anything to Digitize Wardrobe
Alta Daily, a fashion app, uses Meta's Segment Anything Model to process over 20 million images, helping users visualize outfits on their avatars.
-
Meta Unveils Segment Anything Model 3 with Enhanced Video Processing
Meta released Segment Anything Model 3 (SAM 3.1), improving video processing speed by doubling throughput to 32 frames per second using a single H100 GPU.
-
Meta Unveils MTIA AI Chips for Scalable AI Infrastructure
Meta has deployed hundreds of thousands of MTIA chips to power AI experiences for billions of users, with new generations set for 2026 and 2027.
-
AMD Introduces 4-Wave Interleave FP8 GEMM for Enhanced Performance
AMD's 4-wave interleave FP8 GEMM design improves performance by doubling register file size, enabling full 128×128 output tile storage. The update builds on prior 8-wave ping-pong implementation.
-
Hugging Face Server at Capacity, Limits User Access
Hugging Face's server capacity reached 363 sessions, exceeding the 200-session limit, causing disruptions for users since over two days.
-
xAI Enables Grok Integration in Kilo Code
xAI announced today that users with SuperGrok or X Premium+ subscriptions can now use Grok models inside Kilo Code, an open-source agentic coding platform.
-
X Launches Grok Build CLI for Developers
X announced Grok Build, a new coding agent available in early beta for SuperGrok and X Premium Plus users, on May 25, 2026.
-
xAI Integrates Grok into OpenClaw Agent Platform
xAI enables Grok model usage in OpenClaw, an open-source personal assistant, starting May 19, 2026.
-
xAI Launches Grok Skills for Web, iOS, and Android
xAI today launched Grok Skills, a new feature that allows users to teach Grok once and have it remember across all conversations, with no setup required.