AI Blogs#
When a Faster Kernel Doesn't Speed Up Serving: Profiling FP8 KV Cache on AMD Instinct MI308X
Learn how a 34% faster FP8 KV cache kernel delivered 0% E2E speedup, and how profiling attribution exposed the hidden dtype-cast cost on MI308X.
Local Image and Video Generation on AMD Ryzen™ AI Max+ Processor (Windows)
Run ComfyUI natively on Windows on AMD Ryzen AI Max+ with ROCm 7.2.1—SDXL, Flux, and video workflows on the Radeon 8060S, no WSL.
GEAK Agent-Driven Optimization of the DeepSeekV4 MLA Kernel
GEAK Agent accelerates DeepSeekV4 MLA kernel optimization with Triton and delivers SGLang E2E gains on AMD GPUs.
Serving NVFP4 Models on AMD Instinct™ MI355 Accelerators
Learn how to serve NVFP4 models on AMD Instinct™ MI355 using an emulation pipeline in vLLM — no format conversion needed.
QuickReduce INT3 Quantization and Benchmarking on MI355
Learn how QuickReduce uses INT3 quantization to accelerate all-reduce communication and evaluate its performance and accuracy on AMD Instinct MI355 GPUs.
Triton-Based Optimization of Video Sparse Attention on ROCm
Optimize video sparse attention on ROCm with GEAK and linear global context for faster, more stable video generation on AMD GPUs.
Fast Image Generation and Editing with SGLang Diffusion on AMD GPUs
Serve and benchmark diffusion models for image generation and editing on AMD Instinct GPUs using SGLang Diffusion on ROCm.
AMD Instinct™ Network Traffic, Congestion Trends, and Harmonics in Scale-Out Networks for AI Training Clusters
Explore how synchronized GPU collectives create harmonic congestion in AI clusters and the strategies to diagnose and mitigate it.
SGLang-ATOM: Bring ROCm-Native Acceleration to SGLang Serving
Explore how SGLang-ATOM connects SGLang serving applications with ROCm-native ATOM execution to accelerate LLM inference on AMD Instinct GPUs.
Towards Feature Complete Triton Support in JAX-Triton
Learn what new features were added to JAX-Triton and how that could help you write or reuse more efficient and readable GPU kernels in JAX.
Efficient Hyperparameter Optimization for Autonomous Driving Models with AMD Instinct GPU Partitioning
Accelerate HPO for autonomous driving models using AMD MI300X GPU partitioning for higher throughput, efficiency, and parallelism.
Primus Tuning Agent: Closing the Configuration-Search Loop
Use the Primus Tuning Agent to automatically find optimal LLM training configurations on AMD Instinct GPUs.