Software tools & optimizations#
Discover the latest blogs about ROCm software tools, libraries, and performance optimizations to help you get the most out of your AMD hardware.
Scaling MiniMax-M3 Inference with Distributed Serving and Operator Co-Design on AMD Instinct MI355X GPUs
Optimize MiniMax-M3 inference on AMD Instinct™ MI355X GPUs with ATOM online quantization, AITER sparse attention, FP8 KV cache, and EAGLE3.
Understanding Attention Algorithms and Their Backends for Image and Video Generation
Practical guide to attention backends in ComfyUI on AMD describing how to optimize performance, memory, and stability with the right configuration.
SPIR-V on ROCm: A Portable IR for AMD GPUs
Learn how SPIR-V brings compile-once, specialize-on-device portability to AMD GPUs — with a reproducible HIP benchmark, trade-off analysis, and quick-start guide.
Performance Profiling on AMD GPUs – Part 5: Profiling-Driven Kernel Optimization with an AI Code-Assist Tool
Ready to slash HIP kernel runtimes? See how ROCm profiling + an AI code-assist agent delivered a 28.3× speedup on AMD Instinct MI250.
From Vector Search to Agentic RAG: Building an Enterprise Research Analyst with hipVS
Learn how to build an agentic RAG research assistant using hipVS GPU-accelerated vector search on AMD Instinct GPUs, with multi-query decomposition, parallel retrieval, and cited sources synthesis.
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.
LogsLop: A Tiny Summarization Tool for Enormous Log Files
LogsLop deduplicates repetitive log lines so humans and LLMs can find failures in enormous log files.
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.
Porting High-Performance HIP Kernels to FlyDSL
This blog post shows how to port HIP C++ GPU kernels to FlyDSL, AMD's new Python DSL, matching hand-tuned C++ performance with less code.
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.
Occupancy Math on the AMD MI355X GPU (CDNA4): A From-First-Principles Guide
Derive MI355X GPU (CDNA4) occupancy by hand: the four limiters, MXFP8 GEMM examples, and why matrix-bound kernels hit peak throughput at low occupancy.