Systems - Software Tools & Optimizations#
DI Series: Serving Kimi-K3 MXFP4 Wide-EP Disaggregated on AMD Instinct MI300X and MI325X
Learn how to serve a 1.4 TiB Kimi-K3 MXFP4 mixture-of-experts model disaggregated across AMD Instinct MI300X and MI325X GPUs.
Hyperloom: A Multi-Agent Harness for Autonomous Inference Optimization on AMD GPUs
Hyperloom is a multi-agent harness that autonomously optimizes LLM inference on AMD Instinct GPUs, reaching a median 1.73x throughput gain.
veRL on AMD: Production-Ready RL Post-Training on ROCm
Run veRL RL post-training on AMD Instinct GPUs with a turnkey ROCm container, AITER-accelerated rollout, and validated accuracy on MI300 and MI355.
DI Series: Scaling GLM-5.1-FP8 to 64 MI300X GPUs
Scale GLM-5.1-FP8 long-context serving beyond a single node on AMD Instinct MI300X with WideEP prefill-decode disaggregation and MoRI.
Spur: Modern GPU Job Scheduling for HPC and AI Workloads
Explore how Spur addresses pain points in GPU cluster management and how Spur-Cloud extends the platform into a complete GPU-as-a-Service solution.
Introducing ROCm™ AMD Infinity Context: A Purpose-Built KV Cache Tier for Distributed Inference
Explore ROCm AMD Infinity Context (AIC), AMD's open KV cache tier built on AMD Infinity Storage for distributed LLM inference.
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.
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.
RDC and RocProfiler Compared to DCGM for Commonly Used Metrics
Learn how CLI commands and Python code help you evaluate app performance without a profiler, with examples explaining what each metric means.
Primus Tuning Agent: Closing the Configuration-Search Loop
Use the Primus Tuning Agent to automatically find optimal LLM training configurations on AMD Instinct GPUs.
Primus Projection: Estimate Memory and Performance Before You Train
Learn how to use the Primus projection tool to estimate memory and performance for large-scale LLM training on AMD Instinct™ accelerator platforms.
Agentic Diagnosis for LLM Training at Scale
Explore how AI agents diagnose LLM training incidents — from RCCL hangs to throughput regressions — in one prompt with MaxText-Slurm.