AMD ROCm 7.13.0

AMD ROCm 7.13.0#

AMD ROCm is an open, modular, and high‑performance GPU software ecosystem — built collaboratively with the community, maintained transparently, and optimized for consistent, scalable performance across data centers, workstations, and edge devices.

ROCm 7.13.0 is built with TheRock, AMD’s open build and release system. TheRock replaces the previous monolithic release process with a modular workflow that makes ROCm components easier to build, integrate, and distribute. See the release notes for more information.

What is ROCm?#

ROCm is the AMD open software stack for GPU‑accelerated computing. It provides the tools needed to program AMD GPUs — including runtimes, compilers, performance and system utilities, and optimized math and compute libraries. The wider ROCm ecosystem includes ROCm‑enabled HPC applications and deep learning frameworks such as PyTorch.

Some key features:

  • Open source – Transparent development driven by community feedback

  • Cross‑platform – Supports Linux and Windows environments

  • Comprehensive – End‑to‑end toolchain from compilers to libraries

  • Performance‑focused – Tuned for AMD Instinct™, AMD Radeon™, and AMD Ryzen™ devices

Enterprise applications
Customer service
Sales assistant
Human resources
Marketing
Personal assistant
Coding assistant
Drug discovery
Material science
AMD ROCm™ software ecosystem
AMD enterprise AI
AMD Resource Manager
AMD Solution Blueprints
AMD Developer Console
AMD Inference Microservices
AI ecosystem
Hugging Face
PyTorch
JAX
TensorFlow
ONNX
Triton
DeepSpeed
OpenXLA
Cluster operations
Monitoring:
  • Prometheus
  • Grafana
MLOps platforms:
  • mlflow
  • Kubeflow
Model serving runtimes:
  • vLLM
  • PyTorch
Vertical SDKs
ROCm-DS
ROCm-LS
ROCm-Finance
ROCm-Simulation
ROCm Core SDK
Math and compute libraries
Communication libraries
Support libraries
Media libraries
Storage
Runtimes and compilers
Profiling and debugging tools
Control and monitoring tools
Systems and infrastructure
Kubernetes:
  • GPU Operator
  • Device Plugin
  • Container Toolkit
  • Device Metrics Exporter
  • Device Plugin
AMD GPU Driver and runtime firmware
AMD Virtualization Driver
AMD Instinct™ Fabric Manager
Management tools
Supported OSes
Ubuntu
Debian
RHEL
Oracle Linux
Rocky Linux
SLES
Windows
OEM/ODM firmware
Baseboard and GPU firmware (platform-optimized firmware)
AMD GPUs and APUs

ROCm supports AMD GPU architectures spanning data center, workstation, and APU product lines. TheRock enables a unified ROCm user‑space experience across devices.

  • AMD Instinct GPUs – Purpose‑built for large‑scale compute, AI training, and HPC workloads.

  • AMD Radeon GPUs and AMD Ryzen AI APUs – Designed for workstations, desktop computing, and edge AI applications.

See AMD hardware support for the complete list of supported hardware.

ROCm Core SDK#

The ROCm Core SDK provides the foundational components that power the ROCm ecosystem — runtimes, compilers, math libraries, and system utilities for GPGPU computing. See /components/core for more information.

ROCm™ Core SDK
Math and compute libraries
Deep learning and GPU compute kernels:
Composable Kernel, MIOpen, rocWMMA, hipDNN (coming soon)
BLAS dense and sparse libraries:
hipBLAS, hipBLASLt, hipSOLVER, hipSPARSE, hipSPARSELt, rocBLAS, rocSOLVER, rocSPARSE
Kernel primitives:
hipCUB, rocPRIM, rocThrust
Other math:
hipFFT, hipRAND, rocFFT, rocRAND
Communication libraries
RCCL, rocSHMEM
Media libraries
rocDecode, rocJPEG
Storage libraries
hipFile (coming soon)
Runtime and compilers
HIP, HIPIFY, LLVM, SPIRV-LLVM-Translator
Profiling and debugging tools
Profilers:
ROCm Compute Profiler (rocprofiler-compute), ROCm Systems Profiler (rocprofiler-systems), ROCprofiler-SDK
Debuggers:
ROCdbgapi, ROCm Debugger (ROCgdb), ROCr Debug Agent,
Control and monitoring tools
AMD SMI, hipinfo, ROCm Data Center Tool (RDC), rocminfo

ROCm Extra components are supplementary tools for benchmarking, validating, and managing ROCm deployments. These tools are not required for GPU application development but are useful for verifying hardware health, measuring system performance, and managing GPU fleets. For more information, see ROCm Extra components.

Get started#

Contents