innrest.blogg.se

Since AOMP is a clang/llvm compiler, it also supports GPU offloading with HIP, CUDA, and OpenCL.

Supports OpenMP target offload on AMD GPUs. Provides details on AOMP, a scripted build of LLVM and supporting software. GPUperfAPI - GPU Performance API, cloning, system requiments, and source code directory layout Includes details on library source tree, steps to build and run the test ROCm-Tracer - ROCm Tracer - provides a generic independent from specific runtime profiler to trace API and asynchronous activity. ROCm-Profiler - Radeon Compute Profiler- performance analysis tool ROCm-GDB - ROCm-GDB tool includes installtion, configuration, and working of Debugger and APIs GCN Assembler Tools - AMDGPU ISA Assembler Math Library API - Includes HIP MAth API with hcRNG, clBLAS, clSPARSE APIsĭeep Learning API - Includes MIOpen API and MIOpenGEMM APIs Inline GCN ISA Assembly Guide - Covers various concepts of AMDGCN Assembly, DS Permute Instructions, Parameters to a Kernel, GPR Counting. Vega - Provides “Vega” Instruction Set Architecture, Program Organization, Mode register and more details. GCN 2.0 - For information on ISA Manual for Fiji and Polaris (AMD Accelerated Parallel Processing technology) GCN 1.1 - For information on ISA Manual for Hawaii (Sea Islands Series Instruction Set Architecture) Performance and optimization for various device types such as GCN devices HIP-Supported CUDA API Reference Guide v4.5.AMD Instinct™ High Performance Computing and Tuning Guide.