FlagBLAS User Guide

Use FlagBLAS

FlagBLAS integrates directly with PyTorch. Import the package and call operators on CUDA tensors:

import torch
import flag_blas

# Create tensors on CUDA
a = torch.randn(1024, 1024, device='cuda')
b = torch.randn(1024, 1024, device='cuda')

# Matrix multiplication (GEMM)
c = flag_blas.ops.sgemm(a, b)

Operator list

The complete operator registry is maintained at FlagBLAS conf/operators.yaml.