[globalToShmemAsyncCopy] - Starting...
GPU Device 0: "Hopper" with compute capability 9.0

MatrixA(1280,1280), MatrixB(1280,1280)
Running kernel = 0 - AsyncCopyMultiStageLargeChunk
Computing result using CUDA Kernel...
done
Performance= 5289.33 GFlop/s, Time= 0.793 msec, Size= 4194304000 Ops, WorkgroupSize= 256 threads/block
Checking computed result for correctness: Result = PASS

NOTE: The CUDA Samples are not meant for performance measurements. Results may vary when GPU Boost is enabled.