
Perform 3D tensor-matrix multiplication of two tensors:
$$
C[i][j][l] = \sum_{k=0}^{K-1} A[i][j][k] \cdot B[k][l]
$$

## Input
- Tensor $A$ of size $N \times M \times K$
- Matrix $B$ of size $K \times L$

## Output
- Tensor $C$ of size $N \times M \times L$

## Notes:
- All tensors $\text{A}$, $\text{B}$, and $\text{C}$ are stored in row-major order
- This problem is adapted from [KernelBench](https://github.com/ScalingIntelligence/KernelBench/blob/main/KernelBench/level1/10_3D_tensor_matrix_multiplication.py)

Rank	User	GFLOPS	Runtime (ms)	GPU	Language	Date
#1	sagarreddypatil	50.73T	197.44	B200	Triton	4 months ago
#2	sagarreddypatil	41.44T	244.25	H100	Triton	5 months ago
#3	648trindade	11.29T	876.85	B200	CUDA C++	about 2 months ago
#4	648trindade	10.14T	974.95	H100	CUDA C++	about 2 months ago
#5	648trindade	8.91T	1111.02	H200	CUDA C++	2 months ago
#6	648trindade	6.5T	1550.24	L40S	CUDA C++	2 months ago
#7	harmya	6.43T	1538.43	H100	CUDA C++	6 months ago
#8	Firojpaudel	6.08T	1994.18	H100	CUDA C++	5 months ago
#9	ShlokVFX	5.34T	1853.45	H100	CUDA C++	5 months ago