submission 669289

+ # Write your code here#!POPCORN leaderboard vectoradd_v2
   
+ # This is a submission template for popcorn leaderboard 'vectoradd'.
+ # Your task is as follows:
+ # > Implement a float16 vector addition kernel.
   
- import torch
- import triton
- import triton.language as tl
- from task import input_t, output_t
+ # >
   
+ # > Input: tuple(torch.Tensor, torch.Tensor) with tensors of shape (N, N) and type torch.float16. These tensors are from
   
- @triton.jit
- def add_kernel(
-     A_ptr, B_ptr, C_ptr, M, N,
-     BLOCK_SIZE: tl.constexpr,
- ):
-     row_idx = tl.program_id(0) * BLOCK_SIZE + tl.arange(0, BLOCK_SIZE)
-     col_idx = tl.program_id(1) * BLOCK_SIZE + tl.arange(0, BLOCK_SIZE)
-     
-     mask_row = row_idx < M
-     mask_col = col_idx < N
+ # > a normal distribution with mean 0 and variance 1.
   
-     A = tl.load(A_ptr + row_idx[:, None] * N + col_idx[None, :], mask=mask_row[:, None] & mask_col[None, :], other=0.0)
-     B = tl.load(B_ptr + row_idx[:, None] * N + col_idx[None, :], mask=mask_row[:, None] & mask_col[None, :], other=0.0)
+ # > Output: torch.Tensor of shape (N, N) and type torch.float16
   
-     C = A + B
-     tl.store(C_ptr + row_idx[:, None] * N + col_idx[None, :], C, mask=mask_row[:, None] & mask_col[None, :])
+ # The deadline for this leaderboard is 2025-04-30 00:00:00+00:00
   
- def custom_kernel(data: input_t) -> output_t:
-     A, B, C = data
-     M, N = A.shape
+ # You can automatically route this file to specific GPUs by adding a line
+ # `#!POPCORN gpus <GPUs>` to the header of this file.
+ # Happy hacking!
   
-     BLOCK_SIZE = 32
-     grid = (triton.cdiv(M, BLOCK_SIZE), triton.cdiv(N, BLOCK_SIZE))
+ from task import input_t, output_t
   
-     add_kernel[grid](
-         A, B, C, M, N,
-         BLOCK_SIZE=BLOCK_SIZE,
-     )
   
-     return C
   No newline at end of file
+ def custom_kernel(data: input_t) -> output_t:
+     """
+     Custom kernel for vector addition.
+     """
+     # Unpack the input tuple
+     s = data[0]
+     t = data[1]
+  
+     # Perform element-wise addition
+     q = s + t
+  
+     # Return the result
+     return q