Changelog for fixing vec-add sample

Signed-off-by: Jay Gu <jagu@nvidia.com>

Author: haijieg

changelog.d/fix-vec-add.md

@@ -0,0 +1,4 @@
+<!--- SPDX-FileCopyrightText: Copyright (c) <2025> NVIDIA CORPORATION & AFFILIATES. All rights reserved. -->
+<!--- SPDX-License-Identifier: Apache-2.0 -->
+
+Fix redundant check in vectoradd sample (by @NikeNano)

samples/templates/VectorAddition.py

@@ -80,13 +80,6 @@ def vec_add(a: torch.Tensor, b: torch.Tensor, use_gather: bool = False) -> torch
         # (TILE_X * TILE_Y) around 1024 (a common block size limit for threads).
         TILE_X = max(1, 1024 // TILE_Y)
 
-        # Further adjustment to ensure TILE_X * TILE_Y is not excessively large
-        # if N (and thus TILE_Y) is small, or to prevent TILE_X from becoming zero.
-        if TILE_X * TILE_Y > 1024 and TILE_X > 1:
-            TILE_X = 1024 // TILE_Y
-            if TILE_X == 0:
-                TILE_X = 1  # Ensure TILE_X is at least 1
-
         # Calculate the 2D grid dimensions for launching the kernel.
         # `math.ceil(M / TILE_X)` blocks along rows, `math.ceil(N / TILE_Y)` blocks along columns.
         grid = (math.ceil(M / TILE_X), math.ceil(N / TILE_Y), 1)