From 4792e669f82899a4d4ebaf0eebaf73049eba0bbf Mon Sep 17 00:00:00 2001
From: Vivek Trivedi <5340687+trivedivivek@users.noreply.github.com>
Date: Fri, 30 May 2025 07:56:41 -0700
Subject: [PATCH] [ET-VK] Fixing out_limits_scaled calculation for conv2d pw
 ops.

The fix changing the calculation of `out_limits_scaled` from:
```glsl
const int out_limits_scaled[2] = {out_limits.x + (TILE_SIZE_X - 1) * TILE_SIZE_X, out_limits.y + (TILE_SIZE_Y - 1) * TILE_SIZE_Y};
```
to:
```glsl
const int out_limits_scaled[2] = {(out_limits.x + (TILE_SIZE_X - 1)) / TILE_SIZE_X, (out_limits.y + (TILE_SIZE_Y - 1)) / TILE_SIZE_Y};
```

This change ensures that `out_limits_scaled` is calculated correctly, taking into account the tile size and the output limits of the convolution operation.

Differential Revision: [D75575662](https://our.internmc.facebook.com/intern/diff/D75575662/)

[ghstack-poisoned]
---
 backends/vulkan/runtime/graph/ops/glsl/conv2d_pw.glsl      | 2 +-
 backends/vulkan/runtime/graph/ops/glsl/conv2d_pw_s1p0.glsl | 2 +-
 2 files changed, 2 insertions(+), 2 deletions(-)

diff --git a/backends/vulkan/runtime/graph/ops/glsl/conv2d_pw.glsl b/backends/vulkan/runtime/graph/ops/glsl/conv2d_pw.glsl
index c218b8ac8cc..85c1ea1058e 100644
--- a/backends/vulkan/runtime/graph/ops/glsl/conv2d_pw.glsl
+++ b/backends/vulkan/runtime/graph/ops/glsl/conv2d_pw.glsl
@@ -46,7 +46,7 @@ layout(local_size_x_id = 0, local_size_y_id = 1, local_size_z_id = 2) in;
  * size is only 1x1, making it easier to re-use loaded texels from t_kernel.
  */
 void main() {
-  const int out_limits_scaled[2] = {out_limits.x + (TILE_SIZE_X - 1) * TILE_SIZE_X, out_limits.y + (TILE_SIZE_Y - 1) * TILE_SIZE_Y};
+  const int out_limits_scaled[2] = {(out_limits.x + (TILE_SIZE_X - 1)) / TILE_SIZE_X, (out_limits.y + (TILE_SIZE_Y - 1)) / TILE_SIZE_Y};
 
   const int div_by_x = int(gl_GlobalInvocationID.x / out_limits_scaled[0]);
   const int out_pos[3] = {int(gl_GlobalInvocationID.x % out_limits_scaled[0]), div_by_x, int(gl_GlobalInvocationID.y)};
diff --git a/backends/vulkan/runtime/graph/ops/glsl/conv2d_pw_s1p0.glsl b/backends/vulkan/runtime/graph/ops/glsl/conv2d_pw_s1p0.glsl
index 01dcff00e59..03abd91b074 100644
--- a/backends/vulkan/runtime/graph/ops/glsl/conv2d_pw_s1p0.glsl
+++ b/backends/vulkan/runtime/graph/ops/glsl/conv2d_pw_s1p0.glsl
@@ -48,7 +48,7 @@ layout(local_size_x_id = 0, local_size_y_id = 1, local_size_z_id = 2) in;
  * size is only 1x1, making it easier to re-use loaded texels from t_kernel.
  */
 void main() {
-  const int out_limits_scaled[2] = {out_limits.x + (TILE_SIZE_X - 1) * TILE_SIZE_X, out_limits.y + (TILE_SIZE_Y - 1) * TILE_SIZE_Y};
+  const int out_limits_scaled[2] = {(out_limits.x + (TILE_SIZE_X - 1)) / TILE_SIZE_X, (out_limits.y + (TILE_SIZE_Y - 1)) / TILE_SIZE_Y};
 
   const uint16_t div_by_x = uint16_t(gl_GlobalInvocationID.x / out_limits_scaled[0]);
   const uint16_t out_pos_xy[2] = {uint16_t(gl_GlobalInvocationID.x % out_limits_scaled[0]), div_by_x};