[ET-VK] Removing tile input storage variable in conv_pw op and fetchi…

…ng the data in main loop. Also unrolling the main loop for performance improvement.

Differential Revision: D64767314

Pull Request resolved: #6765

backends/vulkan/runtime/graph/ops/glsl/conv2d_pw.glsl

@@ -82,17 +82,15 @@ void main() {
     // During prepacking, the weight tensor has been permuted so that the
     // channel (IC) dim is along the x-axis, and the batch (OC) dim is along
     // the z-axis.
-    vec4 in_tex[TILE_SIZE * TILE_SIZE];
     const vec4 ktex_0 = texelFetch(t_kernel, u16vec2(z + 0, gpos.z), 0);
     const vec4 ktex_1 = texelFetch(t_kernel, u16vec2(z + 1, gpos.z), 0);
     const vec4 ktex_2 = texelFetch(t_kernel, u16vec2(z + 2, gpos.z), 0);
     const vec4 ktex_3 = texelFetch(t_kernel, u16vec2(z + 3, gpos.z), 0);
 
-    for (int i = 0; i < TILE_SIZE * TILE_SIZE; ++i) {
-      in_tex[i] = texelFetch(t_in, u16vec3(ipos[i], z4), 0);
-    }
 
+#pragma unroll
     for (int i = 0; i < TILE_SIZE * TILE_SIZE; ++i) {
+      const vec4 in_tex = texelFetch(t_in, u16vec3(ipos[i], z4), 0);
       // For 2x2 tile size algorithm works as follows.
       // To explain the calculations below, the contents of one in_tex and the
       // group of 4 texels loaded from t_kernel are shown:
@@ -126,10 +124,10 @@ void main() {
       //
       //  which is what is expressed in the following calculations. This is done
       //  for each output position.
-      sum[i] = fma(in_tex[i].xxxx, ktex_0, sum[i]);
-      sum[i] = fma(in_tex[i].yyyy, ktex_1, sum[i]);
-      sum[i] = fma(in_tex[i].zzzz, ktex_2, sum[i]);
-      sum[i] = fma(in_tex[i].wwww, ktex_3, sum[i]);
+      sum[i] = fma(in_tex.xxxx, ktex_0, sum[i]);
+      sum[i] = fma(in_tex.yyyy, ktex_1, sum[i]);
+      sum[i] = fma(in_tex.zzzz, ktex_2, sum[i]);
+      sum[i] = fma(in_tex.wwww, ktex_3, sum[i]);
     }
   }