Removed unnecessary shapes/strides in the shader.

Author: satyajandhyala

js/web/lib/wasm/jsep/webgpu/ops/skip-layer-norm.ts

@@ -7,7 +7,7 @@ import {ShapeUtil} from '../../util';
 import {AttributeWithCacheKey} from '../attribute-with-cache-key';
 import {ComputeContext, ProgramInfo, ProgramUniform} from '../types';
 
-import {castToF32, createTensorShapeVariables, fillVector, getMaxComponents, inputVariable, outputVariable, ShaderHelper, sumVector, tensorTypeToWsglStorageType, UniformsArrayType} from './common';
+import {castToF32, fillVector, getMaxComponents, inputVariable, outputVariable, ShaderHelper, sumVector, tensorTypeToWsglStorageType, UniformsArrayType} from './common';
 
 export interface SkipLayerNormAttributes extends AttributeWithCacheKey {
   epsilon: number;
@@ -98,36 +98,30 @@ const createSkipLayerNormProgramInfo =
             {type: 'uint32', data: hiddenSize},
             {type: 'float32', data: attributes.epsilon},
           ];
-          inputs.forEach((input, _) => {
-            programUniforms.push(...createTensorShapeVariables(input.dims));
-          });
-          const variables = [
-            inputVariable('x', inputs[0].dataType, inputs[0].dims.length, components),
-            inputVariable('skip', inputs[1].dataType, inputs[1].dims.length, components),
-            inputVariable('gamma', inputs[2].dataType, inputs[2].dims.length, components),
-          ];
-          if (hasBetaInput) {
-            variables.push(inputVariable('beta', inputs[3].dataType, inputs[3].dims.length, components));
-          }
-          if (hasBiasInput) {
-            variables.push(inputVariable('bias', inputs[4].dataType, inputs[4].dims.length, components));
-          }
-          variables.push(outputVariable('output', inputs[0].dataType, outputShape.length, components));
-          programUniforms.push(...createTensorShapeVariables(outputShape));
-          if (hasMeanOutput) {
-            variables.push(outputVariable('mean_output', DataType.float, meanInvStdDevDim.length));
-            programUniforms.push(...createTensorShapeVariables(meanInvStdDevDim));
-          }
-          if (hasInvStdDevOutput) {
-            variables.push(outputVariable('inv_std_output', DataType.float, meanInvStdDevDim.length));
-            programUniforms.push(...createTensorShapeVariables(meanInvStdDevDim));
-          }
-          if (hasInputSkipBiasSumOutput) {
-            variables.push(outputVariable('input_skip_bias_sum', inputs[0].dataType, outputShape.length, components));
-            programUniforms.push(...createTensorShapeVariables(outputShape));
-          }
-          const dataType = tensorTypeToWsglStorageType(inputs[0].dataType);
-          const getShaderSource = (shaderHelper: ShaderHelper) => `
+          const getShaderSource = (shaderHelper: ShaderHelper) => {
+            const variables = [
+              inputVariable('x', inputs[0].dataType, inputs[0].dims, components),
+              inputVariable('skip', inputs[1].dataType, inputs[1].dims, components),
+              inputVariable('gamma', inputs[2].dataType, inputs[2].dims, components),
+            ];
+            if (hasBetaInput) {
+              variables.push(inputVariable('beta', inputs[3].dataType, inputs[3].dims, components));
+            }
+            if (hasBiasInput) {
+              variables.push(inputVariable('bias', inputs[4].dataType, inputs[4].dims, components));
+            }
+            variables.push(outputVariable('output', inputs[0].dataType, outputShape, components));
+            if (hasMeanOutput) {
+              variables.push(outputVariable('mean_output', DataType.float, meanInvStdDevDim));
+            }
+            if (hasInvStdDevOutput) {
+              variables.push(outputVariable('inv_std_output', DataType.float, meanInvStdDevDim));
+            }
+            if (hasInputSkipBiasSumOutput) {
+              variables.push(outputVariable('input_skip_bias_sum', inputs[0].dataType, outputShape, components));
+            }
+            const dataType = tensorTypeToWsglStorageType(inputs[0].dataType);
+            return `
       const epsilon: f32 = ${attributes.epsilon};
 
       ${shaderHelper.registerUniforms(uniformsArray).declareVariables(...variables)}
@@ -151,14 +145,15 @@ const createSkipLayerNormProgramInfo =
         }
         let mean = ${sumVector('sum', components)} / f32(uniforms.hidden_size);
         let inv_std_dev = inverseSqrt(${
-              sumVector('squareSum', components)} / f32(uniforms.hidden_size) - mean * mean + uniforms.epsilon);
+                sumVector('squareSum', components)} / f32(uniforms.hidden_size) - mean * mean + uniforms.epsilon);
         ${hasMeanOutput ? 'mean_output[global_idx] = mean;' : ''}
         ${hasInvStdDevOutput ? 'inv_std_output[global_idx] = inv_std_dev;' : ''}
         for (var i: u32 = 0; i < hidden_size_vectorized; i++) {
           output[offset + i] = (output[offset + i] - ${dataType}(mean)) * ${dataType}(inv_std_dev) * gamma[i] + ${
-              hasBetaInput ? 'beta[i]' : '0.0'};
+                hasBetaInput ? 'beta[i]' : '0.0'};
         }
       }`;
+          };
           const outputs = [{dims: outputShape, dataType: inputs[0].dataType}];
           if (outputCount > 1) {
             outputs.push({dims: meanInvStdDevDim, dataType: DataType.float});
@@ -173,7 +168,7 @@ const createSkipLayerNormProgramInfo =
             name: 'SkipLayerNormalization',
             shaderCache: {
               hint: `${components};${hasMeanOutput};${hasInvStdDevOutput};${hasInputSkipBiasSumOutput}`,
-              inputDependencies: inputs.map((_input, _index) => 'rank')
+              inputDependencies: inputs.map((_input, _index) => 'type')
             },
             getShaderSource,
             getRunData: () => ({outputs, dispatchGroup: {x: Math.ceil(outputSize / hiddenSize / 64)}, programUniforms}),