RMS normalization kernel optimization by fusing the reduction kernel and context mapping kernel

llama-tornado

@@ -422,7 +422,7 @@ def create_parser() -> argparse.ArgumentParser:
     )
     debug_group.add_argument(
         "--profiler-dump-dir",
-        default="/home/mikepapadim/repos/gpu-llama3.java/prof.json",
+        default="/home/ruiqi/GPULlama3.java/prof.json",
         help="Directory for profiler output",
     )
 

set_paths

@@ -6,10 +6,10 @@
 
 # Resolve root of this project (LLaMA3) and TornadoVM
 export LLAMA_ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
-export TORNADO_ROOT="${LLAMA_ROOT}/external/tornadovm"
+export TORNADO_ROOT="/home/ruiqi/TornadoVM_OCL/TornadoVM"
 
 # Set the path to TornadoVM SDK binaries
-export TORNADO_SDK="${TORNADO_ROOT}/bin/sdk"
+export TORNADO_SDK="/home/ruiqi/TornadoVM_OCL/TornadoVM/bin/sdk"
 
 # Add TornadoVM and LLaMA bin directories to PATH
 export PATH="${PATH}:${TORNADO_SDK}:${LLAMA_ROOT}"

src/main/java/org/beehive/gpullama3/tornadovm/kernels/TransformerComputeKernelsLayered.java

@@ -35,6 +35,7 @@ public TransformerComputeKernelsLayered() {
      * @param localMemSize
      *         Size of local memory allocation (must match work group size)
      */
+
     public static void reductionOneBlockWithLayer(KernelContext context, FloatArray output, FloatArray x, int size, float ermsNorm, int localMemSize) {
         int gid = context.globalIdx;
         int lid = context.localIdx;
@@ -82,20 +83,92 @@ public static void reductionOneBlockWithLayer(KernelContext context, FloatArray
     }
 
     /**
-     * Applies the computed normalization factor to input and weight elements. This is the second phase of RMS normalization.
+     * Performs RMS (Root Mean Square) normalization using parallel reduction. It first computes the variance and scaling factor across all work groups,
+     * then it applies the computed normalization factor to input and weight elements.
      *
+     * <p>
      * Formula: output[i] = weight[i] * (normalizationFactor * x[i])
      *
+     * Algorithm: 1. Each thread computes square of its input element 2. Work group performs parallel reduction of squares 3. Partial sums stored per work group 4. All thread combines all partial
+     * sums and computes normalization factor 5. Applies the computed normalization factor to input and weight elements.
+     *
      * @param context
      *         Kernel execution context
      * @param output
-     *         Array for normalized output
+     *         Array to store partial sums and final normalization factor
      * @param x
-     *         Input values to normalize
+     *         Input array to normalize
      * @param weights
      *         Weight values for each element
      * @param temp
      *         Temporary array containing normalization factor at index 0
+     * @param size
+     *         Number of elements to process
+     * @param ermsNorm
+     *         Epsilon value squared for numerical stability
+     * @param localMemSize
+     *         Size of local memory allocation (must match work group size)
+     */
+
+    public static void reductionOneBlockWithLayerFuse(KernelContext context, FloatArray output, FloatArray x, FloatArray weights, FloatArray temp, int size, float ermsNorm, int localMemSize) {
+        int gid = context.globalIdx;
+        int lid = context.localIdx;
+        int groupId = context.groupIdx;
+        int groupSize = context.localGroupSizeX;
+
+        // Allocate local memory with the provided size
+        float[] localX = context.allocateFloatLocalArray(localMemSize);
+
+        // Load input value and compute square
+        if (gid < size) {
+            float v = x.get(gid);
+            localX[lid] = v * v;
+        } else {
+            localX[lid] = 0.0f;
+        }
+
+        // Perform parallel reduction within the work group
+        for (int stride = (groupSize / 2); stride > 0; stride /= 2) {
+            context.localBarrier();
+            if (lid < stride) {
+                localX[lid] += localX[lid + stride];
+            }
+        }
+
+        // Each workgroup stores its partial sum in a different location
+        if (lid == 0) {
+            // Store the partial sum from each workgroup
+            temp.set(groupId, localX[0]);
+        }
+
+        context.globalBarrier();
+
+        float localss = 0.0f;
+        int numGroups = (size + groupSize - 1) / groupSize;
+        for (int i = 0; i < numGroups; i++) {  // Assuming 8 workgroups
+            localss += temp.get(i);
+        }
+        localss /= size;
+        localss += ermsNorm;
+        localss = 1.0f / TornadoMath.sqrt(localss);
+
+        if (gid < size) {
+            float in = x.get(gid);
+            float w = weights.get(gid);
+            output.set(gid, w * (localss * in));
+        }
+    }
+
+    /**
+     * Applies the computed normalization factor to input and weight elements. This is the second phase of RMS normalization.
+     * <p>
+     * Formula: output[i] = weight[i] * (normalizationFactor * x[i])
+     *
+     * @param context Kernel execution context
+     * @param output  Array for normalized output
+     * @param x       Input values to normalize
+     * @param weights Weight values for each element
+     * @param temp    Temporary array containing normalization factor at index 0
      */
     public static void reductionOneBlock2WithLayer(KernelContext context, FloatArray output, FloatArray x, FloatArray weights, FloatArray temp) {
         int gid = context.globalIdx;
@@ -106,25 +179,17 @@ public static void reductionOneBlock2WithLayer(KernelContext context, FloatArray
 
     /**
      * Copies keys and values into the key-value cache for attention computation. Enables efficient access to past key-value pairs during autoregressive generation.
-     *
+     * <p>
      * Cache layout: [layer][position][dimension] - Each layer has its own key and value cache - Each position in sequence has a key and value vector
      *
-     * @param destKeyCache
-     *         Destination array for key cache
-     * @param srcKey
-     *         Source keys to copy
-     * @param destValueCache
-     *         Destination array for value cache
-     * @param srcValue
-     *         Source values to copy
-     * @param positioNlayer
-     *         Array containing current position
-     * @param kvDim
-     *         Dimension of key/value vectors
-     * @param layer
-     *         Current transformer layer index
-     * @param contextLength
-     *         Maximum sequence length
+     * @param destKeyCache   Destination array for key cache
+     * @param srcKey         Source keys to copy
+     * @param destValueCache Destination array for value cache
+     * @param srcValue       Source values to copy
+     * @param positioNlayer  Array containing current position
+     * @param kvDim          Dimension of key/value vectors
+     * @param layer          Current transformer layer index
+     * @param contextLength  Maximum sequence length
      */
     public static void copyToCache(FloatArray destKeyCache, FloatArray srcKey, FloatArray destValueCache, FloatArray srcValue, IntArray positioNlayer, int kvDim, int layer, int contextLength) {
 
@@ -160,21 +225,15 @@ public static void splitQKV(FloatArray qkv, FloatArray q, FloatArray k, FloatArr
     /**
      * Applies Rotary Position Encoding (RoPE) to query and key vectors. RoPE rotates pairs of dimensions based on their position in the sequence, enabling the model to learn relative positional
      * information.
-     *
+     * <p>
      * For each pair of dimensions (2*i, 2*i+1): - Compute rotation angle based on position and frequency - Apply 2D rotation to the pair
      *
-     * @param context
-     *         Kernel execution context
-     * @param positionHolder
-     *         Array containing current position
-     * @param sq
-     *         Query vectors to rotate
-     * @param sk
-     *         Key vectors to rotate
-     * @param kv_dim
-     *         Dimension of key/value vectors
-     * @param head_size
-     *         Dimension of each attention head
+     * @param context        Kernel execution context
+     * @param positionHolder Array containing current position
+     * @param sq             Query vectors to rotate
+     * @param sk             Key vectors to rotate
+     * @param kv_dim         Dimension of key/value vectors
+     * @param head_size      Dimension of each attention head
      */
     public static void ropeRotation(KernelContext context, IntArray positionHolder, FloatArray sq, FloatArray sk, int kv_dim, int head_size) {
         int i = context.globalIdx * 2;
@@ -249,31 +308,20 @@ public static void ropeRotationPhi3(KernelContext context, IntArray positionHold
 
     /**
      * Computes attention for a single head. Implements scaled dot-product attention with softmax normalization.
-     *
+     * <p>
      * Steps: 1. Compute attention scores: Q·K / sqrt(head_size) 2. Apply softmax (with max subtraction for numerical stability) 3. Compute weighted sum of values
      *
-     * @param allQ
-     *         All query vectors
-     * @param key_cache
-     *         Cached keys
-     * @param value_cache
-     *         Cached values
-     * @param allXb
-     *         Output buffer
-     * @param h
-     *         Head index to process
-     * @param headSize
-     *         Dimension per head
-     * @param kvDim
-     *         Key/value dimension
-     * @param kvMul
-     *         Key multiplier for grouped attention
-     * @param loff
-     *         Layer offset in cache
-     * @param pos
-     *         Current position
-     * @param wrapAtt
-     *         Attention weights buffer
+     * @param allQ        All query vectors
+     * @param key_cache   Cached keys
+     * @param value_cache Cached values
+     * @param allXb       Output buffer
+     * @param h           Head index to process
+     * @param headSize    Dimension per head
+     * @param kvDim       Key/value dimension
+     * @param kvMul       Key multiplier for grouped attention
+     * @param loff        Layer offset in cache
+     * @param pos         Current position
+     * @param wrapAtt     Attention weights buffer
      */
     private static void processHeadTornado(FloatArray allQ, FloatArray key_cache, FloatArray value_cache, FloatArray allXb, int h, int headSize, int kvDim, int kvMul, long loff, int pos,
             FloatArray wrapAtt) {
@@ -629,23 +677,16 @@ public static void processHeadsFlashAttentionOpt(KernelContext context, FloatArr
 
     /**
      * Performs optimized matrix-vector multiplication where each work group processes one row of the matrix.
-     *
+     * <p>
      * Algorithm: 1. Each work group handles one output dimension 2. Threads in work group compute partial dot products 3. Parallel reduction yields final row result
      *
-     * @param context
-     *         Kernel execution context
-     * @param x
-     *         Input vector
-     * @param hb
-     *         Output vector
-     * @param w
-     *         Weight matrix (row-major)
-     * @param n
-     *         Input dimension
-     * @param d
-     *         Output dimension
-     * @param localWorkGroupSize
-     *         Number of threads per work group
+     * @param context            Kernel execution context
+     * @param x                  Input vector
+     * @param hb                 Output vector
+     * @param w                  Weight matrix (row-major)
+     * @param n                  Input dimension
+     * @param d                  Output dimension
+     * @param localWorkGroupSize Number of threads per work group
      */
     public static void matrixVectorGeneric(KernelContext context, FloatArray x, FloatArray hb, FloatArray w, int n, int d, int localWorkGroupSize) {
         // One row per workgroup (not per thread)

src/main/java/org/beehive/gpullama3/tornadovm/layers/type/fp16/LlamaFP16FFNLayers.java

@@ -31,6 +31,7 @@ public LlamaFP16FFNLayers(String taskGraph, State state, Weights weights, Config
     public GridScheduler updateGridScheduler(GridScheduler tornadoForwardScheduler) {
         WorkerGrid ropeWorker = WorkerGridFactory.genericWorker(config.dim()/2, 128);
         WorkerGrid rmsNormWorker = WorkerGridFactory.createRmsNormWorker(config.dim(), 256);
+        //System.out.println("llama config dim: " + config.dim());
 
         int configDimRowMajorGlobal = config.dim() * LOCAL_WORK_GROUP_SIZE_ALLOC;
         WorkerGrid configDimRowMajorGlobalWorker = WorkerGridFactory.genericWorker(configDimRowMajorGlobal, LOCAL_WORK_GROUP_SIZE_ALLOC);
@@ -54,9 +55,7 @@ public GridScheduler updateGridScheduler(GridScheduler tornadoForwardScheduler)
             tornadoForwardScheduler.addWorkerGrid("layer_" + i + ".projectionTwo", configDimRowMajorGlobalWorker);
             tornadoForwardScheduler.addWorkerGrid("layer_" + i + ".fused_ffn_w1_w3", configHiddenDimRowMajorWorker);
             tornadoForwardScheduler.addWorkerGrid("layer_" + i + ".reductionsOneBlock", rmsNormWorker);
-            tornadoForwardScheduler.addWorkerGrid("layer_" + i + ".mapContext", rmsNormWorker);
             tornadoForwardScheduler.addWorkerGrid("layer_" + i + ".reductionsOneBlockFFN", rmsNormWorker);
-            tornadoForwardScheduler.addWorkerGrid("layer_" + i + ".mapContextFFN", rmsNormWorker);
             tornadoForwardScheduler.addWorkerGrid("layer_" + i + ".parallel-attention", parallelAttentionWorker);
             tornadoForwardScheduler.addWorkerGrid("layer_" + i + ".copyToCaches", copyToCachesWorker);
         }
@@ -112,13 +111,8 @@ TaskGraph setupSingleFFNLayer(LlamaTornadoWeights weights, Configuration config,
                 weights.w3Layered[layerIndex].asHalfFloatArray());
         unifiedLayer = configureLayerDataTransfers(unifiedLayer, layerIndex);
         unifiedLayer
-                .task("reductionsOneBlock", TransformerComputeKernelsLayered::reductionOneBlockWithLayer, context, state.temp, state.wrapX, config.dim(), config.rmsNormEps(), state.localSize);
-                if (shouldUseFinalNormalization()) {
-                    unifiedLayer.task("reductionFinalNormalization", TransformerComputeKernelsLayered::reductionFinalNormalization, context, state.temp,
-                            config.dim(), config.rmsNormEps());
-                }
-                unifiedLayer.task("mapContext", TransformerComputeKernelsLayered::reductionOneBlock2WithLayer, context, state.wrapXb, state.wrapX, weights.rms_att_weightLayered[layerIndex].asFloatArray(), state.temp)
-                .task("qmatmul", TransformerComputeKernelsLayered::matrixVectorGeneric, context, state.wrapXb, state.wrapQ, weights.wqLayered[layerIndex].asHalfFloatArray(), config.dim(), config.dim(),
+                .task("reductionsOneBlock", TransformerComputeKernelsLayered::reductionOneBlockWithLayerFuse, context, state.wrapXb, state.wrapX, weights.rms_att_weightLayered[layerIndex].asFloatArray(), state.temp, config.dim(), config.rmsNormEps(), state.localSize);
+                unifiedLayer.task("qmatmul", TransformerComputeKernelsLayered::matrixVectorGeneric, context, state.wrapXb, state.wrapQ, weights.wqLayered[layerIndex].asHalfFloatArray(), config.dim(), config.dim(),
                         LOCAL_WORK_GROUP_SIZE_ALLOC)
                 .task("kmatmul", TransformerComputeKernelsLayered::matrixVectorGeneric, context, state.wrapXb, state.wrapK, weights.wkLayered[layerIndex].asHalfFloatArray(), config.dim(), config.kvDim(),
                         LOCAL_WORK_GROUP_SIZE_ALLOC)
@@ -130,12 +124,8 @@ TaskGraph setupSingleFFNLayer(LlamaTornadoWeights weights, Configuration config,
                 configureAttention(unifiedLayer, layerIndex);
                 unifiedLayer.task("matmul1", TransformerComputeKernelsLayered::matrixVectorGenericWithResidual, context, state.wrapXb, state.wrapX, weights.woLayered[layerIndex].asHalfFloatArray(), config.dim(), config.dim(),
                         LOCAL_WORK_GROUP_SIZE_ALLOC)
-                .task("reductionsOneBlockFFN", TransformerComputeKernelsLayered::reductionOneBlockWithLayer, context, state.tempFFN, state.wrapX, config.dim(), config.rmsNormEps(), state.localSize);
-                if (shouldUseFinalNormalization()) {
-                    unifiedLayer.task("reductionFinalNormalizationFFN", TransformerComputeKernelsLayered::reductionFinalNormalization, context, state.tempFFN, config.dim(), config.rmsNormEps());
-                }
-                unifiedLayer.task("mapContextFFN", TransformerComputeKernelsLayered::reductionOneBlock2WithLayer, context, state.wrapXb, state.wrapX, weights.rms_ffn_weightLayered[layerIndex].asFloatArray(), state.tempFFN)
-                .task("fused_ffn_w1_w3", TransformerComputeKernelsLayered::fusedFeedForwardWithSiLUAndGLUActivation, context, state.wrapXb, state.wrapHb, weights.w1Layered[layerIndex].asHalfFloatArray(),
+                .task("reductionsOneBlockFFN", TransformerComputeKernelsLayered::reductionOneBlockWithLayerFuse, context, state.wrapXb, state.wrapX, weights.rms_ffn_weightLayered[layerIndex].asFloatArray(), state.tempFFN, config.dim(), config.rmsNormEps(), state.localSize);
+        unifiedLayer.task("fused_ffn_w1_w3", TransformerComputeKernelsLayered::fusedFeedForwardWithSiLUAndGLUActivation, context, state.wrapXb, state.wrapHb, weights.w1Layered[layerIndex].asHalfFloatArray(),
                         weights.w3Layered[layerIndex].asHalfFloatArray(), config.dim(), config.hiddenDim(), LOCAL_WORK_GROUP_SIZE_ALLOC)
                 .task("projectionTwo", TransformerComputeKernelsLayered::matrixVectorGenericWithResidual, context, state.wrapHb, state.wrapX, weights.w2Layered[layerIndex].asHalfFloatArray(), config.hiddenDim(),
                         config.dim(), LOCAL_WORK_GROUP_SIZE_ALLOC).persistOnDevice(state.wrapX);

src/main/java/org/beehive/gpullama3/tornadovm/layers/type/fp16/Phi3FP16FFNLayers.java

@@ -63,7 +63,6 @@ public Phi3FP16FFNLayers(String taskGraphName, Phi3State state, Phi3TornadoWeigh
     public GridScheduler updateGridScheduler(GridScheduler gridScheduler) {
         // RMS norm worker
         WorkerGrid rmsNormWorker = WorkerGridFactory.createRmsNormWorker(config.dim(), state.localSize);
-
         // Combined QKV matmul worker
         int matmulQkvGlobal = opSize * LOCAL_WORK_GROUP_SIZE_ALLOC;
         WorkerGrid matmulQkvRowMajorWorker = WorkerGridFactory.genericWorker(matmulQkvGlobal, LOCAL_WORK_GROUP_SIZE_ALLOC);