llama-model-loader: print warning when using overrides with mmap (#20978)

* llama-model-loader: use pinned memory for tensor overrides

* change to warning

.devops/intel.Dockerfile

@@ -33,6 +33,23 @@ RUN mkdir -p /app/full \
 
 FROM intel/deep-learning-essentials:$ONEAPI_VERSION AS base
 
+ARG IGC_VERSION=v2.30.1
+ARG IGC_VERSION_FULL=2_2.30.1+20950
+ARG COMPUTE_RUNTIME_VERSION=26.09.37435.1
+ARG COMPUTE_RUNTIME_VERSION_FULL=26.09.37435.1-0
+ARG IGDGMM_VERSION=22.9.0
+RUN mkdir /tmp/neo/ && cd /tmp/neo/ \
+  && wget https://github.com/intel/intel-graphics-compiler/releases/download/$IGC_VERSION/intel-igc-core-${IGC_VERSION_FULL}_amd64.deb \
+  && wget https://github.com/intel/intel-graphics-compiler/releases/download/$IGC_VERSION/intel-igc-opencl-${IGC_VERSION_FULL}_amd64.deb \
+  && wget https://github.com/intel/compute-runtime/releases/download/$COMPUTE_RUNTIME_VERSION/intel-ocloc-dbgsym_${COMPUTE_RUNTIME_VERSION_FULL}_amd64.ddeb \
+  && wget https://github.com/intel/compute-runtime/releases/download/$COMPUTE_RUNTIME_VERSION/intel-ocloc_${COMPUTE_RUNTIME_VERSION_FULL}_amd64.deb \
+  && wget https://github.com/intel/compute-runtime/releases/download/$COMPUTE_RUNTIME_VERSION/intel-opencl-icd-dbgsym_${COMPUTE_RUNTIME_VERSION_FULL}_amd64.ddeb \
+  && wget https://github.com/intel/compute-runtime/releases/download/$COMPUTE_RUNTIME_VERSION/intel-opencl-icd_${COMPUTE_RUNTIME_VERSION_FULL}_amd64.deb \
+  && wget https://github.com/intel/compute-runtime/releases/download/$COMPUTE_RUNTIME_VERSION/libigdgmm12_${IGDGMM_VERSION}_amd64.deb \
+  && wget https://github.com/intel/compute-runtime/releases/download/$COMPUTE_RUNTIME_VERSION/libze-intel-gpu1-dbgsym_${COMPUTE_RUNTIME_VERSION_FULL}_amd64.ddeb \
+  && wget https://github.com/intel/compute-runtime/releases/download/$COMPUTE_RUNTIME_VERSION/libze-intel-gpu1_${COMPUTE_RUNTIME_VERSION_FULL}_amd64.deb \
+  && dpkg --install *.deb
+
 RUN apt-get update \
     && apt-get install -y libgomp1 curl\
     && apt autoremove -y \

.github/workflows/python-type-check.yml

@@ -31,7 +31,7 @@ jobs:
         uses: actions/setup-python@v6
         with:
           python-version: "3.11"
-          pip-install: -r requirements/requirements-all.txt ty==0.0.24
+          pip-install: -r requirements/requirements-all.txt ty==0.0.26
       # - name: Type-check with Pyright
       #   uses: jakebailey/pyright-action@v2
       #   with:

convert_hf_to_gguf.py

@@ -31,10 +31,10 @@ import gguf
 from gguf.vocab import MistralTokenizerType, MistralVocab
 
 try:
-    from mistral_common.tokens.tokenizers.base import TokenizerVersion # type: ignore[import-not-found]
-    from mistral_common.tokens.tokenizers.multimodal import DATASET_MEAN as _MISTRAL_COMMON_DATASET_MEAN, DATASET_STD as _MISTRAL_COMMON_DATASET_STD # type: ignore[import-not-found]
-    from mistral_common.tokens.tokenizers.tekken import Tekkenizer # type: ignore[import-not-found]
-    from mistral_common.tokens.tokenizers.sentencepiece import ( # type: ignore[import-not-found]
+    from mistral_common.tokens.tokenizers.base import TokenizerVersion # type: ignore[import-not-found, ty:unresolved-import]
+    from mistral_common.tokens.tokenizers.multimodal import DATASET_MEAN as _MISTRAL_COMMON_DATASET_MEAN, DATASET_STD as _MISTRAL_COMMON_DATASET_STD # type: ignore[import-not-found, ty:unresolved-import]
+    from mistral_common.tokens.tokenizers.tekken import Tekkenizer # type: ignore[import-not-found, ty:unresolved-import]
+    from mistral_common.tokens.tokenizers.sentencepiece import ( # type: ignore[import-not-found, ty:unresolved-import]
         SentencePieceTokenizer,
     )
 

examples/model-conversion/scripts/causal/compare-logits.py

@@ -7,7 +7,7 @@ import os
 
 # Add utils directory to path for direct script execution
 sys.path.insert(0, str(Path(__file__).parent.parent / "utils"))
-from common import get_model_name_from_env_path, compare_tokens, exit_with_warning  # type: ignore[import-not-found]
+from common import get_model_name_from_env_path, compare_tokens, exit_with_warning  # type: ignore[import-not-found, ty:unresolved-import]
 
 def quick_logits_check(pytorch_file, llamacpp_file):
     """Lightweight sanity check before NMSE"""

examples/model-conversion/scripts/utils/check-nmse.py

@@ -5,7 +5,7 @@ import sys
 import os
 import argparse
 from pathlib import Path
-from common import get_model_name_from_env_path  # type: ignore[import-not-found]
+from common import get_model_name_from_env_path  # type: ignore[import-not-found, ty:unresolved-import]
 
 def calculate_nmse(reference, test):
     mse = np.mean((test - reference) ** 2)

examples/model-conversion/scripts/utils/compare_tokens.py

@@ -2,7 +2,7 @@
 
 import argparse
 import sys
-from common import compare_tokens  # type: ignore[import-not-found]
+from common import compare_tokens  # type: ignore[import-not-found, ty:unresolved-import]
 
 
 def parse_arguments():

examples/model-conversion/scripts/utils/semantic_check.py

@@ -7,7 +7,7 @@ import importlib
 from pathlib import Path
 
 from transformers import AutoTokenizer, AutoConfig, AutoModelForCausalLM, AutoModel
-from common import compare_tokens, exit_with_warning  # type: ignore[import-not-found]
+from common import compare_tokens, exit_with_warning  # type: ignore[import-not-found, ty:unresolved-import]
 
 unreleased_model_name = os.getenv('UNRELEASED_MODEL_NAME')
 

examples/sycl/build.sh

@@ -20,4 +20,4 @@ cmake .. -DGGML_SYCL=ON -DCMAKE_C_COMPILER=icx -DCMAKE_CXX_COMPILER=icpx -DLLAMA
 #cmake --build . --config Release --target llama-bench
 
 #build all binary
-cmake --build . --config Release -j -v
+cmake --build . --config Release -j$((($(nproc)+1)/2)) -v

examples/sycl/run-llama2.sh

@@ -23,9 +23,9 @@ if [ $# -gt 0 ]; then
     GGML_SYCL_DEVICE=$1
     echo "use $GGML_SYCL_DEVICE as main GPU"
     #use signle GPU only
-    ZES_ENABLE_SYSMAN=1 ./build/bin/llama-completion -m ${MODEL_FILE} -no-cnv -p "${INPUT_PROMPT}" -n 400 -e -ngl ${NGL} -s 0 -c ${CONTEXT} -mg $GGML_SYCL_DEVICE -sm none ${LOAD_MODE}
+    ZES_ENABLE_SYSMAN=1 ./build/bin/llama-completion -m ${MODEL_FILE} -no-cnv -p "${INPUT_PROMPT}" -n 200 -e -ngl ${NGL} -s 0 -c ${CONTEXT} -mg $GGML_SYCL_DEVICE -sm none ${LOAD_MODE}
 
 else
     #use multiple GPUs with same max compute units
-    ZES_ENABLE_SYSMAN=1 ./build/bin/llama-completion -m ${MODEL_FILE} -no-cnv -p "${INPUT_PROMPT}" -n 400 -e -ngl ${NGL} -s 0 -c ${CONTEXT} ${LOAD_MODE}
+    ZES_ENABLE_SYSMAN=1 ./build/bin/llama-completion -m ${MODEL_FILE} -no-cnv -p "${INPUT_PROMPT}" -n 200 -e -ngl ${NGL} -s 0 -c ${CONTEXT} ${LOAD_MODE}
 fi

ggml/src/ggml-cuda/ggml-cuda.cu

@@ -2343,7 +2343,8 @@ static void ggml_cuda_mul_mat_id(ggml_backend_cuda_context & ctx, ggml_tensor *
         static_assert(MMVQ_MAX_BATCH_SIZE == MMVF_MAX_BATCH_SIZE);
         if (ne2 <= MMVQ_MAX_BATCH_SIZE) {
             if (ggml_is_quantized(src0->type)) {
-                if (ne2 <= MMVQ_MMID_MAX_BATCH_SIZE) {
+                const int mmvq_mmid_max = get_mmvq_mmid_max_batch(src0->type, cc);
+                if (ne2 <= mmvq_mmid_max) {
                     ggml_cuda_mul_mat_vec_q(ctx, src0, src1, ids, dst);
                     return;
                 }
@@ -2946,14 +2947,18 @@ static bool ggml_cuda_graph_check_compability(ggml_cgraph * cgraph) {
         }
 
         // [TAG_MUL_MAT_ID_CUDA_GRAPHS]
-        if (node->op == GGML_OP_MUL_MAT_ID && (!ggml_is_quantized(node->src[0]->type) || node->ne[2] > MMVQ_MMID_MAX_BATCH_SIZE)) {
-            // under these conditions, the mul_mat_id operation will need to synchronize the stream, so we cannot use CUDA graphs
-            // TODO: figure out a way to enable for larger batch sizes, without hurting performance
-            // ref: https://github.com/ggml-org/llama.cpp/pull/18958
-            use_cuda_graph = false;
+        if (node->op == GGML_OP_MUL_MAT_ID) {
+            const int cc = ggml_cuda_info().devices[ggml_cuda_get_device()].cc;
+            const int mmvq_mmid_max = get_mmvq_mmid_max_batch(node->src[0]->type, cc);
+            if (!ggml_is_quantized(node->src[0]->type) || node->ne[2] > mmvq_mmid_max) {
+                // under these conditions, the mul_mat_id operation will need to synchronize the stream, so we cannot use CUDA graphs
+                // TODO: figure out a way to enable for larger batch sizes, without hurting performance
+                // ref: https://github.com/ggml-org/llama.cpp/pull/18958
+                use_cuda_graph = false;
 #ifndef NDEBUG
-            GGML_LOG_DEBUG("%s: disabling CUDA graphs due to unsupported node type\n", __func__);
+                GGML_LOG_DEBUG("%s: disabling CUDA graphs due to unsupported node type\n", __func__);
 #endif
+            }
         }
 
         if (!use_cuda_graph) {

ggml/src/ggml-cuda/mmvq.cu

@@ -97,6 +97,194 @@ static __host__ mmvq_parameter_table_id get_device_table_id(int cc) {
     return MMVQ_PARAMETERS_GENERIC;
 }
 
+// Per-architecture maximum batch size for which MMVQ should be used for MUL_MAT_ID.
+// Returns a value <= MMVQ_MAX_BATCH_SIZE. Default is MMVQ_MAX_BATCH_SIZE.
+// Check https://github.com/ggml-org/llama.cpp/pull/20905#issuecomment-4145835627 for details
+
+static constexpr __host__ __device__ int get_mmvq_mmid_max_batch_pascal_older(ggml_type type) {
+    switch (type) {
+        case GGML_TYPE_IQ1_S:   return 6;
+        case GGML_TYPE_IQ1_M:   return 6;
+        case GGML_TYPE_IQ2_S:   return 4;
+        case GGML_TYPE_IQ2_XS:  return 5;
+        case GGML_TYPE_IQ2_XXS: return 5;
+        case GGML_TYPE_IQ3_S:   return 4;
+        case GGML_TYPE_IQ3_XXS: return 4;
+        case GGML_TYPE_IQ4_NL:  return 6;
+        case GGML_TYPE_IQ4_XS:  return 5;
+        case GGML_TYPE_MXFP4:   return 4;
+        case GGML_TYPE_Q2_K:    return 4;
+        case GGML_TYPE_Q3_K:    return 4;
+        case GGML_TYPE_Q4_0:    return 6;
+        case GGML_TYPE_Q4_1:    return 6;
+        case GGML_TYPE_Q4_K:    return 5;
+        case GGML_TYPE_Q5_0:    return 6;
+        case GGML_TYPE_Q5_1:    return 6;
+        case GGML_TYPE_Q5_K:    return 5;
+        case GGML_TYPE_Q6_K:    return 4;
+        case GGML_TYPE_Q8_0:    return 4;
+        default:                return MMVQ_MAX_BATCH_SIZE;
+    }
+}
+
+static constexpr __host__ __device__ int get_mmvq_mmid_max_batch_turing_plus(ggml_type type) {
+    switch (type) {
+        case GGML_TYPE_IQ2_S:   return 7;
+        case GGML_TYPE_IQ3_S:   return 6;
+        case GGML_TYPE_IQ3_XXS: return 7;
+        case GGML_TYPE_MXFP4:   return 7;
+        case GGML_TYPE_Q2_K:    return 7;
+        case GGML_TYPE_Q3_K:    return 5;
+        default:                return MMVQ_MAX_BATCH_SIZE;
+    }
+}
+
+static constexpr __host__ __device__ int get_mmvq_mmid_max_batch_gcn(ggml_type type) {
+    switch (type) {
+        case GGML_TYPE_IQ1_S:   return 5;
+        case GGML_TYPE_IQ1_M:   return 5;
+        case GGML_TYPE_IQ2_S:   return 4;
+        case GGML_TYPE_IQ2_XS:  return 4;
+        case GGML_TYPE_IQ2_XXS: return 4;
+        case GGML_TYPE_IQ3_S:   return 4;
+        case GGML_TYPE_IQ3_XXS: return 4;
+        case GGML_TYPE_IQ4_NL:  return 6;
+        case GGML_TYPE_IQ4_XS:  return 4;
+        case GGML_TYPE_Q2_K:    return 4;
+        case GGML_TYPE_Q3_K:    return 4;
+        case GGML_TYPE_Q4_0:    return 5;
+        case GGML_TYPE_Q4_1:    return 5;
+        case GGML_TYPE_Q4_K:    return 4;
+        case GGML_TYPE_Q5_K:    return 4;
+        case GGML_TYPE_Q6_K:    return 4;
+        case GGML_TYPE_Q8_0:    return 4;
+        default:                return MMVQ_MAX_BATCH_SIZE;
+    }
+}
+
+static constexpr __host__ __device__ int get_mmvq_mmid_max_batch_cdna(ggml_type type) {
+    switch (type) {
+        case GGML_TYPE_IQ2_S:   return 5;
+        case GGML_TYPE_IQ2_XS:  return 5;
+        case GGML_TYPE_IQ2_XXS: return 5;
+        case GGML_TYPE_IQ3_S:   return 4;
+        case GGML_TYPE_IQ3_XXS: return 5;
+        default:                return MMVQ_MAX_BATCH_SIZE;
+    }
+}
+
+static constexpr __host__ __device__ int get_mmvq_mmid_max_batch_rdna1_rdna2(ggml_type type) {
+    switch (type) {
+        case GGML_TYPE_IQ2_S:   return 4;
+        case GGML_TYPE_IQ2_XS:  return 4;
+        case GGML_TYPE_IQ2_XXS: return 4;
+        case GGML_TYPE_IQ3_S:   return 4;
+        case GGML_TYPE_IQ3_XXS: return 4;
+        case GGML_TYPE_Q2_K:    return 7;
+        case GGML_TYPE_Q3_K:    return 4;
+        case GGML_TYPE_Q4_K:    return 5;
+        case GGML_TYPE_Q5_K:    return 6;
+        case GGML_TYPE_Q6_K:    return 5;
+        default:                return MMVQ_MAX_BATCH_SIZE;
+    }
+}
+
+static constexpr __host__ __device__ int get_mmvq_mmid_max_batch_rdna3(ggml_type type) {
+    switch (type) {
+        case GGML_TYPE_IQ1_S:   return 6;
+        case GGML_TYPE_IQ1_M:   return 6;
+        case GGML_TYPE_IQ2_S:   return 4;
+        case GGML_TYPE_IQ2_XS:  return 4;
+        case GGML_TYPE_IQ2_XXS: return 4;
+        case GGML_TYPE_IQ3_S:   return 4;
+        case GGML_TYPE_IQ3_XXS: return 4;
+        case GGML_TYPE_IQ4_NL:  return 6;
+        case GGML_TYPE_IQ4_XS:  return 6;
+        case GGML_TYPE_Q4_K:    return 4;
+        case GGML_TYPE_Q5_K:    return 4;
+        case GGML_TYPE_Q6_K:    return 4;
+        default:                return MMVQ_MAX_BATCH_SIZE;
+    }
+}
+
+static constexpr __host__ __device__ int get_mmvq_mmid_max_batch_rdna4(ggml_type type) {
+    switch (type) {
+        case GGML_TYPE_IQ1_S:   return 7;
+        case GGML_TYPE_IQ1_M:   return 7;
+        case GGML_TYPE_IQ2_S:   return 4;
+        case GGML_TYPE_IQ2_XS:  return 4;
+        case GGML_TYPE_IQ2_XXS: return 4;
+        case GGML_TYPE_IQ3_S:   return 4;
+        case GGML_TYPE_IQ3_XXS: return 4;
+        case GGML_TYPE_IQ4_NL:  return 7;
+        case GGML_TYPE_IQ4_XS:  return 5;
+        case GGML_TYPE_MXFP4:   return 5;
+        case GGML_TYPE_Q3_K:    return 4;
+        case GGML_TYPE_Q4_0:    return 7;
+        case GGML_TYPE_Q4_1:    return 7;
+        case GGML_TYPE_Q4_K:    return 4;
+        case GGML_TYPE_Q5_0:    return 7;
+        case GGML_TYPE_Q5_1:    return 7;
+        case GGML_TYPE_Q5_K:    return 5;
+        case GGML_TYPE_Q6_K:    return 5;
+        case GGML_TYPE_Q8_0:    return 7;
+        default:                return MMVQ_MAX_BATCH_SIZE;
+    }
+}
+
+// Host function: returns the max batch size for the current arch+type at runtime.
+int get_mmvq_mmid_max_batch(ggml_type type, int cc) {
+    // NVIDIA: Volta, Ada Lovelace, and Blackwell always use MMVQ for MUL_MAT_ID.
+    if (cc == GGML_CUDA_CC_VOLTA || cc >= GGML_CUDA_CC_ADA_LOVELACE) {
+        return MMVQ_MAX_BATCH_SIZE;
+    }
+    if (cc >= GGML_CUDA_CC_TURING) {
+        return get_mmvq_mmid_max_batch_turing_plus(type);
+    }
+    if (GGML_CUDA_CC_IS_NVIDIA(cc)) {
+        return get_mmvq_mmid_max_batch_pascal_older(type);
+    }
+    // AMD
+    if (GGML_CUDA_CC_IS_RDNA4(cc)) {
+        return get_mmvq_mmid_max_batch_rdna4(type);
+    }
+    if (GGML_CUDA_CC_IS_RDNA3(cc)) {
+        return get_mmvq_mmid_max_batch_rdna3(type);
+    }
+    if (GGML_CUDA_CC_IS_RDNA1(cc) || GGML_CUDA_CC_IS_RDNA2(cc)) {
+        return get_mmvq_mmid_max_batch_rdna1_rdna2(type);
+    }
+    if (GGML_CUDA_CC_IS_CDNA(cc)) {
+        return get_mmvq_mmid_max_batch_cdna(type);
+    }
+    if (GGML_CUDA_CC_IS_GCN(cc)) {
+        return get_mmvq_mmid_max_batch_gcn(type);
+    }
+    return MMVQ_MAX_BATCH_SIZE;
+}
+
+// Device constexpr: returns the max batch size for the current arch+type at compile time.
+template <ggml_type type>
+static constexpr __device__ int get_mmvq_mmid_max_batch_for_device() {
+#if defined(RDNA4)
+    return get_mmvq_mmid_max_batch_rdna4(type);
+#elif defined(RDNA3)
+    return get_mmvq_mmid_max_batch_rdna3(type);
+#elif defined(RDNA2) || defined(RDNA1)
+    return get_mmvq_mmid_max_batch_rdna1_rdna2(type);
+#elif defined(CDNA)
+    return get_mmvq_mmid_max_batch_cdna(type);
+#elif defined(GCN)
+    return get_mmvq_mmid_max_batch_gcn(type);
+#elif defined(__CUDA_ARCH__) && (__CUDA_ARCH__ == GGML_CUDA_CC_VOLTA || __CUDA_ARCH__ >= GGML_CUDA_CC_ADA_LOVELACE)
+    return MMVQ_MAX_BATCH_SIZE;
+#elif defined(__CUDA_ARCH__) && __CUDA_ARCH__ >= GGML_CUDA_CC_TURING
+    return get_mmvq_mmid_max_batch_turing_plus(type);
+#else
+    return get_mmvq_mmid_max_batch_pascal_older(type);
+#endif
+}
+
 static constexpr __host__ __device__ int calc_nwarps(ggml_type type, int ncols_dst, mmvq_parameter_table_id table_id) {
     if (table_id == MMVQ_PARAMETERS_GENERIC) {
         switch (ncols_dst) {
@@ -195,7 +383,7 @@ static constexpr __host__ __device__ int calc_rows_per_block(int ncols_dst, int
     return 1;
 }
 
-template <ggml_type type, int ncols_dst, bool has_fusion, bool is_multi_token_id = false, bool small_k = false>
+template <ggml_type type, int ncols_dst, bool has_fusion, bool small_k = false>
 __launch_bounds__(calc_nwarps(type, ncols_dst, get_device_table_id())*ggml_cuda_get_physical_warp_size(), 1)
 static __global__ void mul_mat_vec_q(
         const void * __restrict__ vx, const void * __restrict__ vy, const int32_t * __restrict__ ids, const ggml_cuda_mm_fusion_args_device fusion, float * __restrict__ dst,
@@ -222,22 +410,13 @@ static __global__ void mul_mat_vec_q(
 
     const uint32_t channel_dst = blockIdx.y;
 
-    uint32_t token_idx = 0;
     uint32_t channel_x;
     uint32_t channel_y;
     uint32_t sample_dst;
 
-    if constexpr (is_multi_token_id) {
-        // Multi-token MUL_MAT_ID path, adding these in the normal path causes a perf regression for n_tokens=1 case
-        token_idx  = blockIdx.z;
-        channel_x  = ids[channel_dst + token_idx * ids_stride];
-        channel_y  = fastmodulo(channel_dst, nchannels_y);
-        sample_dst = 0;
-    } else {
-        channel_x  = ncols_dst == 1 && ids ? ids[channel_dst]                     : fastdiv(channel_dst, channel_ratio);
-        channel_y  = ncols_dst == 1 && ids ? fastmodulo(channel_dst, nchannels_y) : channel_dst;
-        sample_dst = blockIdx.z;
-    }
+    channel_x  = ncols_dst == 1 && ids ? ids[channel_dst]                     : fastdiv(channel_dst, channel_ratio);
+    channel_y  = ncols_dst == 1 && ids ? fastmodulo(channel_dst, nchannels_y) : channel_dst;
+    sample_dst = blockIdx.z;
 
     const uint32_t sample_x    = fastdiv(sample_dst, sample_ratio);
     const uint32_t sample_y    = sample_dst;
@@ -294,9 +473,6 @@ static __global__ void mul_mat_vec_q(
     float tmp_gate[ncols_dst][rows_per_cuda_block] = {{0.0f}};
 
     const block_q8_1 * y = ((const block_q8_1 *) vy) + sample_y*stride_sample_y + channel_y*stride_channel_y;
-    if constexpr (is_multi_token_id) {
-        y += token_idx*stride_col_y;
-    }
     const int kbx_offset = sample_x*stride_sample_x + channel_x*stride_channel_x + row0*stride_row_x;
 
     for (int kbx = tid / (qi/vdr); kbx < blocks_per_row_x; kbx += blocks_per_iter) {
@@ -350,10 +526,6 @@ static __global__ void mul_mat_vec_q(
 
     dst += sample_dst*stride_sample_dst + channel_dst*stride_channel_dst + row0;
 
-    if constexpr (is_multi_token_id) {
-        dst += token_idx*stride_col_dst;
-    }
-
     // sum up partial sums and write back result
 #pragma unroll
     for (int j = 0; j < ncols_dst; ++j) {
@@ -413,6 +585,69 @@ static __global__ void mul_mat_vec_q(
     }
 }
 
+// Dedicated MoE multi-token kernel.
+// Grid: (ceil(nrows_x / c_rows_per_block), nchannels_dst)
+// Block: (warp_size, ncols_dst) - each warp handles one token independently.
+// No shared memory reduction needed since each warp works alone.
+template <ggml_type type, int c_rows_per_block>
+__launch_bounds__(get_mmvq_mmid_max_batch_for_device<type>()*ggml_cuda_get_physical_warp_size(), 1)
+static __global__ void mul_mat_vec_q_moe(
+        const void * __restrict__ vx, const void * __restrict__ vy, const int32_t * __restrict__ ids,
+        float * __restrict__ dst,
+        const uint32_t ncols_x, const uint3 nchannels_y, const uint32_t nrows_x,
+        const uint32_t stride_row_x, const uint32_t stride_col_y, const uint32_t stride_col_dst,
+        const uint32_t stride_channel_x, const uint32_t stride_channel_y, const uint32_t stride_channel_dst,
+        const uint32_t ncols_dst, const uint32_t ids_stride) {
+
+    constexpr int qk  = ggml_cuda_type_traits<type>::qk;
+    constexpr int qi  = ggml_cuda_type_traits<type>::qi;
+    constexpr int vdr = get_vdr_mmvq(type);
+    constexpr int warp_size = ggml_cuda_get_physical_warp_size();
+
+    constexpr vec_dot_q_cuda_t vec_dot_q_cuda = get_vec_dot_q_cuda(type);
+
+    const uint32_t token_idx   = threadIdx.y;
+    const int      row0        = c_rows_per_block*blockIdx.x;
+    const int      blocks_per_row_x = ncols_x / qk;
+    constexpr int  blocks_per_iter  = vdr * warp_size / qi;
+
+    const uint32_t channel_dst = blockIdx.y;
+
+    if (token_idx >= ncols_dst) {
+        return;
+    }
+
+    const uint32_t channel_x = ids[channel_dst + token_idx * ids_stride];
+    const uint32_t channel_y = fastmodulo(channel_dst, nchannels_y);
+
+    const block_q8_1 * y = ((const block_q8_1 *) vy) + channel_y*stride_channel_y + token_idx*stride_col_y;
+    const int kbx_offset  = channel_x*stride_channel_x + row0*stride_row_x;
+
+    // partial sum for each thread
+    float tmp[c_rows_per_block] = {0.0f};
+
+    for (int kbx = threadIdx.x / (qi/vdr); kbx < blocks_per_row_x; kbx += blocks_per_iter) {
+        const int kby = kbx * (qk/QK8_1);
+        const int kqs = vdr * (threadIdx.x % (qi/vdr));
+
+#pragma unroll
+        for (int i = 0; i < c_rows_per_block; ++i) {
+            tmp[i] += vec_dot_q_cuda(vx, &y[kby], kbx_offset + i*stride_row_x + kbx, kqs);
+        }
+    }
+
+    // Warp-level reduction only - no shared memory needed
+#pragma unroll
+    for (int i = 0; i < c_rows_per_block; ++i) {
+        tmp[i] = warp_reduce_sum<warp_size>(tmp[i]);
+    }
+
+    // Write results
+    if (threadIdx.x < c_rows_per_block && (c_rows_per_block == 1 || uint32_t(row0 + threadIdx.x) < nrows_x)) {
+        dst[channel_dst*stride_channel_dst + token_idx*stride_col_dst + row0 + threadIdx.x] = tmp[threadIdx.x];
+    }
+}
+
 template<ggml_type type>
 static std::pair<dim3, dim3> calc_launch_params(
         const int ncols_dst, const int nrows_x, const int nchannels_dst, const int nsamples_or_ntokens,
@@ -425,7 +660,7 @@ static std::pair<dim3, dim3> calc_launch_params(
     return {block_nums, block_dims};
 }
 
-template<ggml_type type, int c_ncols_dst, bool is_multi_token_id = false, bool small_k = false>
+template<ggml_type type, int c_ncols_dst, bool small_k = false>
 static void mul_mat_vec_q_switch_fusion(
         const void * vx, const void * vy, const int32_t * ids, const ggml_cuda_mm_fusion_args_device fusion, float * dst,
         const uint32_t ncols_x, const uint3 nchannels_y, const uint32_t stride_row_x, const uint32_t stride_col_y,
@@ -438,7 +673,7 @@ static void mul_mat_vec_q_switch_fusion(
     const bool has_fusion = fusion.gate != nullptr || fusion.x_bias != nullptr || fusion.gate_bias != nullptr;
     if constexpr (c_ncols_dst == 1) {
         if (has_fusion) {
-            mul_mat_vec_q<type, c_ncols_dst, true, is_multi_token_id, small_k><<<block_nums, block_dims, nbytes_shared, stream>>>
+            mul_mat_vec_q<type, c_ncols_dst, true, small_k><<<block_nums, block_dims, nbytes_shared, stream>>>
                 (vx, vy, ids, fusion, dst, ncols_x, nchannels_y, stride_row_x, stride_col_y, stride_col_dst,
                  channel_ratio, stride_channel_x, stride_channel_y, stride_channel_dst,
                  sample_ratio, stride_sample_x, stride_sample_y, stride_sample_dst, ids_stride);
@@ -448,12 +683,33 @@ static void mul_mat_vec_q_switch_fusion(
 
     GGML_ASSERT(!has_fusion && "fusion only supported for ncols_dst=1");
 
-    mul_mat_vec_q<type, c_ncols_dst, false, is_multi_token_id, small_k><<<block_nums, block_dims, nbytes_shared, stream>>>
+    mul_mat_vec_q<type, c_ncols_dst, false, small_k><<<block_nums, block_dims, nbytes_shared, stream>>>
         (vx, vy, ids, fusion, dst, ncols_x, nchannels_y, stride_row_x, stride_col_y, stride_col_dst,
         channel_ratio, stride_channel_x, stride_channel_y, stride_channel_dst,
         sample_ratio, stride_sample_x, stride_sample_y, stride_sample_dst, ids_stride);
 }
 
+template <ggml_type type>
+static void mul_mat_vec_q_moe_launch(
+        const void * vx, const void * vy, const int32_t * ids, float * dst,
+        const uint32_t ncols_x, const uint3 nchannels_y, const uint32_t nrows_x,
+        const uint32_t stride_row_x, const uint32_t stride_col_y, const uint32_t stride_col_dst,
+        const uint32_t stride_channel_x, const uint32_t stride_channel_y, const uint32_t stride_channel_dst,
+        const uint32_t ncols_dst, const uint32_t ids_stride,
+        const int warp_size, const int nchannels_dst, cudaStream_t stream) {
+
+    constexpr int rows_per_block = 2; // 2 gives best perf based on tuning
+    const int64_t nblocks_rows = (nrows_x + rows_per_block - 1) / rows_per_block;
+    const dim3 block_nums(nblocks_rows, nchannels_dst);
+    const dim3 block_dims(warp_size, ncols_dst);
+
+    mul_mat_vec_q_moe<type, rows_per_block><<<block_nums, block_dims, 0, stream>>>(
+        vx, vy, ids, dst, ncols_x, nchannels_y, nrows_x,
+        stride_row_x, stride_col_y, stride_col_dst,
+        stride_channel_x, stride_channel_y, stride_channel_dst,
+        ncols_dst, ids_stride);
+}
+
 template <ggml_type type>
 static void mul_mat_vec_q_switch_ncols_dst(
         const void * vx, const void * vy, const int32_t * ids, const ggml_cuda_mm_fusion_args_device fusion, float * dst,
@@ -472,20 +728,62 @@ static void mul_mat_vec_q_switch_ncols_dst(
     const uint3 sample_ratio_fd  = init_fastdiv_values(nsamples_dst  / nsamples_x);
 
     const int device = ggml_cuda_get_device();
+    const int                     cc        = ggml_cuda_info().devices[device].cc;
     const int warp_size = ggml_cuda_info().devices[device].warp_size;
-    const mmvq_parameter_table_id table_id = get_device_table_id(ggml_cuda_info().devices[device].cc);
+    const mmvq_parameter_table_id table_id  = get_device_table_id(cc);
 
     const bool has_fusion = fusion.gate != nullptr || fusion.x_bias != nullptr || fusion.gate_bias != nullptr;
     const bool has_ids = ids != nullptr;
 
+    const auto should_use_small_k = [&](int c_ncols_dst) {
+        // When K is small, increase rows_per_block to match nwarps so each warp has more work to do
+        // Trigger when the full thread block covers all K blocks in a single loop iteration and few threads remain idle.
+        constexpr int qk                    = ggml_cuda_type_traits<type>::qk;
+        constexpr int qi                    = ggml_cuda_type_traits<type>::qi;
+        constexpr int vdr                   = get_vdr_mmvq(type);
+        const int     blocks_per_row_x      = ncols_x / qk;
+        const int     blocks_per_iter_1warp = vdr * warp_size / qi;
+        const int     nwarps                = calc_nwarps(type, c_ncols_dst, table_id);
+        bool          use                   = nwarps > 1 && blocks_per_row_x < nwarps * blocks_per_iter_1warp;
+
+        constexpr std::array<ggml_type, 2> iq_slow_turing = {
+            GGML_TYPE_IQ3_XXS,
+            GGML_TYPE_IQ3_S,
+        };
+        constexpr std::array<ggml_type, 8> iq_slow_other = {
+            GGML_TYPE_IQ1_S, GGML_TYPE_IQ1_M,   GGML_TYPE_IQ2_XXS, GGML_TYPE_IQ2_XS,
+            GGML_TYPE_IQ2_S, GGML_TYPE_IQ3_XXS, GGML_TYPE_IQ3_S,   GGML_TYPE_IQ4_XS,
+        };
+        constexpr std::array<ggml_type, 3> slow_pascal = {
+            GGML_TYPE_IQ3_S,
+            GGML_TYPE_Q2_K,
+            GGML_TYPE_Q3_K,
+        };
+
+        const bool is_nvidia_turing_plus  = GGML_CUDA_CC_IS_NVIDIA(cc) && cc >= GGML_CUDA_CC_TURING;
+        const bool is_nvidia_pascal_older = GGML_CUDA_CC_IS_NVIDIA(cc) && cc < GGML_CUDA_CC_VOLTA;
+
+        if (is_nvidia_turing_plus) {
+            if (ncols_dst == 1 &&
+                    std::find(iq_slow_turing.begin(), iq_slow_turing.end(), type) != iq_slow_turing.end()) {
+                use = false;
+            }
+        } else if ((ncols_dst == 1 && std::find(iq_slow_other.begin(), iq_slow_other.end(), type) != iq_slow_other.end()) ||
+                (is_nvidia_pascal_older && std::find(slow_pascal.begin(), slow_pascal.end(), type) != slow_pascal.end()) ||
+                GGML_CUDA_CC_IS_RDNA(cc)) {
+            use = false;
+        }
+
+        return use;
+    };
+
     if (has_ids && ncols_dst > 1) {
-        // Multi-token MUL_MAT_ID path only - single-token goes through regular path below
-        constexpr int c_ncols_dst = 1;
-        std::pair<dim3, dim3> dims = calc_launch_params<type>(c_ncols_dst, nrows_x, nchannels_dst, ncols_dst, warp_size, table_id);
-        mul_mat_vec_q_switch_fusion<type, c_ncols_dst, true>(vx, vy, ids, fusion, dst, ncols_x, nchannels_y_fd, stride_row_x, stride_col_y, stride_col_dst,
-             channel_ratio_fd, stride_channel_x, stride_channel_y, stride_channel_dst,
-             sample_ratio_fd, stride_sample_x, stride_sample_y, stride_sample_dst,
-             dims.first, dims.second, 0, ids_stride, stream);
+        // Multi-token MUL_MAT_ID path - dedicated MoE kernel
+        mul_mat_vec_q_moe_launch<type>(
+            vx, vy, ids, dst, ncols_x, nchannels_y_fd, nrows_x,
+            stride_row_x, stride_col_y, stride_col_dst,
+            stride_channel_x, stride_channel_y, stride_channel_dst,
+            ncols_dst, ids_stride, warp_size, nchannels_dst, stream);
         return;
     }
 
@@ -493,31 +791,24 @@ static void mul_mat_vec_q_switch_ncols_dst(
         case 1: {
             constexpr int c_ncols_dst = 1;
 
-            // When K is small, increase rows_per_block to match nwarps so each warp has more work to do
-            // Trigger when the full thread block covers all K blocks in a single loop iteration and few threads remain idle.
-            constexpr int qk  = ggml_cuda_type_traits<type>::qk;
-            constexpr int qi  = ggml_cuda_type_traits<type>::qi;
-            constexpr int vdr = get_vdr_mmvq(type);
-            const int blocks_per_row_x = ncols_x / qk;
-            const int blocks_per_iter_1warp = vdr * warp_size / qi;
-            const int nwarps = calc_nwarps(type, c_ncols_dst, table_id);
-            const bool use_small_k = nwarps > 1 && blocks_per_row_x < nwarps * blocks_per_iter_1warp;
+            bool use_small_k = should_use_small_k(c_ncols_dst);
+
             if (use_small_k) {
-                std::pair<dim3, dim3> dims = calc_launch_params<type>(c_ncols_dst, nrows_x, nchannels_dst, nsamples_dst,
-                                                                    warp_size, table_id, true);
-                mul_mat_vec_q_switch_fusion<type, c_ncols_dst, false, true>(
+                std::pair<dim3, dim3> dims = calc_launch_params<type>(c_ncols_dst, nrows_x, nchannels_dst,
+                                                                        nsamples_dst, warp_size, table_id, true);
+                mul_mat_vec_q_switch_fusion<type, c_ncols_dst, true>(
                     vx, vy, ids, fusion, dst, ncols_x, nchannels_y_fd, stride_row_x, stride_col_y, stride_col_dst,
-                    channel_ratio_fd, stride_channel_x, stride_channel_y, stride_channel_dst,
-                    sample_ratio_fd, stride_sample_x, stride_sample_y, stride_sample_dst,
-                    dims.first, dims.second, 0, ids_stride, stream);
+                    channel_ratio_fd, stride_channel_x, stride_channel_y, stride_channel_dst, sample_ratio_fd,
+                    stride_sample_x, stride_sample_y, stride_sample_dst, dims.first, dims.second, 0, ids_stride,
+                    stream);
             } else {
-                std::pair<dim3, dim3> dims = calc_launch_params<type>(c_ncols_dst, nrows_x, nchannels_dst, nsamples_dst,
-                                                                    warp_size, table_id);
+                std::pair<dim3, dim3> dims = calc_launch_params<type>(c_ncols_dst, nrows_x, nchannels_dst,
+                                                                        nsamples_dst, warp_size, table_id);
                 mul_mat_vec_q_switch_fusion<type, c_ncols_dst>(
                     vx, vy, ids, fusion, dst, ncols_x, nchannels_y_fd, stride_row_x, stride_col_y, stride_col_dst,
-                    channel_ratio_fd, stride_channel_x, stride_channel_y, stride_channel_dst,
-                    sample_ratio_fd, stride_sample_x, stride_sample_y, stride_sample_dst,
-                    dims.first, dims.second, 0, ids_stride, stream);
+                    channel_ratio_fd, stride_channel_x, stride_channel_y, stride_channel_dst, sample_ratio_fd,
+                    stride_sample_x, stride_sample_y, stride_sample_dst, dims.first, dims.second, 0, ids_stride,
+                    stream);
             }
         } break;
         case 2: {

ggml/src/ggml-cuda/mmvq.cuh

@@ -1,7 +1,10 @@
 #include "common.cuh"
 
 #define MMVQ_MAX_BATCH_SIZE 8 // Max. batch size for which to use MMVQ kernels.
-#define MMVQ_MMID_MAX_BATCH_SIZE 4 // Max. batch size for which to use MMVQ kernels for MUL_MAT_ID
+
+// Returns the maximum batch size for which MMVQ should be used for MUL_MAT_ID,
+// based on the quantization type and GPU architecture (compute capability).
+int get_mmvq_mmid_max_batch(ggml_type type, int cc);
 
 void ggml_cuda_mul_mat_vec_q(ggml_backend_cuda_context & ctx,
     const ggml_tensor * src0, const ggml_tensor * src1, const ggml_tensor * ids, ggml_tensor * dst, const ggml_cuda_mm_fusion_args_host * fusion = nullptr);

ggml/src/ggml-hexagon/htp/flash-attn-ops.c

@@ -346,6 +346,9 @@ static void flash_attn_ext_f16_thread(unsigned int nth, unsigned int ith, void *
 
     const HVX_Vector logit_cap = hvx_vec_splat_f32(factx->logit_softcap);
 
+    dma_cache m_cache;
+    dma_cache_init(&m_cache, spad_m, factx->size_m_block, DMA_CACHE_MAX_SIZE);
+
     for (uint32_t ir = ir0; ir < ir1; ++ir) {
         const uint32_t iq3 = fastdiv(ir, &factx->src0_div21);
         const uint32_t iq2 = fastdiv(ir - iq3*neq2*neq1, &factx->src0_div1);
@@ -389,9 +392,8 @@ static void flash_attn_ext_f16_thread(unsigned int nth, unsigned int ith, void *
             // Mask
             if (mask) {
                 const uint8_t * m_src = (const uint8_t *) (mp_base + ic_start);
-                uint8_t * m_dst = spad_m + (ib % 2) * factx->size_m_block;
                 // Mask is 1D contiguous for this row
-                dma_queue_push(dma, dma_make_ptr(m_dst, m_src), current_block_size * 2, current_block_size * 2, current_block_size * 2, 1);
+                dma_cache_push(dma, &m_cache, m_src, current_block_size * 2, current_block_size * 2, current_block_size * 2, 1);
             }
 
             // FARF(HIGH, "fa %u: prefetch KVM: ir %u ib %u iq1 %u iq2 %u iq3 %u : size_k_row %u size_v_row %u bs %u: usec %u",
@@ -554,7 +556,7 @@ static void flash_attn_ext_f16_thread(unsigned int nth, unsigned int ith, void *
                 // Mask
                 if (mask) {
                     const uint8_t * m_src = (const uint8_t *) (mp_base + next_ic_start);
-                    dma_queue_push(dma, dma_make_ptr(m_base, m_src), next_block_size * 2, next_block_size * 2, next_block_size * 2, 1);
+                    dma_cache_push(dma, &m_cache, m_src, next_block_size * 2, next_block_size * 2, next_block_size * 2, 1);
                 }
 
                 // FARF(HIGH, "fa %u: prefetch KVM: ir %u ib %u : iq1 %u iq2 %u iq3 %u : size_k_row %u size_v_row %u bs %u: usec %u",
@@ -684,7 +686,7 @@ int op_flash_attn_ext(struct htp_ops_context * octx) {
     octx->src0_spad.size_per_thread = size_q_block * 1;
     octx->src1_spad.size_per_thread = factx.size_k_block * 2;
     octx->src2_spad.size_per_thread = factx.size_v_block * 2;
-    octx->src3_spad.size_per_thread = mask ? factx.size_m_block * 2 : 0;
+    octx->src3_spad.size_per_thread = mask ? factx.size_m_block * DMA_CACHE_MAX_SIZE : 0;
     octx->dst_spad.size_per_thread  = size_vkq_acc;
 
     octx->src0_spad.size = octx->src0_spad.size_per_thread * octx->n_threads;
@@ -705,6 +707,8 @@ int op_flash_attn_ext(struct htp_ops_context * octx) {
     octx->src3_spad.data = octx->src2_spad.data + octx->src2_spad.size;
     octx->dst_spad.data  = octx->src3_spad.data + octx->src3_spad.size;
 
+    // FARF(ERROR, "fa: qrows-per-thread %u", factx.qrows_per_thread);
+
     if (!(octx->flags & HTP_OPFLAGS_SKIP_COMPUTE)) {
         worker_pool_run_func(octx->ctx->worker_pool, flash_attn_ext_f16_thread, &factx, octx->n_threads);
     }

ggml/src/ggml-hexagon/htp/hex-dma.h

@@ -143,7 +143,7 @@ static inline bool dma_queue_push_single_1d(dma_queue * q, dma_ptr dptr, size_t
     desc->desc_size  = 0; // 1D mode
     desc->src_bypass = dma_src_l2_bypass_on;
     desc->dst_bypass = dma_dst_l2_bypass_on;
-    desc->order      = 1;
+    desc->order      = 0;
     desc->done       = 0;
     desc->src        = (void *) dptr.src;
     desc->dst        = (void *) dptr.dst;
@@ -151,8 +151,12 @@ static inline bool dma_queue_push_single_1d(dma_queue * q, dma_ptr dptr, size_t
 
     q->dptr[q->push_idx] = dptr;
 
-    dmlink(q->tail, desc);
-    q->tail = (dma_descriptor_2d *) desc;
+    if (size) {
+        dmlink(q->tail, desc);
+        q->tail = (dma_descriptor_2d *) desc;
+    } else {
+        desc->done = 1;
+    }
 
     // FARF(ERROR, "dma-push: i %u row-size %u nrows %d dst %p src %p\n", q->push_idx, row_size, nrows, dptr.dst, dptr.src);
     q->push_idx = (q->push_idx + 1) & q->idx_mask;
@@ -175,7 +179,7 @@ static inline bool dma_queue_push_single_2d(dma_queue * q, dma_ptr dptr, size_t
     desc->dst_bypass     = dma_dst_l2_bypass_on;
     desc->src_comp       = 0;
     desc->dst_comp       = 0;
-    desc->order          = 1;
+    desc->order          = 0;
     desc->done           = 0;
     desc->src_stride     = src_stride;
     desc->dst_stride     = dst_stride;
@@ -197,8 +201,12 @@ static inline bool dma_queue_push_single_2d(dma_queue * q, dma_ptr dptr, size_t
 
     q->dptr[q->push_idx] = dptr;
 
-    dmlink(q->tail, desc);
-    q->tail = desc;
+    if (nrows) {
+        dmlink(q->tail, desc);
+        q->tail = desc;
+    } else {
+        desc->done = 1;
+    }
 
     // FARF(ERROR, "dma-push: i %u row-size %u nrows %d dst %p src %p\n", q->push_idx, row_size, nrows, dptr.dst, dptr.src);
     q->push_idx = (q->push_idx + 1) & q->idx_mask;
@@ -215,12 +223,9 @@ static inline dma_ptr dma_queue_pop(dma_queue * q) {
     dma_descriptor_2d * desc = &q->desc[q->pop_idx];
 
     // Wait for desc to complete
-    while (1) {
-        dmpoll();
-        if (desc->done) {
-            break;
-        }
+    while (!desc->done) {
         // FARF(ERROR, "dma-pop: waiting for DMA : %u\n", q->pop_idx);
+        dmpoll();
     }
 
     dptr = q->dptr[q->pop_idx];
@@ -312,6 +317,54 @@ static inline bool dma_queue_push_vtcm_to_ddr(dma_queue * q, dma_ptr dptr, size_
     return dma_queue_push(q, dptr, dst_row_size, src_row_size, dst_row_size, nrows);
 }
 
+#define DMA_CACHE_MAX_SIZE 64U
+
+typedef struct {
+    uint8_t *base;
+    uint32_t line_size;
+    uint32_t capacity;
+    uint32_t src[DMA_CACHE_MAX_SIZE];
+    uint16_t age[DMA_CACHE_MAX_SIZE];
+} dma_cache;
+
+static inline void dma_cache_init(dma_cache *c, uint8_t *base, uint32_t line_size, uint32_t capacity)
+{
+    c->capacity  = (capacity > DMA_CACHE_MAX_SIZE) ? DMA_CACHE_MAX_SIZE : capacity;
+    c->base      = base;
+    c->line_size = line_size;
+
+    for (unsigned i=0; i < c->capacity; i++) {
+        c->src[i] = 0;
+        c->age[i] = 0;
+    }
+}
+
+static inline bool dma_cache_push(dma_queue *q, dma_cache *c, const uint8_t * src, uint32_t dst_stride, uint32_t src_stride, uint32_t row_size, uint32_t nrows)
+{
+    uint32_t o_idx = 0;
+    uint16_t o_age = 0;
+    uint8_t *  dst = 0;
+
+    for (unsigned i=0; i < c->capacity; i++) {
+        if (c->src[i] == (uint32_t) src) {
+            c->age[i] = 0;
+            dst = c->base + (i * c->line_size); nrows = 0; // dummy dma
+            // FARF(ERROR, "dma-cache: found %p", src);
+        } else {
+            c->age[i]++;
+            if (c->age[i] > o_age) { o_age = c->age[i]; o_idx = i; }
+        }
+    }
+    if (!dst) {
+        // FARF(ERROR, "dma-cache: replacing #%u : age %u %p -> %p", o_idx, c->age[o_idx], (void *) c->src[o_idx], src);
+        c->age[o_idx] = 0;
+        c->src[o_idx] = (uint32_t) src;
+        dst = c->base + o_idx * c->line_size; // normal nrows dma
+    }
+
+    return dma_queue_push(q, dma_make_ptr(dst, src), dst_stride, src_stride, row_size, nrows);
+}
+
 #ifdef __cplusplus
 }  // extern "C"
 #endif

ggml/src/ggml-hexagon/htp/rope-ops.c

@@ -333,8 +333,8 @@ static void rope_job_f32(unsigned int nth, unsigned int ith, void * data) {
                     //         (unsigned) HAP_perf_qtimer_count_to_us(HAP_perf_get_qtimer_count() - rctx->t_start));
                 }
 
-                // Skip DMA transactions from prev block (if any)
-                // No need to wait for these since the DMA is setup for in-order processing
+                // Skip output DMA transactions from prev block (if any)
+                // No need to wait for those here since we're explicitly waiting for the latest prefecthes below.
                 for (uint32_t d=0; d < dma_depth; d++) { dma_queue_pop_nowait(dma_queue); }
 
                 // Compute loop

gguf-py/gguf/vocab.py

@@ -14,12 +14,12 @@ except ImportError:
     SentencePieceProcessor: Any = None
 
 try:
-    from mistral_common.tokens.tokenizers.mistral import MistralTokenizer # type: ignore[import-not-found]
-    from mistral_common.tokens.tokenizers.tekken import Tekkenizer # type: ignore[import-not-found]
-    from mistral_common.tokens.tokenizers.utils import ( # type: ignore[import-not-found]
+    from mistral_common.tokens.tokenizers.mistral import MistralTokenizer # type: ignore[import-not-found, ty:unresolved-import]
+    from mistral_common.tokens.tokenizers.tekken import Tekkenizer # type: ignore[import-not-found, ty:unresolved-import]
+    from mistral_common.tokens.tokenizers.utils import ( # type: ignore[import-not-found, ty:unresolved-import]
         _filter_valid_tokenizer_files,
     )
-    from mistral_common.tokens.tokenizers.sentencepiece import ( # type: ignore[import-not-found]
+    from mistral_common.tokens.tokenizers.sentencepiece import ( # type: ignore[import-not-found, ty:unresolved-import]
         SentencePieceTokenizer,
     )
 except ImportError:
@@ -32,7 +32,7 @@ else:
     _mistral_common_installed = True
 
 try:
-    from mistral_common.tokens.tokenizers.utils import ( # type: ignore[import-not-found]
+    from mistral_common.tokens.tokenizers.utils import ( # type: ignore[import-not-found, ty:unresolved-import]
         get_one_valid_tokenizer_file,
     )
 except ImportError:

scripts/gen-unicode-data.py

@@ -147,7 +147,7 @@ ranges_nfd: list[tuple[int, int, int]] = [(0, 0, 0)]  # start, last, nfd
 for codepoint, norm in table_nfd:
     start = ranges_nfd[-1][0]
     if ranges_nfd[-1] != (start, codepoint - 1, norm):
-        ranges_nfd.append(None)  # type: ignore[arg-type]  # dummy, will be replaced below
+        ranges_nfd.append((0, 0, 0))  # dummy, will be replaced below
         start = codepoint
     ranges_nfd[-1] = (start, codepoint, norm)
 

src/llama-arch.cpp

@@ -557,6 +557,8 @@ static std::set<llm_tensor> llm_get_tensor_names(llm_arch arch) {
                 LLM_TENSOR_OUTPUT_NORM,
                 LLM_TENSOR_OUTPUT,
                 LLM_TENSOR_ROPE_FREQS,
+                LLM_TENSOR_ROPE_FACTORS_LONG,
+                LLM_TENSOR_ROPE_FACTORS_SHORT,
                 LLM_TENSOR_ATTN_NORM,
                 LLM_TENSOR_ATTN_Q,
                 LLM_TENSOR_ATTN_K,

src/llama-model-loader.cpp

@@ -1158,6 +1158,12 @@ struct ggml_tensor * llama_model_loader::create_tensor(
                     if (overrides->buft == ggml_backend_cpu_buffer_type()) {
                         // when overriding to a CPU buffer, consider the extra buffer types
                         buft = select_weight_buft(hparams, t_meta, op, buft_list_cpu);
+                        if (use_mmap) {
+                            static std::once_flag once;
+                            std::call_once(once, [] {
+                                LLAMA_LOG_WARN("llama_model_loader: tensor overrides to CPU are used with mmap enabled - consider using --no-mmap for better performance\n");
+                            });
+                        }
                     } else {
                         buft = overrides->buft;
                     }

tools/server/server-cors-proxy.h

@@ -32,13 +32,22 @@ static server_http_res_ptr proxy_request(const server_http_req & req, std::strin
 
     SRV_INF("proxying %s request to %s://%s:%i%s\n", method.c_str(), parsed_url.scheme.c_str(), parsed_url.host.c_str(), parsed_url.port, parsed_url.path.c_str());
 
+    std::map<std::string, std::string> headers;
+    for (auto [key, value] : req.headers) {
+        auto new_key = key;
+        if (string_starts_with(new_key, "X-Proxy-Header-")) {
+            string_replace_all(new_key, "X-Proxy-Header-", "");
+        }
+        headers[new_key] = value;
+    }
+
     auto proxy = std::make_unique<server_http_proxy>(
             method,
             parsed_url.scheme,
             parsed_url.host,
             parsed_url.port,
             parsed_url.path,
-            req.headers,
+            headers,
             req.body,
             req.should_stop,
             600, // timeout_read (default to 10 minutes)

tools/server/server-http.h

@@ -35,7 +35,7 @@ using server_http_res_ptr = std::unique_ptr<server_http_res>;
 
 struct server_http_req {
     std::map<std::string, std::string> params; // path_params + query_params
-    std::map<std::string, std::string> headers; // reserved for future use
+    std::map<std::string, std::string> headers; // used by MCP proxy
     std::string path;
     std::string query_string; // query parameters string (e.g. "action=save")
     std::string body;

tools/server/tests/utils.py

@@ -116,7 +116,7 @@ class ServerProcess:
             self.server_port = int(os.environ["PORT"])
         self.external_server = "DEBUG_EXTERNAL" in os.environ
 
-    def start(self, timeout_seconds: int | None = DEFAULT_HTTP_TIMEOUT) -> None:
+    def start(self, timeout_seconds: int = DEFAULT_HTTP_TIMEOUT) -> None:
         if self.external_server:
             print(f"[external_server]: Assuming external server running on {self.server_host}:{self.server_port}")
             return

tools/server/webui/src/lib/services/mcp.service.ts

@@ -39,7 +39,13 @@ import type {
 	MCPResourceContent,
 	MCPReadResourceResult
 } from '$lib/types';
-import { buildProxiedUrl, throwIfAborted, isAbortError, createBase64DataUrl } from '$lib/utils';
+import {
+	buildProxiedUrl,
+	buildProxiedHeaders,
+	throwIfAborted,
+	isAbortError,
+	createBase64DataUrl
+} from '$lib/utils';
 
 interface ToolResultContentItem {
 	type: string;
@@ -118,7 +124,7 @@ export class MCPService {
 		const requestInit: RequestInit = {};
 
 		if (config.headers) {
-			requestInit.headers = config.headers;
+			requestInit.headers = buildProxiedHeaders(config.headers);
 		}
 
 		if (config.credentials) {

tools/server/webui/src/lib/utils/cors-proxy.ts

@@ -19,6 +19,21 @@ export function buildProxiedUrl(targetUrl: string): URL {
 	return proxyUrl;
 }
 
+/**
+ * Wrap original headers for proxying through the CORS proxy. This avoids issues with duplicated llama.cpp-specific and target headers when using the CORS proxy.
+ * @param headers - The original headers to be proxied to target
+ * @returns List of "wrapped" headers to be sent to the CORS proxy
+ */
+export function buildProxiedHeaders(headers: Record<string, string>): Record<string, string> {
+	const proxiedHeaders: Record<string, string> = {};
+
+	for (const [key, value] of Object.entries(headers)) {
+		proxiedHeaders[`X-Proxy-Header-${key}`] = value;
+	}
+
+	return proxiedHeaders;
+}
+
 /**
  * Get a proxied URL string for use in fetch requests.
  * @param targetUrl - The original URL to proxy

tools/server/webui/src/lib/utils/index.ts

@@ -38,7 +38,7 @@ export { highlightCode, detectIncompleteCodeBlock, type IncompleteCodeBlock } fr
 export { setConfigValue, getConfigValue, configToParameterRecord } from './config-helpers';
 
 // CORS Proxy
-export { buildProxiedUrl, getProxiedUrlString } from './cors-proxy';
+export { buildProxiedUrl, getProxiedUrlString, buildProxiedHeaders } from './cors-proxy';
 
 // Conversation utilities
 export { createMessageCountMap, getMessageCount } from './conversation-utils';