Merge branch 'threadpool' of https://github.com/CodeLinaro/llama.cpp into threadpool

Author: fmz

ggml-cuda.cu

@@ -2495,13 +2495,9 @@ static bool ggml_graph_node_has_matching_properties(ggml_tensor * node, ggml_gra
     return true;
 }
 
-GGML_CALL static enum ggml_status ggml_backend_cuda_graph_compute(
-                   ggml_backend_t   backend,
-                      ggml_cgraph * cgraph,
-        ggml_compute_threadpool_t   threadpool) {
-
-    GGML_UNUSED(threadpool);
+GGML_CALL static enum ggml_status ggml_backend_cuda_graph_compute(ggml_backend_t backend, ggml_cgraph * cgraph) {
     ggml_backend_cuda_context * cuda_ctx = (ggml_backend_cuda_context *)backend->context;
+
     ggml_cuda_set_device(cuda_ctx->device);
 
 #ifdef USE_CUDA_GRAPH

ggml-kompute.cpp

@@ -1948,12 +1948,7 @@ static ggml_backend_buffer_type_t ggml_backend_kompute_get_default_buffer_type(g
     return ggml_backend_kompute_buffer_type(ctx->device);
 }
 
-static ggml_status ggml_backend_kompute_graph_compute(
-                  ggml_backend_t   backend,
-              struct ggml_cgraph * cgraph
-       ggml_compute_threadpool_t   threadpool) {
-
-    GGML_UNUSED(threadpool);
+static ggml_status ggml_backend_kompute_graph_compute(ggml_backend_t backend, struct ggml_cgraph * cgraph) {
     auto * ctx = static_cast<ggml_kompute_context *>(backend->context);
     ggml_vk_graph_compute(ctx, cgraph);
     return GGML_STATUS_SUCCESS;

ggml-metal.m

@@ -3103,12 +3103,7 @@ GGML_CALL static ggml_backend_buffer_type_t ggml_backend_metal_get_default_buffe
     UNUSED(backend);
 }
 
-GGML_CALL static enum ggml_status ggml_backend_metal_graph_compute(
-                   ggml_backend_t   backend,
-               struct ggml_cgraph * cgraph,
-        ggml_compute_threadpool_t   threadpool) {
-
-    UNUSED(threadpool);
+GGML_CALL static enum ggml_status ggml_backend_metal_graph_compute(ggml_backend_t backend, struct ggml_cgraph * cgraph) {
     struct ggml_metal_context * metal_ctx = (struct ggml_metal_context *)backend->context;
 
     return ggml_metal_graph_compute(metal_ctx, cgraph);

ggml-opencl.cpp

@@ -2235,12 +2235,7 @@ static ggml_backend_buffer_type_t ggml_backend_opencl_get_default_buffer_type(gg
     GGML_UNUSED(backend);
 }
 
-static ggml_status ggml_backend_opencl_graph_compute(
-                   ggml_backend_t   backend,
-                      ggml_cgraph * graph,
-        ggml_compute_threadpool_t   threadpool) {
-
-    GGML_UNUSED(threadpool);
+static ggml_status ggml_backend_opencl_graph_compute(ggml_backend_t backend, ggml_cgraph * graph) {
     for (int i = 0; i < graph->n_nodes; ++i) {
         ggml_tensor * node = graph->nodes[i];
 

ggml-rpc.cpp

@@ -585,8 +585,7 @@ static void serialize_graph(const ggml_cgraph * cgraph, std::vector<uint8_t> & o
     memcpy(out_tensors, tensors.data(), n_tensors * sizeof(rpc_tensor));
 }
 
-GGML_CALL static enum ggml_status ggml_backend_rpc_graph_compute(ggml_backend_t backend, ggml_cgraph * cgraph, ggml_compute_threadpool * tp) {
-    UNUSED(tp);
+GGML_CALL static enum ggml_status ggml_backend_rpc_graph_compute(ggml_backend_t backend, ggml_cgraph * cgraph) {
     ggml_backend_rpc_context * rpc_ctx = (ggml_backend_rpc_context *)backend->context;
     std::vector<uint8_t> input;
     serialize_graph(cgraph, input);
@@ -1021,7 +1020,7 @@ bool rpc_server::graph_compute(const std::vector<uint8_t> & input, std::vector<u
     for (uint32_t i = 0; i < n_nodes; i++) {
         graph->nodes[i] = create_node(nodes[i], ctx, tensor_ptrs, tensor_map);
     }
-    ggml_status status = ggml_backend_graph_compute(backend, graph, NULL);
+    ggml_status status = ggml_backend_graph_compute(backend, graph);
     // output serialization format: | status (1 byte) |
     output.resize(1, 0);
     output[0] = status;

ggml-sycl.cpp

@@ -17022,13 +17022,7 @@ catch (sycl::exception const &exc) {
   std::exit(1);
 }
 
-GGML_CALL static ggml_status ggml_backend_sycl_graph_compute(
-                   ggml_backend_t   backend,
-                      ggml_cgraph * cgraph,
-        ggml_compute_threadpool_t   threadpool) {
-
-    GGML_UNUSED(threadpool);
-
+GGML_CALL static ggml_status ggml_backend_sycl_graph_compute(ggml_backend_t backend, ggml_cgraph * cgraph) {
     ggml_backend_sycl_context * sycl_ctx = (ggml_backend_sycl_context *)backend->context;
     ggml_sycl_set_main_device(sycl_ctx->device);
 

ggml-vulkan.cpp

@@ -6225,12 +6225,7 @@ static bool ggml_vk_is_empty(ggml_tensor * node) {
     return ggml_is_empty(node) || node->op == GGML_OP_NONE || node->op == GGML_OP_RESHAPE || node->op == GGML_OP_TRANSPOSE || node->op == GGML_OP_VIEW || node->op == GGML_OP_PERMUTE;
 }
 
-GGML_CALL static ggml_status ggml_backend_vk_graph_compute(
-                   ggml_backend_t   backend,
-                      ggml_cgraph * cgraph,
-        ggml_compute_threadpool_t   threadpool) {
-
-    GGML_UNUSED(threadpool);
+GGML_CALL static ggml_status ggml_backend_vk_graph_compute(ggml_backend_t backend, ggml_cgraph * cgraph) {
 #ifdef GGML_VULKAN_DEBUG
     std::cerr << "ggml_backend_vk_graph_compute(" << cgraph->n_nodes << " nodes)" << std::endl;
 #endif