vbatts/llama.cpp
1
0
Fork 0
Code Issues Pull requests Projects Releases Packages Wiki Activity Actions

threadpool: update backend interface in ggml-rpc

Browse source
This commit is contained in:
Max Krasnyansky 2024-05-27 22:13:07 -07:00
parent a67dbcc538
commit 1d9d39a18e
1 changed files with 3 additions and 2 deletions
Download patch file Download diff file Expand all files Collapse all files

5
ggml-rpc.cpp
View file

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