summaryrefslogtreecommitdiff
path: root/ggml-rpc.h
diff options
context:
space:
mode:
authorRadoslav Gerganov <rgerganov@gmail.com>2024-05-14 14:27:19 +0300
committerGitHub <noreply@github.com>2024-05-14 14:27:19 +0300
commit5e31828d3e35c76ecfee665bc23771a4bec1d130 (patch)
tree7f5f2edc7c3fc3e7655904316897e32202edd5d6 /ggml-rpc.h
parent541600201e6480f54ae09e58d16b154d4b4b331d (diff)
ggml : add RPC backend (#6829)
* ggml : add RPC backend The RPC backend proxies all operations to a remote server which runs a regular backend (CPU, CUDA, Metal, etc). * set TCP_NODELAY * add CI workflows * Address review comments * fix warning * implement llama_max_devices() for RPC * Address review comments * Address review comments * wrap sockfd into a struct * implement get_alignment and get_max_size * add get_device_memory * fix warning * win32 support * add README * readme : trim trailing whitespace * Address review comments * win32 fix * Address review comments * fix compile warnings on macos
Diffstat (limited to 'ggml-rpc.h')
-rw-r--r--ggml-rpc.h24
1 files changed, 24 insertions, 0 deletions
diff --git a/ggml-rpc.h b/ggml-rpc.h
new file mode 100644
index 00000000..aa144832
--- /dev/null
+++ b/ggml-rpc.h
@@ -0,0 +1,24 @@
+#pragma once
+
+#include "ggml.h"
+#include "ggml-backend.h"
+
+#ifdef __cplusplus
+extern "C" {
+#endif
+
+#define GGML_RPC_MAX_SERVERS 16
+
+// backend API
+GGML_API GGML_CALL ggml_backend_t ggml_backend_rpc_init(const char * endpoint);
+GGML_API GGML_CALL bool ggml_backend_is_rpc(ggml_backend_t backend);
+
+GGML_API GGML_CALL ggml_backend_buffer_type_t ggml_backend_rpc_buffer_type(const char * endpoint);
+
+GGML_API GGML_CALL void ggml_backend_rpc_get_device_memory(const char * endpoint, size_t * free, size_t * total);
+
+GGML_API GGML_CALL void start_rpc_server(ggml_backend_t backend, const char * endpoint, size_t free_mem, size_t total_mem);
+
+#ifdef __cplusplus
+}
+#endif