fix cmake

cmake/paddlepaddle.cmake

@@ -49,7 +49,7 @@ endif()
 
 SET(PADDLE_LIB_PATH "http://paddle-inference-lib.bj.bcebos.com/${PADDLE_LIB_VERSION}/fluid_inference.tgz")
 MESSAGE(STATUS "PADDLE_LIB_PATH=${PADDLE_LIB_PATH}")
-
+if (WITH_GPU OR WITH_MKLML)
 ExternalProject_Add(
     "extern_paddle"
     ${EXTERNAL_PROJECT_LOG_ARGS}
@@ -62,11 +62,24 @@ ExternalProject_Add(
     INSTALL_COMMAND
         ${CMAKE_COMMAND} -E copy_directory ${PADDLE_DOWNLOAD_DIR}/paddle/include ${PADDLE_INSTALL_DIR}/include &&
         ${CMAKE_COMMAND} -E copy_directory ${PADDLE_DOWNLOAD_DIR}/paddle/lib ${PADDLE_INSTALL_DIR}/lib &&
-        ${CMAKE_COMMAND} -E copy_directory ${PADDLE_DOWNLOAD_DIR}/third_party ${PADDLE_INSTALL_DIR}/third_party 
+        ${CMAKE_COMMAND} -E copy_directory ${PADDLE_DOWNLOAD_DIR}/third_party ${PADDLE_INSTALL_DIR}/third_party &&
+        ${CMAKE_COMMAND} -E copy ${PADDLE_INSTALL_DIR}/third_party/install/mkldnn/lib/libmkldnn.so.0 ${PADDLE_INSTALL_DIR}/third_party/install/mkldnn/lib/libmkldnn.so 
+)
+else()
+ExternalProject_Add(
+    "extern_paddle"
+    ${EXTERNAL_PROJECT_LOG_ARGS}
+    URL                 "${PADDLE_LIB_PATH}"
+    PREFIX              "${PADDLE_SOURCES_DIR}"
+    DOWNLOAD_DIR        "${PADDLE_DOWNLOAD_DIR}"
+    CONFIGURE_COMMAND   ""
+    BUILD_COMMAND       ""
+    UPDATE_COMMAND      ""
+    INSTALL_COMMAND
+        ${CMAKE_COMMAND} -E copy_directory ${PADDLE_DOWNLOAD_DIR}/paddle/include ${PADDLE_INSTALL_DIR}/include &&
+        ${CMAKE_COMMAND} -E copy_directory ${PADDLE_DOWNLOAD_DIR}/paddle/lib ${PADDLE_INSTALL_DIR}/lib &&
+        ${CMAKE_COMMAND} -E copy_directory ${PADDLE_DOWNLOAD_DIR}/third_party ${PADDLE_INSTALL_DIR}/third_party
 )
-
-if (WITH_MKLML)
-   file(COPY ${PADDLE_INSTALL_DIR}/third_party/install/mkldnn/lib/libmkldnn.so.0 DESTINATION ${PADDLE_INSTALL_DIR}/third_party/install/mkldnn/lib/libmkldnn.so FOLLOW_SYMLINK_CHAIN)
 endif()
 
 INCLUDE_DIRECTORIES(${PADDLE_INCLUDE_DIR})

core/configure/CMakeLists.txt

@@ -55,6 +55,7 @@ if (NOT CLIENT_ONLY)
 py_proto_compile(server_config_py_proto SRCS proto/server_configure.proto)
 add_custom_target(server_config_py_proto_init ALL COMMAND ${CMAKE_COMMAND} -E touch __init__.py)
 add_dependencies(server_config_py_proto server_config_py_proto_init)
+if (NOT WITH_GPU)
 add_custom_command(TARGET server_config_py_proto POST_BUILD
 		COMMAND ${CMAKE_COMMAND} -E make_directory ${PADDLE_SERVING_BINARY_DIR}/python/paddle_serving_server/proto
 		COMMAND cp *.py ${PADDLE_SERVING_BINARY_DIR}/python/paddle_serving_server/proto
@@ -66,6 +67,24 @@ add_custom_command(TARGET general_model_config_py_proto POST_BUILD
 		COMMAND cp *.py ${PADDLE_SERVING_BINARY_DIR}/python/paddle_serving_server/proto
 		COMMENT "Copy generated general_model_config proto file into directory paddle_serving_server/proto."
 		WORKING_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR})
+else()
+add_custom_command(TARGET server_config_py_proto POST_BUILD
+		COMMAND ${CMAKE_COMMAND} -E make_directory
+        ${PADDLE_SERVING_BINARY_DIR}/python/paddle_serving_server_gpu/proto
+		COMMAND cp *.py
+        ${PADDLE_SERVING_BINARY_DIR}/python/paddle_serving_server_gpu/proto
+		COMMENT "Copy generated python proto into directory
+        paddle_serving_server_gpu/proto."
+		WORKING_DIRECTORY ${CMAKE_CURRENT_BINRARY_DIR})
 
+add_custom_command(TARGET general_model_config_py_proto POST_BUILD
+		COMMAND ${CMAKE_COMMAND} -E make_directory
+        ${PADDLE_SERVING_BINARY_DIR}/python/paddle_serving_server_gpu/proto
+		COMMAND cp *.py
+        ${PADDLE_SERVING_BINARY_DIR}/python/paddle_serving_server_gpu/proto
+		COMMENT "Copy generated general_model_config proto file into directory
+        paddle_serving_server_gpu/proto."
+		WORKING_DIRECTORY ${CMAKE_CURRENT_BINARY_DIR})
+endif()
 endif()
 

core/general-server/CMakeLists.txt

@@ -31,7 +31,7 @@ if(WITH_GPU)
     target_link_libraries(serving ${CUDA_LIBRARIES})
 endif()
 
-if(WITH_MKL)
+if(WITH_MKL OR WITH_GPU)
     target_link_libraries(serving -liomp5 -lmklml_intel -lmkldnn -lpthread -lcrypto -lm -lrt -lssl -ldl -lz -lbz2)
 else()
     target_link_libraries(serving openblas -lpthread -lcrypto -lm -lrt -lssl -ldl -lz -lbz2)

python/paddle_serving_server/__init__.py

@@ -16,24 +16,32 @@ import os
 from .proto import server_configure_pb2 as server_sdk
 from .proto import general_model_config_pb2 as m_config
 import google.protobuf.text_format
+import tarfile
+import paddle_serving_server as paddle_serving_server
+from version import serving_server_version
+
 
 class OpMaker(object):
     def __init__(self):
-        self.op_dict = {"general_infer":"GeneralInferOp",
-                        "general_reader":"GeneralReaderOp",
-                        "general_single_kv":"GeneralSingleKVOp",
-                        "general_dist_kv":"GeneralDistKVOp"}
+        self.op_dict = {
+            "general_infer": "GeneralInferOp",
+            "general_reader": "GeneralReaderOp",
+            "general_single_kv": "GeneralSingleKVOp",
+            "general_dist_kv": "GeneralDistKVOp"
+        }
 
     # currently, inputs and outputs are not used
     # when we have OpGraphMaker, inputs and outputs are necessary
     def create(self, name, inputs=[], outputs=[]):
         if name not in self.op_dict:
-            raise Exception("Op name {} is not supported right now".format(name))
+            raise Exception("Op name {} is not supported right now".format(
+                name))
         node = server_sdk.DAGNode()
         node.name = "{}_op".format(name)
         node.type = self.op_dict[name]
         return node
 
+
 class OpSeqMaker(object):
     def __init__(self):
         self.workflow = server_sdk.Workflow()
@@ -53,6 +61,7 @@ class OpSeqMaker(object):
         workflow_conf.workflows.extend([self.workflow])
         return workflow_conf
 
+
 class Server(object):
     def __init__(self):
         self.server_handle_ = None
@@ -72,6 +81,8 @@ class Server(object):
         self.num_threads = 0
         self.port = 8080
         self.reload_interval_s = 10
+        self.module_path = os.path.dirname(paddle_serving_server.__file__)
+        self.cur_path = os.getcwd()
 
     def set_max_concurrency(self, concurrency):
         self.max_concurrency = concurrency
@@ -129,7 +140,8 @@ class Server(object):
 
     def _prepare_resource(self, workdir):
         if self.resource_conf == None:
-            with open("{}/{}".format(workdir, self.general_model_config_fn), "w") as fout:
+            with open("{}/{}".format(workdir, self.general_model_config_fn),
+                      "w") as fout:
                 fout.write(str(self.model_conf))
             self.resource_conf = server_sdk.ResourceConf()
             self.resource_conf.model_toolkit_path = workdir
@@ -150,6 +162,54 @@ class Server(object):
         # check config here
         # print config here
 
+    def get_device_version(self):
+        avx_flag = False
+        mkl_flag = False
+        openblas_flag = False
+        r = os.system("cat /proc/cpuinfo | grep avx > /dev/null 2>&1")
+        if r == 0:
+            avx_flag = True
+        r = os.system("which mkl")
+        if r == 0:
+            mkl_flag = True
+        if avx_flag:
+            if mkl_flag:
+                device_version = "serving-cpu-avx-mkl-"
+            else:
+                device_version = "serving-cpu-avx-openblas-"
+        else:
+            device_version = "serving-cpu-noavx-openblas-"
+        return device_version
+
+    def download_bin(self):
+        os.chdir(self.module_path)
+        need_download = False
+        device_version = self.get_device_version()
+        floder_name = device_version + serving_server_version
+        tar_name = floder_name + ".tar.gz"
+        bin_url = "https://paddle-serving.bj.bcebos.com/bin/" + tar_name
+        self.server_path = os.path.join(self.module_path, floder_name)
+        if not os.path.exists(self.server_path):
+            print('Frist time run, downloading PaddleServing components ...')
+            r = os.system('wget ' + bin_url + ' --no-check-certificate')
+            if r != 0:
+                print('Download failed')
+                if os.path.exists(tar_name):
+                    os.remove(tar_name)
+            else:
+                try:
+                    print('Decompressing files ..')
+                    tar = tarfile.open(tar_name)
+                    tar.extractall()
+                    tar.close()
+                except:
+                    if os.path.exists(exe_path):
+                        os.remove(exe_path)
+                finally:
+                    os.remove(tar_name)
+        os.chdir(self.cur_path)
+        self.bin_path = self.server_path + "/serving"
+
     def prepare_server(self, workdir=None, port=9292, device="cpu"):
         if workdir == None:
             workdir = "./tmp"
@@ -176,8 +236,9 @@ class Server(object):
     def run_server(self):
         # just run server with system command
         # currently we do not load cube
-        command = "/home/xulongteng/github/Serving/build_server/core/general-server/serving" \
-                  " -enable_model_toolkit " \
+        self.download_bin()
+        command = "{} " \
+                  "-enable_model_toolkit " \
                   "-inferservice_path {} " \
                   "-inferservice_file {} " \
                   "-max_concurrency {} " \
@@ -189,6 +250,7 @@ class Server(object):
                   "-workflow_path {} " \
                   "-workflow_file {} " \
                   "-bthread_concurrency {} ".format(
+                      self.bin_path,
                       self.workdir,
                       self.infer_service_fn,
                       self.max_concurrency,
@@ -201,5 +263,3 @@ class Server(object):
                       self.workflow_fn,
                       self.num_threads,)
         os.system(command)
-
-