milvus/cpp
yu yunfeng 52e23c4760 merge 0.3.0
Former-commit-id: 9e30f64b77a970d48e627bd5c518a7c17f498fd5
2019-06-14 09:59:28 +08:00
..
cmake fix faiss ptx issue; change cuda gencode 2019-06-13 13:20:06 +08:00
conf MS-82 & MS-83 Update vecwise to Milvus 2019-06-13 16:02:25 +08:00
scripts MS-82 & MS-83 Update vecwise to Milvus 2019-06-13 16:02:25 +08:00
src merge 0.3.0 2019-06-14 09:59:28 +08:00
thirdparty cmake: change faiss version to 1.5.2; add CUDA gencode 2019-06-12 11:27:50 +08:00
unittest merge 0.3.0 2019-06-14 09:59:28 +08:00
.gitignore automaticlly generate build version infor 2019-06-06 10:24:40 +08:00
CHANGELOG.md Merge remote-tracking branch 'upstream/branch-0.3.0' into branch-0.3.0 2019-06-13 16:04:04 +08:00
CMakeLists.txt fix faiss ptx issue; change cuda gencode 2019-06-13 13:20:06 +08:00
LICENSE.txt Add LICENSE.txt and RELEASE.md 2019-04-14 20:12:34 +08:00
README.md Change README.md in cpp 2019-06-12 11:05:33 +08:00
RELEASE.md Add LICENSE.txt and RELEASE.md 2019-04-14 20:12:34 +08:00
build.sh Build: make -j 4 2019-05-30 09:20:47 +08:00
start_server.sh refine code 2019-04-18 17:26:03 +08:00
stop_server.sh refine code 2019-04-18 17:26:03 +08:00
version.h.macro automaticlly generate build version infor 2019-06-06 10:45:07 +08:00

README.md

Compilation

Step 1: install necessery tools

centos7 : 
    yum install gfortran flex bison
    
ubuntu16.04 : 
    sudo apt-get install gfortran flex bison

Step 2: build(output to cmake_build folder)

cmake_build/src/vecwise_server is the server

cmake_build/src/libvecwise_engine.a is the static library

cd [sourcecode path]/cpp
./build.sh -t Debug
./build.sh -t Release
./build.sh -g # Build GPU version

If you encounter the following error when building: protocol https not supported or disabled in libcurl

  1. Install libcurl4-openssl-dev

  2. Install cmake 3.14:

    ./bootstrap --system-curl 
    make 
    sudo make install
    

To build unittest:

./build.sh -u
or
./build.sh --unittest

Launch server

Set config in cpp/conf/server_config.yaml

Then launch server with config: cd [build output path] start_server.sh stop_server.sh

Launch test_client(only for debug)

If you want to test remote api, you can build test_client. test_client use same config file with server:

cd [build output path]/test_client
test_client -c [sourcecode path]/cpp/conf/server_config.yaml

License Generate

Use get_sys_info to get system info file.

./get_sys_info                  # system.info will be generated

Use license_generator to generate license file.

./license_generator -s system.info -l system.license -b 2019-05-15 -e 2019-08-14                 

Copy the license file to path assigned by license_path in server config file.