cann: update image

Author: wangshuai09

.devops/llama-cli-cann.Dockerfile

@@ -7,8 +7,8 @@ WORKDIR /app
 COPY . .
 
 RUN yum install -y gcc g++ cmake make
-ENV LIBRARY_PATH=/usr/local/Ascend/ascend-toolkit/latest/lib64:$LIBRARY_PATH
 ENV ASCEND_TOOLKIT_HOME=/usr/local/Ascend/ascend-toolkit/latest
+ENV LIBRARY_PATH=${ASCEND_TOOLKIT_HOME}/lib64:$LIBRARY_PATH
 ENV LD_LIBRARY_PATH=${ASCEND_TOOLKIT_HOME}/lib64:${ASCEND_TOOLKIT_HOME}/lib64/plugin/opskernel:${ASCEND_TOOLKIT_HOME}/lib64/plugin/nnengine:${ASCEND_TOOLKIT_HOME}/opp/built-in/op_impl/ai_core/tbe/op_tiling:${LD_LIBRARY_PATH}
 ENV PYTHONPATH=${ASCEND_TOOLKIT_HOME}/python/site-packages:${ASCEND_TOOLKIT_HOME}/opp/built-in/op_impl/ai_core/tbe:${PYTHONPATH}
 ENV PATH=${ASCEND_TOOLKIT_HOME}/bin:${ASCEND_TOOLKIT_HOME}/compiler/ccec_compiler/bin:${PATH}
@@ -17,17 +17,28 @@ ENV ASCEND_OPP_PATH=${ASCEND_TOOLKIT_HOME}/opp
 ENV TOOLCHAIN_HOME=${ASCEND_TOOLKIT_HOME}/toolkit
 ENV ASCEND_HOME_PATH=${ASCEND_TOOLKIT_HOME}
 
+# find libascend_hal.so, because the drive hasn`t been mounted.
+ENV LD_LIBRARY_PATH=${ASCEND_TOOLKIT_HOME}/runtime/lib64/stub:$LD_LIBRARY_PATH
 
 RUN echo "Building with static libs" && \
-    # source /usr/local/Ascend/ascend-toolkit/set_env.sh && \
-    cmake -B build -DGGML_CANN=ON -DBUILD_SHARED_LIBS=OFF -DCMAKE_LIBRARY_PATH=${ASCEND_TOOLKIT_HOME}\lib64 && \
+    source /usr/local/Ascend/ascend-toolkit/set_env.sh --force && \
+    cmake -B build -DGGML_CANN=ON -DBUILD_SHARED_LIBS=OFF  && \
     cmake --build build --config Release --target llama-cli
 
 # TODO: use image with NNRT
 FROM cosdt/cann:$ASCEND_VERSION AS runtime
-
 COPY --from=build /app/build/bin/llama-cli /llama-cli
 
 ENV LC_ALL=C.utf8
 
-ENTRYPOINT [ "/llama-cli" ]
+ENV ASCEND_TOOLKIT_HOME=/usr/local/Ascend/ascend-toolkit/latest
+ENV LIBRARY_PATH=${ASCEND_TOOLKIT_HOME}/lib64:$LIBRARY_PATH
+ENV LD_LIBRARY_PATH=${ASCEND_TOOLKIT_HOME}/lib64:${ASCEND_TOOLKIT_HOME}/lib64/plugin/opskernel:${ASCEND_TOOLKIT_HOME}/lib64/plugin/nnengine:${ASCEND_TOOLKIT_HOME}/opp/built-in/op_impl/ai_core/tbe/op_tiling:${LD_LIBRARY_PATH}
+ENV PYTHONPATH=${ASCEND_TOOLKIT_HOME}/python/site-packages:${ASCEND_TOOLKIT_HOME}/opp/built-in/op_impl/ai_core/tbe:${PYTHONPATH}
+ENV PATH=${ASCEND_TOOLKIT_HOME}/bin:${ASCEND_TOOLKIT_HOME}/compiler/ccec_compiler/bin:${PATH}
+ENV ASCEND_AICPU_PATH=${ASCEND_TOOLKIT_HOME}
+ENV ASCEND_OPP_PATH=${ASCEND_TOOLKIT_HOME}/opp
+ENV TOOLCHAIN_HOME=${ASCEND_TOOLKIT_HOME}/toolkit
+ENV ASCEND_HOME_PATH=${ASCEND_TOOLKIT_HOME}
+
+ENTRYPOINT ["/llama-cli" ]

docs/backend/CANN.md

@@ -120,12 +120,11 @@ The llama.cpp CANN backend is designed to support Ascend NPU. It utilize the abi
 
 ## Docker
 
-### Get Images
-You can get a pre-build image at cosdt/cann:8.0.rc2.alpha003-910b-openeuler22.03-py3.8-llama.cpp and use llama-cli directly without building llama.cpp in this image.
+### Build Images
+You can get a image with llama.cpp in one command.
 ```sh
-docker pull cosdt/cann:8.0.rc2.alpha003-910b-ubuntu22.04-py3.8-llama.cpp
+docker build -t llama-cpp-cann -f .devops/llama-cli-cann.Dockerfile .
 ```
-!!!!!! Add content for build image or get pre-build image.
 
 ### Run container
 
@@ -134,8 +133,8 @@ docker pull cosdt/cann:8.0.rc2.alpha003-910b-ubuntu22.04-py3.8-llama.cpp
 npu-smi info
 
 # Select the cards that you want to use, make sure these cards are not used by someone.
-# Following using cards of device0 and device1.
-docker run --name llamacpp --device /dev/davinci0  --device /dev/davinci1 --device /dev/davinci_manager --device /dev/devmm_svm --device /dev/hisi_hdc -v /usr/local/dcmi:/usr/local/dcmi -v /usr/local/bin/npu-smi:/usr/local/bin/npu-smi -v /usr/local/Ascend/driver/lib64/:/usr/local/Ascend/driver/lib64/ -v /usr/local/Ascend/driver/version.info:/usr/local/Ascend/driver/version.info -v /PATH_TO_YOUR_MODELS/:/app/models -itd cosdt/cann:8.0.rc2.alpha003-910b-ubuntu22.04-py3.8-llama.cpp -m /app/models/MODEL_PATH -ngl 32 -p "Building a website can be done in 10 simple steps:"
+# Following using cards of device0.
+docker run --name llamacpp --device /dev/davinci0  --device /dev/davinci_manager --device /dev/devmm_svm --device /dev/hisi_hdc -v /usr/local/dcmi:/usr/local/dcmi -v /usr/local/bin/npu-smi:/usr/local/bin/npu-smi -v /usr/local/Ascend/driver/lib64/:/usr/local/Ascend/driver/lib64/ -v /usr/local/Ascend/driver/version.info:/usr/local/Ascend/driver/version.info -v /PATH_TO_YOUR_MODELS/:/app/models -it llama-cpp-cann -m /app/models/MODEL_PATH -ngl 32 -p "Building a website can be done in 10 simple steps:"
 ```
 
 *Notes:*
@@ -234,8 +233,6 @@ cmake --build build --config release
     - Single device: Use one device target specified by the user.
     - Multiple devices: Automatically choose the devices with the same backend.
 
-    In two device selection modes, the default SYCL backend is level_zero, you can choose other backend supported by SYCL by setting environment variable ONEAPI_DEVICE_SELECTOR.
-
     | Device selection | Parameter                              |
     |:----------------:|:--------------------------------------:|
     | Single device    | --split-mode none --main-gpu DEVICE_ID |

docs/build.md

@@ -365,7 +365,7 @@ cmake -B build -DGGML_CANN=on -DCMAKE_BUILD_TYPE=release
 cmake --build build --config release
 ```
 
-`llama.cpp based on CANN` does not currently supprot quantization. You can test with:
+You can test with:
 
 `./build/llama-cli -m PATH_TO_MODEL -p "Building a website can be done in 10 steps:" -ngl 32`