You signed in with another tab or window. Reload to refresh your session.You signed out in another tab or window. Reload to refresh your session.You switched accounts on another tab or window. Reload to refresh your session.Dismiss alert
When built with CLANG complier for Android with git command and cmake commands and executing benchmark_genai, the following error is reported.
Model used: Almost all models report this issue.
Error output:
Error sample 1:
134|base_aaos:/data/local/tmp/npu # ./benchmark_genai -m ../../npu/ov-models/openla/open_llama_7b_v2-int4-ov/ -d GPU -n2
Exception from src/inference/src/cpp/core.cpp:112:
Exception from src/inference/src/dev/plugin.cpp:53:
Exception from src/plugins/intel_cpu/src/dnnl_extension_utils.cpp:84:
CPU plugin does not support i64 for use with oneDNN.
Error Sample 2:
base_aaos:/data/local/tmp/npu # ./benchmark_genai -m ../../npu/ov-models/mistral-7b-instruct-v0.1-int4-ov/ -d GPU -n2
Exception from src/inference/src/cpp/core.cpp:112:
Exception from src/inference/src/dev/plugin.cpp:53:
Exception from src/plugins/intel_cpu/src/dnnl_extension_utils.cpp:84:
CPU plugin does not support i64 for use with oneDNN.
Workaround:
For this error and below patch causes to fix the issue and benchmark_genai starts to execute and run:
I understand this is not correct data type to be used. Please provide an appropriate resolution.
celadon@biaas-d119:~/workspace/openvino-android/openvino/src/plugins/intel_cpu
# git diff
diff --git a/src/plugins/intel_cpu/src/dnnl_extension_utils.cpp b/src/plugins/intel_cpu/src/dnnl_extension_utils.cpp
index 1c5598b6d5..3a723b8da0 100644
--- a/src/plugins/intel_cpu/src/dnnl_extension_utils.cpp
+++ b/src/plugins/intel_cpu/src/dnnl_extension_utils.cpp
@@ -80,6 +80,8 @@ dnnl::memory::data_type DnnlExtensionUtils::ElementTypeToDataType(const ov::elem
return memory::data_type::f4_e2m1;
case ov::element::undefined:
return memory::data_type::undef;
+ case ov::element::i64:
+ return memory::data_type::f32;
default: {
OPENVINO_THROW("CPU plugin does not support ", elementType.to_string(), " for use with oneDNN.");
}
The text was updated successfully, but these errors were encountered:
Openvino version: Master branch
Openvino genai: Master branch
When built with CLANG complier for Android with git command and cmake commands and executing benchmark_genai, the following error is reported.
Model used: Almost all models report this issue.
Error output:
Error sample 1:
134|base_aaos:/data/local/tmp/npu # ./benchmark_genai -m ../../npu/ov-models/openla/open_llama_7b_v2-int4-ov/ -d GPU -n2
Exception from src/inference/src/cpp/core.cpp:112:
Exception from src/inference/src/dev/plugin.cpp:53:
Exception from src/plugins/intel_cpu/src/dnnl_extension_utils.cpp:84:
CPU plugin does not support i64 for use with oneDNN.
Error Sample 2:
base_aaos:/data/local/tmp/npu # ./benchmark_genai -m ../../npu/ov-models/mistral-7b-instruct-v0.1-int4-ov/ -d GPU -n2
Exception from src/inference/src/cpp/core.cpp:112:
Exception from src/inference/src/dev/plugin.cpp:53:
Exception from src/plugins/intel_cpu/src/dnnl_extension_utils.cpp:84:
CPU plugin does not support i64 for use with oneDNN.
Git clone and build command:
https://github.com/openvinotoolkit/openvino/blob/master/docs/dev/build_android.md
This is my source tree structure in local machine:
CMAKE command: cmake -S $OPV_HOME_DIR/openvino -B $OPV_HOME_DIR/openvino-androidbuild -DCMAKE_INSTALL_PREFIX=$OPV_HOME_DIR/openvino-androidinstall -DCMAKE_TOOLCHAIN_FILE=$CURRENT_CMAKE_TOOLCHAIN_FILE -DCMAKE_BUILD_TYPE=Debug -DANDROID_ABI=$CURRENT_ANDROID_ABI -DANDROID_PLATFORM=$CURRENT_ANDROID_PLATFORM -DANDROID_STL=$CURRENT_ANDROID_STL -DTBB_DIR=$OPV_HOME_DIR/one-tbb-androidinstall/lib/cmake/TBB -DCMAKE_CXX_FLAGS="-fsanitize=address -fno-omit-frame-pointer" -DOPENVINO_EXTRA_MODULES=$OPV_HOME_DIR/openvino.genai -DCPACK_ARCHIVE_COMPONENT_INSTALL=OFF -DENABLE_INTEL_NPU=OFF
Build Command:
cmake --build $OPV_HOME_DIR/openvino-androidbuild --target package -j
Workaround:
For this error and below patch causes to fix the issue and benchmark_genai starts to execute and run:
I understand this is not correct data type to be used. Please provide an appropriate resolution.
The text was updated successfully, but these errors were encountered: