[CANN] Compile bug: no matching function for call to 'CastIntrinsicsImpl' Ascend NPU issues specific to Ascend NPUs

[Cikaros]

I encountered the same issue(#10556 ) in Ascend310B1 as well.

root@orangepiaipro-20t:/data/llama.cpp# cmake -B build -DGGML_CANN=on -DCMAKE_BUILD_TYPE=release
-- Warning: ccache not found - consider installing it for faster compilation or disable this warning with GGML_CCACHE=OFF
-- CMAKE_SYSTEM_PROCESSOR: aarch64
-- Including CPU backend
-- ARM detected
-- ARM feature FMA enabled
-- Adding CPU backend variant ggml-cpu:  
-- CANN: updated CANN_INSTALL_DIR from ASCEND_TOOLKIT_HOME=/usr/local/Ascend/ascend-toolkit/latest
-- CANN: SOC_VERSION auto-detected is:Ascend310B1
-- CANN: compile ascend kernels witch SOC_TYPE:Ascend310B1, SOC_VERSION:ascend310b1, compile macro:-DASCEND_310B.
-- CANN: CANN_INCLUDE_DIRS =  /usr/local/Ascend/ascend-toolkit/latest/include;/usr/local/Ascend/ascend-toolkit/latest/include/aclnn;/usr/local/Ascend/ascend-toolkit/latest/acllib/include
-- CANN: CANN_LIBRARIES =  ascendcl;nnopbase;opapi;acl_op_compiler;ascendc_kernels
-- Including CANN backend
-- Configuring done
-- Generating done
-- Build files have been written to: /data/llama.cpp/build

root@orangepiaipro-20t:/data/llama.cpp# cmake --build build --config release
Consolidate compiler generated dependencies of target ggml-base
[  3%] Built target ggml-base
[  3%] Performing build step for 'ascendc_kernels_precompile'
Consolidate compiler generated dependencies of target precompile_obj
[ 12%] Building CXX object CMakeFiles/precompile_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/dup.cpp.o
[ 25%] Building CXX object CMakeFiles/precompile_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_f16.cpp.o
[ 37%] Building CXX object CMakeFiles/precompile_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_f32.cpp.o
[ 50%] Building CXX object CMakeFiles/precompile_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_q4_0.cpp.o
[ 62%] Building CXX object CMakeFiles/precompile_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_q8_0.cpp.o
[ 75%] Building CXX object CMakeFiles/precompile_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/quantize_f16_q8_0.cpp.o
[ 87%] Building CXX object CMakeFiles/precompile_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/quantize_f32_q8_0.cpp.o
[100%] Building CXX object CMakeFiles/precompile_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/quantize_float_to_q4_0.cpp.o
[100%] Built target precompile_obj
[100%] Built target check_src_template
[  4%] No install step for 'ascendc_kernels_precompile'
[  4%] Completed 'ascendc_kernels_precompile'
[  6%] Built target ascendc_kernels_precompile
[  7%] Performing build step for 'ascendc_kernels_preprocess'
-- Configuring done
-- Generating done
-- Build files have been written to: /data/llama.cpp/build/ggml/src/ggml-cann/kernels/ascendc_kernels_preprocess-prefix/src/ascendc_kernels_preprocess-build
Consolidate compiler generated dependencies of target preprocess_obj
[  6%] Building CXX object CMakeFiles/preprocess_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/dup.cpp.o
[ 12%] Building CXX object CMakeFiles/preprocess_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_f16.cpp.o
[ 18%] Building CXX object CMakeFiles/preprocess_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_f32.cpp.o
[ 25%] Building CXX object CMakeFiles/preprocess_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_q4_0.cpp.o
[ 31%] Building CXX object CMakeFiles/preprocess_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_q8_0.cpp.o
[ 37%] Building CXX object CMakeFiles/preprocess_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/quantize_f16_q8_0.cpp.o
[ 43%] Building CXX object CMakeFiles/preprocess_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/quantize_f32_q8_0.cpp.o
[ 50%] Building CXX object CMakeFiles/preprocess_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/quantize_float_to_q4_0.cpp.o
[ 50%] Built target preprocess_obj
Consolidate compiler generated dependencies of target m200_obj
[ 56%] Building CXX object CMakeFiles/m200_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/dup.cpp.o
In file included from /data/llama.cpp/ggml/src/ggml-cann/kernels/dup.cpp:1:
In file included from /usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/kernel_operator.h:28:
In file included from /usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/inner_interface/inner_kernel_operator_intf.h:24:
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/inner_interface/inner_kernel_operator_data_copy_intf.cppm:947:9: warning: 'DataCopyPadUB2GMImpl<half>' is deprecated: NOTICE: DataCopyPad is not deprecated. Currently, DataCopyPad is an unsupported API on current device. Please check your code! [-Wdeprecated-declarations]
        DataCopyPadUB2GMImpl((__gm__ PrimType*)dstGlobal.GetPhyAddr(), (__ubuf__ PrimType*)srcLocal.GetPhyAddr(),
        ^
/data/llama.cpp/ggml/src/ggml-cann/kernels/dup.cpp:83:9: note: in instantiation of function template specialization 'AscendC::DataCopyPad<half>' requested here
        DataCopyPad(dst_gm, dst_local, dataCopyParams);
        ^
/data/llama.cpp/ggml/src/ggml-cann/kernels/dup.cpp:101:9: note: in instantiation of member function 'DupByRows<half, half>::copy_out' requested here
        copy_out();
        ^
/data/llama.cpp/ggml/src/ggml-cann/kernels/dup.cpp:166:8: note: in instantiation of member function 'DupByRows<half, half>::dup' requested here
    op.dup();
       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_data_copy_impl.h:1187:3: note: 'DataCopyPadUB2GMImpl<half>' has been explicitly marked deprecated here
[[deprecated("NOTICE: DataCopyPad is not deprecated. Currently, DataCopyPad is an unsupported API on current device. "
  ^
In file included from /data/llama.cpp/ggml/src/ggml-cann/kernels/dup.cpp:1:
In file included from /usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/kernel_operator.h:28:
In file included from /usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/inner_interface/inner_kernel_operator_intf.h:24:
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/inner_interface/inner_kernel_operator_data_copy_intf.cppm:947:9: warning: 'DataCopyPadUB2GMImpl<float>' is deprecated: NOTICE: DataCopyPad is not deprecated. Currently, DataCopyPad is an unsupported API on current device. Please check your code! [-Wdeprecated-declarations]
        DataCopyPadUB2GMImpl((__gm__ PrimType*)dstGlobal.GetPhyAddr(), (__ubuf__ PrimType*)srcLocal.GetPhyAddr(),
        ^
/data/llama.cpp/ggml/src/ggml-cann/kernels/dup.cpp:83:9: note: in instantiation of function template specialization 'AscendC::DataCopyPad<float>' requested here
        DataCopyPad(dst_gm, dst_local, dataCopyParams);
        ^
/data/llama.cpp/ggml/src/ggml-cann/kernels/dup.cpp:101:9: note: in instantiation of member function 'DupByRows<float, float>::copy_out' requested here
        copy_out();
        ^
/data/llama.cpp/ggml/src/ggml-cann/kernels/dup.cpp:188:8: note: in instantiation of member function 'DupByRows<float, float>::dup' requested here
    op.dup();
       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_data_copy_impl.h:1187:3: note: 'DataCopyPadUB2GMImpl<float>' has been explicitly marked deprecated here
[[deprecated("NOTICE: DataCopyPad is not deprecated. Currently, DataCopyPad is an unsupported API on current device. "
  ^
2 warnings generated.
[ 62%] Building CXX object CMakeFiles/m200_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_f16.cpp.o
In file included from /data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_f16.cpp:1:
In file included from /usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/kernel_operator.h:28:
In file included from /usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/inner_interface/inner_kernel_operator_intf.h:24:
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/inner_interface/inner_kernel_operator_data_copy_intf.cppm:947:9: warning: 'DataCopyPadUB2GMImpl<float>' is deprecated: NOTICE: DataCopyPad is not deprecated. Currently, DataCopyPad is an unsupported API on current device. Please check your code! [-Wdeprecated-declarations]
        DataCopyPadUB2GMImpl((__gm__ PrimType*)dstGlobal.GetPhyAddr(), (__ubuf__ PrimType*)srcLocal.GetPhyAddr(),
        ^
/data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_f16.cpp:95:13: note: in instantiation of function template specialization 'AscendC::DataCopyPad<float>' requested here
            DataCopyPad(output_gm[offset + len], output_local[len],
            ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_data_copy_impl.h:1187:3: note: 'DataCopyPadUB2GMImpl<float>' has been explicitly marked deprecated here
[[deprecated("NOTICE: DataCopyPad is not deprecated. Currently, DataCopyPad is an unsupported API on current device. "
  ^
1 warning generated.
[ 68%] Building CXX object CMakeFiles/m200_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_f32.cpp.o
In file included from /data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_f32.cpp:1:
In file included from /usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/kernel_operator.h:28:
In file included from /usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/inner_interface/inner_kernel_operator_intf.h:24:
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/inner_interface/inner_kernel_operator_data_copy_intf.cppm:947:9: warning: 'DataCopyPadUB2GMImpl<float>' is deprecated: NOTICE: DataCopyPad is not deprecated. Currently, DataCopyPad is an unsupported API on current device. Please check your code! [-Wdeprecated-declarations]
        DataCopyPadUB2GMImpl((__gm__ PrimType*)dstGlobal.GetPhyAddr(), (__ubuf__ PrimType*)srcLocal.GetPhyAddr(),
        ^
/data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_f32.cpp:89:13: note: in instantiation of function template specialization 'AscendC::DataCopyPad<float>' requested here
            DataCopyPad(output_gm[offset + len], output_local[len],
            ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_data_copy_impl.h:1187:3: note: 'DataCopyPadUB2GMImpl<float>' has been explicitly marked deprecated here
[[deprecated("NOTICE: DataCopyPad is not deprecated. Currently, DataCopyPad is an unsupported API on current device. "
  ^
1 warning generated.
[ 75%] Building CXX object CMakeFiles/m200_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_q4_0.cpp.o
In file included from /data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_q4_0.cpp:1:
In file included from /usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/kernel_operator.h:27:
In file included from /usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/interface/kernel_operator_intf.h:48:
In file included from /usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/interface/kernel_operator_vec_vconv_intf.h:28:
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:455:5: error: no matching function for call to 'CastIntrinsicsImpl'
    CastIntrinsicsImpl(dst, src, roundMode, 1, repeatParams);
    ^~~~~~~~~~~~~~~~~~
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/inner_interface/inner_kernel_operator_vec_vconv_intf.cppm:128:5: note: in instantiation of function template specialization 'AscendC::CastImpl<half, AscendC::IntegerSubType<4, true>>' requested here
    CastImpl((__ubuf__ T1*)dstLocal.GetPhyAddr(), (__ubuf__ T2*)srcLocal.GetPhyAddr(), round_mode, calCount);
    ^
/data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_q4_0.cpp:122:9: note: in instantiation of function template specialization 'AscendC::Cast<half, AscendC::IntegerSubType<4, true>>' requested here
        Cast(cast_local, input_local, RoundMode::CAST_NONE, QK4_0);
        ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:26:24: note: candidate function not viable: no known conversion from '__ubuf__ AscendC::IntegerSubType<4, true> *' to '__ubuf__ int32_t *' (aka '__ubuf__ int *') for 2nd argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ half* dst, __ubuf__ int32_t* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:33:24: note: candidate function not viable: no known conversion from '__ubuf__ AscendC::IntegerSubType<4, true> *' to '__ubuf__ int8_t *' (aka '__ubuf__ signed char *') for 2nd argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ half* dst, __ubuf__ int8_t* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:47:24: note: candidate function not viable: no known conversion from '__ubuf__ AscendC::IntegerSubType<4, true> *' to '__ubuf__ uint8_t *' (aka '__ubuf__ unsigned char *') for 2nd argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ half* dst, __ubuf__ uint8_t* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:215:24: note: candidate function not viable: no known conversion from '__ubuf__ AscendC::IntegerSubType<4, true> *' to '__ubuf__ float *' for 2nd argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ half* dst, __ubuf__ float* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:334:24: note: candidate function not viable: no known conversion from '__ubuf__ AscendC::IntegerSubType<4, true> *' to '__ubuf__ int16_t *' (aka '__ubuf__ short *') for 2nd argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ half* dst, __ubuf__ int16_t* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:61:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ float *' for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ float* dst, __ubuf__ int32_t* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:75:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ float *' for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ float* dst, __ubuf__ half* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:89:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ int32_t *' (aka '__ubuf__ int *') for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ int32_t* dst, __ubuf__ half* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:131:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ int8_t *' (aka '__ubuf__ signed char *') for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ int8_t* dst, __ubuf__ half* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:173:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ uint8_t *' (aka '__ubuf__ unsigned char *') for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ uint8_t* dst, __ubuf__ half* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:263:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ int32_t *' (aka '__ubuf__ int *') for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ int32_t* dst, __ubuf__ float* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:305:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ int16_t *' (aka '__ubuf__ short *') for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ int16_t* dst, __ubuf__ half* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:322:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ uint8_t *' (aka '__ubuf__ unsigned char *') for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ uint8_t* dst, __ubuf__ int16_t* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:328:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ int8_t *' (aka '__ubuf__ signed char *') for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ int8_t* dst, __ubuf__ int16_t* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:351:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ float *' for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ float* dst, __ubuf__ float* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:357:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ int64_t *' (aka '__ubuf__ long *') for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ int64_t* dst, __ubuf__ float* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:363:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ int16_t *' (aka '__ubuf__ short *') for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ int16_t* dst, __ubuf__ float* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:369:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ float *' for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ float* dst, __ubuf__ int16_t* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:375:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ int64_t *' (aka '__ubuf__ long *') for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ int64_t* dst, __ubuf__ int32_t* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:381:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ float *' for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ float* dst, __ubuf__ int64_t* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:387:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ int32_t *' (aka '__ubuf__ int *') for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ int32_t* dst, __ubuf__ int64_t* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:393:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ AscendC::int4b_t *' (aka '__ubuf__ IntegerSubType<INT4_BIT_NUM, true> *') for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ int4b_t* dst, __ubuf__ half* src, const RoundMode& roundMode,
                       ^
/usr/local/Ascend/ascend-toolkit/latest/tools/tikcpp/tikcfw/impl/dav_m200/kernel_operator_vec_vconv_impl.h:408:24: note: candidate function not viable: no known conversion from '__ubuf__ half *' to '__ubuf__ int16_t *' (aka '__ubuf__ short *') for 1st argument
__aicore__ inline void CastIntrinsicsImpl(__ubuf__ int16_t* dst, __ubuf__ int32_t* src, const RoundMode& roundMode,
                       ^
1 error generated.
gmake[5]: *** [CMakeFiles/m200_obj.dir/build.make:118: CMakeFiles/m200_obj.dir/data/llama.cpp/ggml/src/ggml-cann/kernels/get_row_q4_0.cpp.o] Error 1
gmake[4]: *** [CMakeFiles/Makefile2:115: CMakeFiles/m200_obj.dir/all] Error 2
gmake[3]: *** [Makefile:91: all] Error 2
gmake[2]: *** [ggml/src/ggml-cann/kernels/CMakeFiles/ascendc_kernels_preprocess.dir/build.make:86: ggml/src/ggml-cann/kernels/ascendc_kernels_preprocess-prefix/src/ascendc_kernels_preprocess-stamp/ascendc_kernels_preprocess-build] Error 2
gmake[1]: *** [CMakeFiles/Makefile2:1897: ggml/src/ggml-cann/kernels/CMakeFiles/ascendc_kernels_preprocess.dir/all] Error 2
gmake: *** [Makefile:146: all] Error 2