Skip to content

compile error when try to install mxnet on rocm platform #7

Open
Open
compile error when try to install mxnet on rocm platform#7
@andyzhanged

Description

@andyzhanged
andyzhanged
opened on Jan 5, 2021

Environment info

OS:
$ lsb_release -a
No LSB modules are available.
Distributor ID: Ubuntu
Description: Ubuntu 18.04.5 LTS
Release: 18.04
Codename: bionic

Compiler:
$ hipcc --version
HIP version: 3.3.20126-2dbba46b
HCC clang version 10.0.0 (/data/jenkins-workspace/compute-rocm-rel-3.3/external/hcc-tot/llvm-project/clang 1ce0fe5e88b2124494b9500817b4c2c66bdfa5aa) (based on HCC 3.1.20114-6776c83f-1ce0fe5e88b )
Target: x86_64-unknown-linux-gnu
Thread model: posix
InstalledDir: /opt/rocm-3.3.0/hcc/bin

Package used (Python/R/Scala/Julia): python

MXNET version: install from source.

MXNet commit hash (git rev-parse HEAD):
0cd2b0b

Python version and distribution:
$ python --version
Python 3.6.9

ROCM Version: 3.3.0

Error Message:

I try to install mxnet follow the guide https://github.com/ROCmSoftwarePlatform/mxnet#installation-guide.

$ make -j1  CXX=g++-6 USE_GPU=1              
Makefile:346: WARNING: could not find nvcc compiler, the specified path was: hipcc
Running CUDA_ARCH: --amdgpu-target=gfx801 --amdgpu-target=gfx802 --amdgpu-target=gfx803 --amdgpu-target=gfx900 --amdgpu-target=gfx906
cd /disk/zhanged/code/mxnet/3rdparty/dmlc-core; make libdmlc.a USE_SSE=1 config=/disk/zhanged/code/mxnet/make/config.mk; cd /disk/zhanged/code/mxnet
make[1]: Entering directory '/disk/zhanged/code/mxnet/3rdparty/dmlc-core'
make[1]: 'libdmlc.a' is up to date.
make[1]: Leaving directory '/disk/zhanged/code/mxnet/3rdparty/dmlc-core'
hipcc -std=c++11 -Xcompiler -D_FORCE_INLINES -g -O3   --amdgpu-target=gfx801 --amdgpu-target=gfx802 --amdgpu-target=gfx803 --amdgpu-target=gfx900 --amdgpu-target=gfx906 -Xcompiler "-DMSHADOW_FORCE_STREAM -Wall -Wsign-compare -O3 -DNDEBUG=1 -I. -I/opt/rocm/include -I/opt/rocm/hipblas/include -I/opt/rocm/hiprand/include -I/opt/rocm/rocfft/include -I/opt/rocm/hipcub/include/ -I/opt/rocm/rocblas/include -I/opt/rocm/rocrand/include -I/disk/zhanged/code/mxnet/3rdparty/mshadow/ -I/disk/zhanged/code/mxnet/3rdparty/dmlc-core/include -fPIC -I/disk/zhanged/code/mxnet/3rdparty/tvm/nnvm/include -I/disk/zhanged/code/mxnet/3rdparty/dlpack/include -I/disk/zhanged/code/mxnet/3rdparty/tvm/include -Iinclude -funroll-loops -Wno-unused-parameter -Wno-unknown-pragmas -Wno-unused-local-typedefs -msse3 -mf16c -DMSHADOW_USE_CBLAS=1 -DMSHADOW_USE_MKL=0 -DMSHADOW_RABIT_PS=0 -DMSHADOW_DIST_PS=0 -DMSHADOW_USE_PASCAL=0 -DMXNET_USE_OPENCV=1 -I/usr/include/opencv -fopenmp -DMXNET_USE_OPERATOR_TUNING=1 -DMXNET_USE_LAPACK  -I/opt/rocm/hipcub/include/hipcub/rocprim -DMXNET_USE_RCCL=0 -DMXNET_USE_LIBJPEG_TURBO=0" -M -MT build/src/operator/nn/ctc_loss_gpu.o src/operator/nn/ctc_loss.cu >build/src/operator/nn/ctc_loss_gpu.d
clang-10: warning: argument unused during compilation: '-Xcompiler' [-Wunused-command-line-argument]
clang-10: warning: argument unused during compilation: '--amdgpu-target=gfx801' [-Wunused-command-line-argument]
clang-10: warning: argument unused during compilation: '--amdgpu-target=gfx802' [-Wunused-command-line-argument]
clang-10: warning: argument unused during compilation: '--amdgpu-target=gfx803' [-Wunused-command-line-argument]
clang-10: warning: argument unused during compilation: '--amdgpu-target=gfx900' [-Wunused-command-line-argument]
clang-10: warning: argument unused during compilation: '--amdgpu-target=gfx906' [-Wunused-command-line-argument]
clang-10: warning: argument unused during compilation: '-Xcompiler' [-Wunused-command-line-argument]
hipcc -c -o build/src/operator/nn/ctc_loss_gpu.o -std=c++11 -Xcompiler -D_FORCE_INLINES -g -O3   --amdgpu-target=gfx801 --amdgpu-target=gfx802 --amdgpu-target=gfx803 --amdgpu-target=gfx900 --amdgpu-target=gfx906 -Xcompiler "-DMSHADOW_FORCE_STREAM -Wall -Wsign-compare -O3 -DNDEBUG=1 -I. -I/opt/rocm/include -I/opt/rocm/hipblas/include -I/opt/rocm/hiprand/include -I/opt/rocm/rocfft/include -I/opt/rocm/hipcub/include/ -I/opt/rocm/rocblas/include -I/opt/rocm/rocrand/include -I/disk/zhanged/code/mxnet/3rdparty/mshadow/ -I/disk/zhanged/code/mxnet/3rdparty/dmlc-core/include -fPIC -I/disk/zhanged/code/mxnet/3rdparty/tvm/nnvm/include -I/disk/zhanged/code/mxnet/3rdparty/dlpack/include -I/disk/zhanged/code/mxnet/3rdparty/tvm/include -Iinclude -funroll-loops -Wno-unused-parameter -Wno-unknown-pragmas -Wno-unused-local-typedefs -msse3 -mf16c -DMSHADOW_USE_CBLAS=1 -DMSHADOW_USE_MKL=0 -DMSHADOW_RABIT_PS=0 -DMSHADOW_DIST_PS=0 -DMSHADOW_USE_PASCAL=0 -DMXNET_USE_OPENCV=1 -I/usr/include/opencv -fopenmp -DMXNET_USE_OPERATOR_TUNING=1 -DMXNET_USE_LAPACK  -I/opt/rocm/hipcub/include/hipcub/rocprim -DMXNET_USE_RCCL=0 -DMXNET_USE_LIBJPEG_TURBO=0" src/operator/nn/ctc_loss.cu
clang-10: warning: argument unused during compilation: '-Xcompiler' [-Wunused-command-line-argument]
clang-10: warning: argument unused during compilation: '--amdgpu-target=gfx801' [-Wunused-command-line-argument]
clang-10: warning: argument unused during compilation: '--amdgpu-target=gfx802' [-Wunused-command-line-argument]
clang-10: warning: argument unused during compilation: '--amdgpu-target=gfx803' [-Wunused-command-line-argument]
clang-10: warning: argument unused during compilation: '--amdgpu-target=gfx900' [-Wunused-command-line-argument]
clang-10: warning: argument unused during compilation: '--amdgpu-target=gfx906' [-Wunused-command-line-argument]
clang-10: warning: argument unused during compilation: '-Xcompiler' [-Wunused-command-line-argument]
In file included from src/operator/nn/ctc_loss.cu:26:
In file included from src/operator/nn/./ctc_loss-inl.h:29:
In file included from include/mxnet/operator_util.h:43:
In file included from include/mxnet/./base.h:32:
In file included from /disk/zhanged/code/mxnet/3rdparty/mshadow/mshadow/tensor.h:16:
In file included from /disk/zhanged/code/mxnet/3rdparty/mshadow/mshadow/./base.h:29:
In file included from ./hip-wrappers.h:8:
In file included from /opt/rocm/include/hip/hip_runtime.h:56:
In file included from /opt/rocm/include/hip/hcc_detail/hip_runtime.h:105:
/opt/rocm/include/hip/hcc_detail/surface_functions.h:37:18: warning: comparison of integers of different signs: 'int32_t' (aka 'int') and 'size_t' (aka 'unsigned long') [-Wsign-compare]
    if ((xOffset > width) || (xOffset < 0) || (y > height) || (y < 0)) {
         ~~~~~~~ ^ ~~~~~
/opt/rocm/include/hip/hcc_detail/surface_functions.h:37:50: warning: comparison of integers of different signs: 'int' and 'size_t' (aka 'unsigned long') [-Wsign-compare]
    if ((xOffset > width) || (xOffset < 0) || (y > height) || (y < 0)) {
                                               ~ ^ ~~~~~~
/opt/rocm/include/hip/hcc_detail/surface_functions.h:54:20: warning: comparison of integers of different signs: 'int32_t' (aka 'int') and 'size_t' (aka 'unsigned long') [-Wsign-compare]
    if (!((xOffset > width) || (xOffset < 0) || (y > height) || (y < 0))) {
           ~~~~~~~ ^ ~~~~~
/opt/rocm/include/hip/hcc_detail/surface_functions.h:54:52: warning: comparison of integers of different signs: 'int' and 'size_t' (aka 'unsigned long') [-Wsign-compare]
    if (!((xOffset > width) || (xOffset < 0) || (y > height) || (y < 0))) {
                                                 ~ ^ ~~~~~~
In file included from src/operator/nn/ctc_loss.cu:26:
In file included from src/operator/nn/./ctc_loss-inl.h:34:
src/operator/nn/./sequence_mask-inl.h:55:5: warning: misleading indentation; statement is not part of the previous 'if' [-Wmisleading-indentation]
    for (index_t s = lengths[batch]; s < smax; ++s)
    ^
src/operator/nn/./sequence_mask-inl.h:51:3: note: previous statement is here
  if (batch >= bmax)
  ^
In file included from src/operator/nn/ctc_loss.cu:26:
In file included from src/operator/nn/./ctc_loss-inl.h:35:
In file included from src/operator/nn/../sequence_op_common.h:31:
In file included from src/operator/nn/.././operator_common.h:42:
src/operator/nn/../../common/cuda_utils.h:246:11: warning: enumeration values 'HIPRAND_STATUS_DOUBLE_PRECISION_REQUIRED' and 'HIPRAND_STATUS_NOT_IMPLEMENTED' not handled in switch [-Wswitch]
  switch (status) {
          ^
In file included from src/operator/nn/ctc_loss.cu:27:
In file included from src/operator/nn/../../../3rdparty/ctc_include/detail/gpu_ctc.h:25:
In file included from src/operator/nn/../../../3rdparty/ctc_include/detail/gpu_ctc_kernels.h:23:
In file included from src/operator/nn/../../../3rdparty/ctc_include/detail/../contrib/moderngpu/include/device/ctascan.cuh:38:
src/operator/nn/../../../3rdparty/ctc_include/detail/../contrib/moderngpu/include/device/deviceutil.cuh:68:13: error: no matching function for call to 'min'
        range.x += min(block, task.y);
                   ^~~
src/operator/nn/../../../3rdparty/ctc_include/detail/../contrib/moderngpu/include/device/devicetypes.cuh:260:23: note: candidate function not viable: no known conversion from 'int' to 'int2' (aka 'HIP_vector_type<int, 2>') for 1st argument
MGPU_HOST_DEVICE int2 min(int2 a, int2 b) {
                      ^
src/operator/nn/../../../3rdparty/ctc_include/detail/../contrib/moderngpu/include/device/devicetypes.cuh:243:20: note: candidate template ignored: deduced conflicting types for parameter 'T' ('int' vs. 'hip_impl::Scalar_accessor<int, int __attribute__((ext_vector_type(2))), 1>')
MGPU_HOST_DEVICE T min(T a, T b) {
                   ^
In file included from src/operator/nn/ctc_loss.cu:26:
In file included from src/operator/nn/./ctc_loss-inl.h:29:
In file included from include/mxnet/operator_util.h:43:
In file included from include/mxnet/base.h:32:
In file included from /disk/zhanged/code/mxnet/3rdparty/mshadow/mshadow/./cuda/../tensor.h:16:
In file included from /disk/zhanged/code/mxnet/3rdparty/mshadow/mshadow/base.h:29:
In file included from ./hip-wrappers.h:8:
In file included from /opt/rocm/include/hip/hip_runtime.h:56:
In file included from /opt/rocm/include/hip/hcc_detail/hip_runtime.h:57:
In file included from /opt/rocm/include/hip/hip_runtime_api.h:348:
In file included from /opt/rocm/include/hip/hcc_detail/hip_runtime_api.h:44:
In file included from /opt/rocm/include/hip/hcc_detail/hip_texture_types.h:38:
In file included from /opt/rocm/include/hip/hcc_detail/channel_descriptor.h:28:
/opt/rocm/include/hip/hcc_detail/hip_vector_types.h:176:22: warning: unused variable 'r' [-Wunused-variable]
                auto r{data[idx]};
                     ^
/opt/rocm/rocrand/include/rocrand_philox4x32_10.h:284:26: note: in instantiation of member function 'hip_impl::Scalar_accessor<unsigned int, unsigned int __attribute__((ext_vector_type(4))), 0>::operator++' requested here
        m_state.counter.x++;
                         ^
7 warnings and 1 error generated.
In file included from src/operator/nn/ctc_loss.cu:26:
In file included from src/operator/nn/./ctc_loss-inl.h:29:
In file included from include/mxnet/operator_util.h:43:
In file included from include/mxnet/./base.h:32:
In file included from /disk/zhanged/code/mxnet/3rdparty/mshadow/mshadow/tensor.h:16:
In file included from /disk/zhanged/code/mxnet/3rdparty/mshadow/mshadow/./base.h:29:
In file included from ./hip-wrappers.h:8:
In file included from /opt/rocm/include/hip/hip_runtime.h:56:
In file included from /opt/rocm/include/hip/hcc_detail/hip_runtime.h:105:
/opt/rocm/include/hip/hcc_detail/surface_functions.h:37:18: warning: comparison of integers of different signs: 'int32_t' (aka 'int') and 'size_t' (aka 'unsigned long') [-Wsign-compare]
    if ((xOffset > width) || (xOffset < 0) || (y > height) || (y < 0)) {
         ~~~~~~~ ^ ~~~~~
/opt/rocm/include/hip/hcc_detail/surface_functions.h:37:50: warning: comparison of integers of different signs: 'int' and 'size_t' (aka 'unsigned long') [-Wsign-compare]
    if ((xOffset > width) || (xOffset < 0) || (y > height) || (y < 0)) {
                                               ~ ^ ~~~~~~
/opt/rocm/include/hip/hcc_detail/surface_functions.h:54:20: warning: comparison of integers of different signs: 'int32_t' (aka 'int') and 'size_t' (aka 'unsigned long') [-Wsign-compare]
    if (!((xOffset > width) || (xOffset < 0) || (y > height) || (y < 0))) {
           ~~~~~~~ ^ ~~~~~
/opt/rocm/include/hip/hcc_detail/surface_functions.h:54:52: warning: comparison of integers of different signs: 'int' and 'size_t' (aka 'unsigned long') [-Wsign-compare]
    if (!((xOffset > width) || (xOffset < 0) || (y > height) || (y < 0))) {
                                                 ~ ^ ~~~~~~
In file included from src/operator/nn/ctc_loss.cu:26:
In file included from src/operator/nn/./ctc_loss-inl.h:34:
src/operator/nn/./sequence_mask-inl.h:55:5: warning: misleading indentation; statement is not part of the previous 'if' [-Wmisleading-indentation]
    for (index_t s = lengths[batch]; s < smax; ++s)
    ^
src/operator/nn/./sequence_mask-inl.h:51:3: note: previous statement is here
  if (batch >= bmax)
  ^
In file included from src/operator/nn/ctc_loss.cu:26:
In file included from src/operator/nn/./ctc_loss-inl.h:35:
In file included from src/operator/nn/../sequence_op_common.h:31:
In file included from src/operator/nn/.././operator_common.h:42:
src/operator/nn/../../common/cuda_utils.h:246:11: warning: enumeration values 'HIPRAND_STATUS_DOUBLE_PRECISION_REQUIRED' and 'HIPRAND_STATUS_NOT_IMPLEMENTED' not handled in switch [-Wswitch]
  switch (status) {
          ^
In file included from src/operator/nn/ctc_loss.cu:27:
In file included from src/operator/nn/../../../3rdparty/ctc_include/detail/gpu_ctc.h:25:
In file included from src/operator/nn/../../../3rdparty/ctc_include/detail/gpu_ctc_kernels.h:23:
In file included from src/operator/nn/../../../3rdparty/ctc_include/detail/../contrib/moderngpu/include/device/ctascan.cuh:38:
src/operator/nn/../../../3rdparty/ctc_include/detail/../contrib/moderngpu/include/device/deviceutil.cuh:68:13: error: no matching function for call to 'min'
        range.x += min(block, task.y);
                   ^~~
src/operator/nn/../../../3rdparty/ctc_include/detail/../contrib/moderngpu/include/device/devicetypes.cuh:260:23: note: candidate function not viable: no known conversion from 'int' to 'int2' (aka 'HIP_vector_type<int, 2>') for 1st argument
MGPU_HOST_DEVICE int2 min(int2 a, int2 b) {
                      ^
src/operator/nn/../../../3rdparty/ctc_include/detail/../contrib/moderngpu/include/device/devicetypes.cuh:243:20: note: candidate template ignored: deduced conflicting types for parameter 'T' ('int' vs. 'hip_impl::Scalar_accessor<int, int __attribute__((ext_vector_type(2))), 1>')
MGPU_HOST_DEVICE T min(T a, T b) {
                   ^
In file included from src/operator/nn/ctc_loss.cu:26:
In file included from src/operator/nn/./ctc_loss-inl.h:29:
In file included from include/mxnet/operator_util.h:43:
In file included from include/mxnet/base.h:32:
In file included from /disk/zhanged/code/mxnet/3rdparty/mshadow/mshadow/./cuda/../tensor.h:16:
In file included from /disk/zhanged/code/mxnet/3rdparty/mshadow/mshadow/base.h:29:
In file included from ./hip-wrappers.h:8:
In file included from /opt/rocm/include/hip/hip_runtime.h:56:
In file included from /opt/rocm/include/hip/hcc_detail/hip_runtime.h:57:
In file included from /opt/rocm/include/hip/hip_runtime_api.h:348:
In file included from /opt/rocm/include/hip/hcc_detail/hip_runtime_api.h:44:
In file included from /opt/rocm/include/hip/hcc_detail/hip_texture_types.h:38:
In file included from /opt/rocm/include/hip/hcc_detail/channel_descriptor.h:28:
/opt/rocm/include/hip/hcc_detail/hip_vector_types.h:176:22: warning: unused variable 'r' [-Wunused-variable]
                auto r{data[idx]};
                     ^
/opt/rocm/rocrand/include/rocrand_philox4x32_10.h:284:26: note: in instantiation of member function 'hip_impl::Scalar_accessor<unsigned int, unsigned int __attribute__((ext_vector_type(4))), 0>::operator++' requested here
        m_state.counter.x++;
                         ^
7 warnings and 1 error generated.
Makefile:507: recipe for target 'build/src/operator/nn/ctc_loss_gpu.o' failed
make: *** [build/src/operator/nn/ctc_loss_gpu.o] Error 1

What have i tried to solve it?

The compiler has confused about min function. So I try to modify the code as following show:

diff --git a/3rdparty/ctc_include/contrib/moderngpu/include/device/deviceutil.cuh b/3rdparty/ctc_include/contrib/moderngpu/include/device/deviceutil.cuh
index e18807f38..fb4f08f21 100644
--- a/3rdparty/ctc_include/contrib/moderngpu/include/device/deviceutil.cuh
+++ b/3rdparty/ctc_include/contrib/moderngpu/include/device/deviceutil.cuh
@@ -65,7 +65,7 @@ MGPU_HOST int2 DivideTaskRange(int numItems, int numWorkers) {
 MGPU_HOST_DEVICE int2 ComputeTaskRange(int block, int2 task) {
        int2 range;
        range.x = task.x * block;
-       range.x += min(block, task.y);
+       range.x += min(block, (int)task.y);
        range.y = range.x + task.x + (block < task.y);
        return range;
 }

After the modify, Mxnet build success and the demo bdk_demo.py run success on my vega20 card.
So the reason is int2 type or template function min. Is some one has some idea? thanks.

Metadata

Metadata

Assignees

No one assigned

    Labels

    No labels
    No labels

    Type

    No type

    Projects

    No projects

    Milestone

    No milestone

    Relationships

    None yet

    Development

    No branches or pull requests

    Issue actions