HCC : An open source C++ compiler for heterogeneous devices

This repository hosts HCC compiler implementation project. The goal is to implement a compiler that takes a program conforming parallel programming standards such as C++ AMP, HC, C++ 17 ParallelSTL, or OpenMP and transforms it into AMD GCN ISA.

The project is based on LLVM+CLANG. For more information, please visit the hcc wiki:

https://github.com/RadeonOpenCompute/hcc/wiki

Git submodules

The project now employs git submodules to manage external components it depends upon. It it advised to add --recursive when you clone the project so all submodules are fetched automatically.

For example:

# automatically fetches all submodules
git clone --recursive -b clang_tot_upgrade https://github.com/RadeonOpenCompute/hcc.git

For more information about git submodules, please refer to git documentation.

Device libraries

HCC device library is a part of ROCm-Device-Libs. When compiling device code with hcc, the rocm-device-libs package needs to be installed.

In case rocm-device-libs package is not present, you are required to build it from source. Please refer to ROCm-Device-Libs build procedure for more details.

Once it's built, run make install and config ToT HCC like:

cmake \
    -DCMAKE_BUILD_TYPE=Release \
    -DHSA_AMDGPU_GPU_TARGET=<AMD GPU ISA version string> \
    -DROCM_DEVICE_LIB_DIR=<location of the ROCm-Device-Libs bitcode> \
    <ToT HCC checkout directory>

An example would be:

# Use gfx803 AMD GPU ISA
# ROCm-Device-Libs is built at ~/ocml/build , bitcodes are at
# ~/ocml/build/dist/lib
cmake \
    -DCMAKE_BUILD_TYPE=Release \
    -DHSA_AMDGPU_GPU_TARGET=gfx803 \
    -DROCM_DEVICE_LIB_DIR=~/ocml/build/dist/lib \
    ..

Multiple ISA

HCC now supports having multiple GCN ISAs in one executable file. You can do it in different ways:

use --amdgpu-target= command line option

It's possible to specify multiple --amdgpu-target= option. Example:

# ISA for Hawaii(gfx701), Carrizo(gfx801), Tonga(gfx802) and Fiji(gfx803) would 
# be produced
hcc `hcc-config --cxxflags --ldflags` \
    --amdgpu-target=gfx701 \
    --amdgpu-target=gfx801 \
    --amdgpu-target=gfx802 \
    --amdgpu-target=gfx803 \
    foo.cpp

use HCC_AMDGPU_TARGET env var

Use , to delimit each AMDGPU target in HCC. Example:

export HCC_AMDGPU_TARGET=gfx701,gfx801,gfx802,gfx803
# ISA for Hawaii(gfx801), Carrizo(gfx801), Tonga(gfx802) and Fiji(gfx803) would 
# be produced
hcc `hcc-config --cxxflags --ldflags` foo.cpp

configure HCC use CMake HSA_AMDGPU_GPU_TARGET variable

If you build HCC from source, it's possible to configure it to automatically produce multiple ISAs via HSA_AMDGPU_GPU_TARGET CMake variable.

Use ; to delimit each AMDGPU target. Example:

# ISA for Hawaii(gfx701), Carrizo(gfx801), Tonga(gfx802) and Fiji(gfx803) would 
# be produced by default
cmake \
    -DCMAKE_BUILD_TYPE=Release \
    -DROCM_DEVICE_LIB_DIR=~hcc/ROCm-Device-Libs/build/dist/lib \
    -DHSA_AMDGPU_GPU_TARGET="gfx701;gfx801;gfx802;gfx803" \
    ../hcc

Building clang_tot_upgrade branch on Ubuntu 16.04.1

The following issue is common when building HCC tot branch.

In file included from /home/aditya/rocm/hcc.lc.tot/lib/mcwamp.cpp:8:
In file included from /usr/include/c++/v1/iostream:38:
In file included from /usr/include/c++/v1/ios:216:
In file included from /usr/include/c++/v1/__locale:15:
/usr/include/c++/v1/string:1938:44: error: 'basic_string<_CharT, _Traits, _Allocator>' is
      missing exception specification
      'noexcept(is_nothrow_copy_constructible<allocator_type>::value)'
basic_string<_CharT, _Traits, _Allocator>::basic_string(const allocator_type& __a)
                                           ^
/usr/include/c++/v1/string:1326:40: note: previous declaration is here
    _LIBCPP_INLINE_VISIBILITY explicit basic_string(const allocator_type& __a)
                                       ^
1 error generated.
lib/CMakeFiles/mcwamp.dir/build.make:62: recipe for target 'lib/CMakeFiles/mcwamp.dir/mcwamp.cpp.o' failed
make[2]: *** [lib/CMakeFiles/mcwamp.dir/mcwamp.cpp.o] Error 1
CMakeFiles/Makefile2:229: recipe for target 'lib/CMakeFiles/mcwamp.dir/all' failed
make[1]: *** [lib/CMakeFiles/mcwamp.dir/all] Error 2
Makefile:149: recipe for target 'all' failed
make: *** [all] Error 2

This is because of the libc++ package (version 3.7.x) that ships with Ubuntu 16.04.1 being broken. This can be solved by installing the libc++1 and libc++-dev packages from upstream Debian.

The latest working version, at the time of this writing, is 3.9.0-3. For example, the following instruction sequence performs the update:

wget http://ftp.us.debian.org/debian/pool/main/libc/libc++/libc++-dev_3.9.0-3_amd64.deb
wget http://ftp.us.debian.org/debian/pool/main/libc/libc++/libc++1_3.9.0-3_amd64.deb

dpkg -i libc++1_3.9.0-3_amd64.deb
dpkg -i libc++-dev_3.9.0-3_amd64.deb

This replaces the previous version of libc++.

CodeXL Activity Logger

To enable the CodeXL Activity Logger, use the USE_CODEXL_ACTIVITY_LOGGER environment variable.

Configure the build in the following way:

cmake \
    -DCMAKE_BUILD_TYPE=Release \
    -DHSA_AMDGPU_GPU_TARGET=<AMD GPU ISA version string> \
    -DROCM_DEVICE_LIB_DIR=<location of the ROCm-Device-Libs bitcode> \
    -DUSE_CODEXL_ACTIVITY_LOGGER=1 \
    <ToT HCC checkout directory>

In your application compiled using hcc, include the CodeXL Activiy Logger header:

#include <CXLActivityLogger.h>

For information about the usage of the Activity Logger for profiling, please refer to its documentation.