Older gcc, e.g., 5.x, see an ambiguity in some calls. Example error
output as seen with gcc 5.5:
In file included from benchmark_wino.cpp:1:
In file included from ./miopen.hpp:13:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/regex:38:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/algorithm:62:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/bits/stl_algo.h:66:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/random:51:
/usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/bits/random.tcc:1324:27: error: call to 'abs' is ambiguous
const double __y = -std::abs(__n) * __param._M_sm - 1;
^~~~~~~~
/opt/rocm/hip/include/hip/hcc_detail/hip_complex.h:345:31: note: candidate function
__DEFINE_HIP_COMPLEX_REAL_FUN(abs, hipCabs)
Older gcc, e.g., 5.x, see an ambiguity in some calls. Example error
output as seen with gcc 5.5:
In file included from benchmark_wino.cpp:1:
In file included from ./miopen.hpp:13:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/regex:38:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/algorithm:62:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/bits/stl_algo.h:66:
In file included from /usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/random:51:
/usr/lib/gcc/x86_64-linux-gnu/5.5.0/../../../../include/c++/5.5.0/bits/random.tcc:1324:27: error: call to 'abs' is ambiguous
const double __y = -std::abs(__n) * __param._M_sm - 1;
^~~~~~~~
/opt/rocm/hip/include/hip/hcc_detail/hip_complex.h:345:31: note: candidate function
__DEFINE_HIP_COMPLEX_REAL_FUN(abs, hipCabs)
- Removed hcc integration logic for now
- No longer publish docker images to artifactory or docker hub
- Switch to new docker images and new test node
Change-Id: Idb06bf0a52658d1869367cb38bb03ef4f1649103
hcc_detail/math_functions.h used to include hcc_detail/hip_runtime.h.
Removing it has caused regression in TensorFlow 1.8.
Put it back for backward compatibiliity.
When shareWithAll memory (e.g., host memory) is allocated, set appId
in hc::AmPointerInfo to -1 to indicate that this memory is not mapped
to any device. Peer checking in ihipStream_t::canSeeMemory is not
necessary if memory is shared with all devices. Thus, it is skipped.
Note that earlier host memory is always mapped to device 0 and HIP
always performs peer checking for all kinds of hipMemcpy. Since the
peer checking process requires context locking, hipMemcpy from/to host
memory always grabs device 0's context lock. Therefore, if there is
another thread holding the context lock of device 0 (e.g.,
hipDeviceSynchronize on device 0), hipMemcpy will have to wait for the
lock until it can actually perform memcpy. This can significantly
deteriorate execution performance.
Signed-off-by: Sarunya Pumma <sarunya.pumma@amd.com>