GEMM seems to be hanging at compilation for the Cypress GPU (and oddly enough, only the Cypress GPU) with AMD APP 2.6 on atlanta. Runs fine on Tahiti GPU, CPU, and other OpenCL impls.
@rothpc I'm considering this a bug with the AMD compiler and advise we proceed with a 1.1.4 release.
GDB Output:
0x00002aaaac8bf550 in ?? () from /usr/lib64/libaticaldd.so
Missing separate debuginfos, use: debuginfo-install glibc-2.12-1.47.el6_2.9.x86_64 libX11-1.3-2.el6.x86_64 libXau-1.0.5-1.el6.x86_64 libXext-1.1-3.el6.x86_64 libXinerama-1.1-1.el6.x86_64 libgcc-4.4.6-3.el6.x86_64 libstdc++-4.4.6-3.el6.x86_64 libxcb-1.5-1.el6.x86_64
(gdb) where
#0 0x00002aaaac8bf550 in ?? () from /usr/lib64/libaticaldd.so
#1 0x00002aaaac9b15f8 in ?? () from /usr/lib64/libaticaldd.so
#2 0x00002aaaac9b234b in ?? () from /usr/lib64/libaticaldd.so
#3 0x00002aaaac9b6948 in ?? () from /usr/lib64/libaticaldd.so
#4 0x00002aaaac9b7056 in ?? () from /usr/lib64/libaticaldd.so
#5 0x00002aaaac9b76a2 in ?? () from /usr/lib64/libaticaldd.so
#6 0x00002aaaac8a9333 in ?? () from /usr/lib64/libaticaldd.so
#7 0x00002aaaac8a96e5 in ?? () from /usr/lib64/libaticaldd.so
#8 0x00002aaaac8abd67 in ?? () from /usr/lib64/libaticaldd.so
#9 0x00002aaaacc0506d in ?? () from /usr/lib64/libaticaldd.so
#10 0x00002aaaac83e14a in ?? () from /usr/lib64/libaticaldd.so
#11 0x00002aaaac83f9ed in ?? () from /usr/lib64/libaticaldd.so
#12 0x00002aaaaaeb91c8 in gpu::NullKernel::create(std::basic_string<char, std::char_traits, std::allocator > const&, std::basic_string<char, std::char_traits, std::allocator > const&, void const*, unsigned long) ()
from /usr/lib64/libamdocl64.so
#13 0x00002aaaaaebb473 in gpu::Kernel::create(std::basic_string<char, std::char_traits, std::allocator > const&, std::basic_string<char, std::char_traits, std::allocator > const&, void const*, unsigned long) ()
from /usr/lib64/libamdocl64.so
#14 0x00002aaaaaecaa16 in gpu::Program::createKernel(std::basic_string<char, std::char_traits, std::allocator > const&, gpu::Kernel::InitData const_, std::basic_string<char, std::char_traits, std::allocator > const&, std::basic_string<char, std::char_traits, std::allocator > const&, bool_, void const*, unsigned long) ()
from /usr/lib64/libamdocl64.so
#15 0x00002aaaaaec761c in gpu::NullProgram::linkImpl(amd::option::Options*) () from /usr/lib64/libamdocl64.so
#16 0x00002aaaaae6c9bf in device::Program::build(std::basic_string<char, std::char_traits, std::allocator > const&, char const_, amd::option::Options_) () from /usr/lib64/libamdocl64.so
---Type to continue, or q to quit---
#17 0x00002aaaaae7bad7 in amd::Program::build(std::vector<amd::Device*, std::allocatoramd::Device* > const&, char const_, void ()(cl_program, void), void*, bool) () from /usr/lib64/libamdocl64.so
#18 0x00002aaaaae5d444 in clBuildProgram () from /usr/lib64/libamdocl64.so
#19 0x000000000040c735 in runTest (testName="DGEMM", dev=0xffb340, ctx=0xd5a4d0, queue=0xd8b2f0, resultDB=...,
op=..., compileFlags="-DK_DOUBLE_PRECISION ") at SGEMM.cpp:197
#20 0x0000000000406556 in RunBenchmark (devcpp=, ctxcpp=,
queuecpp=<value optimized out>, resultDB=..., op=...) at SGEMM.cpp:114
#21 0x0000000000404322 in main (argc=3, argv=) at ../../common/main.cpp:208