Add trace support for jit
- CPU Backend - Nothing added as there is compile/link stage.
- CUDA Backend
- OpenCL Backend
Sample output from CUDA backend.
Pure JIT kernels
[==========] Running 1 test from 1 test case.
[----------] Global test environment set-up.
[----------] 1 test from JIT
[ RUN ] JIT.CPP_JIT_HASH
[jit][022615] [ ../src/backend/cuda/nvrtc/cache.cpp:242 ] Fetch kernel: KER15803858198028938426
[jit][022615] [ ../src/backend/cuda/nvrtc/cache.cpp:244 ] Compile option: --gpu-architecture=compute_61
[jit][022615] [ ../src/backend/cuda/nvrtc/cache.cpp:244 ] Compile option: --std=c++14
[jit][022615] [ ../src/backend/cuda/nvrtc/cache.cpp:242 ] Fetch kernel: KER4770152843556411233
[jit][022615] [ ../src/backend/cuda/nvrtc/cache.cpp:244 ] Compile option: --gpu-architecture=compute_61
[jit][022615] [ ../src/backend/cuda/nvrtc/cache.cpp:244 ] Compile option: --std=c++14
[jit][022615] [ ../src/backend/cuda/nvrtc/cache.cpp:242 ] Fetch kernel: KER9160382770965065811
[jit][022615] [ ../src/backend/cuda/nvrtc/cache.cpp:244 ] Compile option: --gpu-architecture=compute_61
[jit][022615] [ ../src/backend/cuda/nvrtc/cache.cpp:244 ] Compile option: --std=c++14
[ OK ] JIT.CPP_JIT_HASH (1036 ms)
[----------] 1 test from JIT (1037 ms total)
[----------] Global test environment tear-down
[==========] 1 test from 1 test case ran. (1037 ms total)
[ PASSED ] 1 test.
NVRTC compilation code path
[==========] Running 1 test from 1 test case.
[----------] Global test environment set-up.
[----------] 1 test from Transpose/0, where TypeParam = float
[ RUN ] Transpose/0.Square512x512
[jit][022725] [ ../src/backend/cuda/nvrtc/cache.cpp:242 ] Fetch kernel: cuda::transpose<float,false,true>
[jit][022725] [ ../src/backend/cuda/nvrtc/cache.cpp:244 ] Compile option: --gpu-architecture=compute_61
[jit][022725] [ ../src/backend/cuda/nvrtc/cache.cpp:244 ] Compile option: --std=c++14
[jit][022725] [ ../src/backend/cuda/nvrtc/cache.cpp:244 ] Compile option: -D TILE_DIM=32
[jit][022725] [ ../src/backend/cuda/nvrtc/cache.cpp:244 ] Compile option: -D THREADS_Y=8
[jit][022725] [ ../src/backend/cuda/nvrtc/cache.cpp:244 ] Compile option: --device-as-default-execution-space
[ OK ] Transpose/0.Square512x512 (800 ms)
[----------] 1 test from Transpose/0 (800 ms total)
[----------] Global test environment tear-down
[==========] 1 test from 1 test case ran. (800 ms total)
[ PASSED ] 1 test.