Use CUDA native functions in math namespace #5782
stephenswatbuilds.yml
on: pull_request
gitlab-benchmark
3m 24s
Matrix: device-container
Matrix: host-container
Matrix: native
Annotations
3 warnings
|
FP64 instructions emitted:
core/include/detray/utils/tuple.hpp#L65
Instruction(s) generated are 10 × `cvt.rn.f64.s64`, 10 × `mul.rn.f64`, and 10 × `cvt.rn.f32.f64` in translation unit(s) `propagation_kernel.ptx`, `benchmark_propagator_cuda_kernel.ptx`, `benchmark_propagator_cuda_kernel.ptx`, `propagator_cuda_kernel.ptx`, and `propagator_cuda_kernel.ptx`.
|
|
FP64 instructions emitted:
tests/unit_tests/device/cuda/sf_finders_grid_cuda_kernel.cu#L222
Instruction(s) generated are 24 × `cvt.f64.f32` and 24 × `st.f64` in translation unit(s) `sf_finders_grid_cuda_kernel.ptx` and `sf_finders_grid_cuda_kernel.ptx`.
|
|
FP64 instructions emitted:
build/_deps/algebraplugins-src/math/common/include/algebra/math/common.hpp#L39
Instruction(s) generated are 2 × `cvt.rn.f64.s64`, 2 × `mul.rn.f64`, and 2 × `cvt.rn.f32.f64` in translation unit(s) `mask_store_cuda_kernel.ptx` and `mask_store_cuda_kernel.ptx`.
|