Implement CUDA Multipath Channel Simulation

This commit introduces the CUDA implementation of the multipath channel
model in `multipath_channel.cu`.

- Implements `multipath_channel_kernel` using shared memory for
  efficient, high-performance convolution.

- Provides a C-callable wrapper, `multipath_channel_cuda`, to
  orchestrate the full pipeline of data transfers and kernel
  launches.

- The wrapper includes support for multiple memory models (Explicit
  Copy, Unified Memory, ATS) via preprocessor directives to adapt to
  different build configurations.

- Adds a `interleave_channel_output_cuda` utility to convert signal
  data from the CPU's planar format to the GPU's interleaved format.
This commit is contained in:
Nika Ghaderi
2026-02-09 09:29:36 +00:00
committed by Jaroslava Fiedlerova
parent c54f0f7955
commit 77d854bdf3
4 changed files with 284 additions and 3 deletions
+33
View File
@@ -36,6 +36,23 @@ if(CUDA_ENABLE)
set(CMAKE_CUDA_ARCHITECTURES 90)
endif()
option(USE_UNIFIED_MEMORY "Enable CUDA Unified Memory path" OFF)
option(USE_ATS_MEMORY "Enable ATS Hybrid Memory path" ON)
if(USE_UNIFIED_MEMORY AND USE_ATS_MEMORY)
message(FATAL_ERROR "Cannot enable both USE_UNIFIED_MEMORY and USE_ATS_MEMORY at the same time.")
endif()
if(USE_UNIFIED_MEMORY)
add_compile_definitions(USE_UNIFIED_MEMORY)
message(STATUS "CUDA Unified Memory path enabled.")
elseif(USE_ATS_MEMORY)
add_compile_definitions(USE_ATS_MEMORY)
message(STATUS "CUDA ATS Hybrid Memory path enabled.")
else()
message(STATUS "CUDA Explicit Copy path enabled (default ATS was overridden).")
endif()
add_compile_definitions(ENABLE_CUDA)
endif()
@@ -297,6 +314,22 @@ if(GIT_FOUND)
)
endif()
option(USE_UNIFIED_MEMORY "Enable CUDA Unified Memory path instead of explicit copies" OFF)
option(USE_ATS_MEMORY "Enable Hybrid ATS (CPU->GPU) and Explicit Copy (GPU->CPU) path" OFF)
if(USE_UNIFIED_MEMORY AND USE_ATS_MEMORY)
message(FATAL_ERROR "Cannot enable both USE_UNIFIED_MEMORY and USE_ATS_MEMORY at the same time.")
endif()
if(USE_UNIFIED_MEMORY)
add_compile_definitions(USE_UNIFIED_MEMORY)
message(STATUS "CUDA Unified Memory path enabled.")
elseif(USE_ATS_MEMORY)
add_compile_definitions(USE_ATS_MEMORY)
message(STATUS "CUDA ATS Hybrid Memory path enabled.")
endif()
# Debug related options
#########################################
# asn1c skeletons have hardcoded this flag to make customized debug logs