Merge remote-tracking branch 'origin/multipath-testing-ci' into integration_2026_w07 (!3886)

Integrate GPU-Accelerated Channel Simulation + CI Test Pipeline

This MR builds on top of MR !3588 (feat: GPU Accelerated Channel
Simulation Pipeline).

The main portion of the changes is directly taken from MR !3588, and
focuses on integrating the GPU-accelerated channel simulation. On top of
that, this MR adds CI-related configurations to introduce a new channel
simulation testing pipeline into the CI --> RAN-Channel-Simulation

This MR introduces a CUDA-based pipeline to offload the
multipath_channel and add_noise channel simulation functions to the GPU.
This enables large-scale, real-time physical layer simulations not
possible on the CPU.

For full documentation, please see the file at openair1/SIMULATION/TOOLS/DOC/gpu_acceleration.md.

System & Compiler Requirements: An NVIDIA GPU with a compatible driver
and the CUDA Toolkit installed.

This feature was developed and tested on the GH200 server using the GCC
12 toolchain. To ensure reproducibility, building with GCC 12 is highly
recommended.  Build Instructions

The primary flag to enable this feature is -DCUDA_ENABLE=ON. The
following command shows a sample configuration used during development
on the GH200 server:

    ./build_oai --phy_simulators --cmake-opt "\
      -DCUDA_ENABLE=ON \
      -DUSE_UNIFIED_MEMORY=ON \
      -DUSE_ATS_MEMORY=OFF \
      -DCMAKE_C_COMPILER=/usr/bin/gcc-12 \
      -DCMAKE_CXX_COMPILER=/usr/bin/g++-12 \
      -DENABLE_TESTS=ON"

Memory Model Configuration

The pipeline supports multiple CUDA memory management modes, which can
be configured at compile time. ATS Memory (-DUSE_ATS_MEMORY=ON) is
enabled by default. To use another memory model, you must first disable
ATS.

For a detailed explanation of these models, please refer to Section 3 of
the gpu_acceleration.md report.

CMakeLists.txt

@@ -24,6 +24,38 @@ project (OpenAirInterface LANGUAGES C CXX)
 set(CMAKE_CXX_STANDARD 17)
 set(OAI_VERSION 2.4.0)
 
+option(CUDA_ENABLE "Enable CUDA accelerated channel simulation" OFF)
+if(CUDA_ENABLE)
+    find_package(CUDA REQUIRED)
+    find_package(CUDAToolkit REQUIRED)
+    message(STATUS "CUDA explicitly enabled, building with GPU acceleration support.")
+
+    enable_language(CUDA)
+
+    if(CMAKE_SYSTEM_PROCESSOR STREQUAL "aarch64")
+        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()
+
 #########################################################
 # Base directories, compatible with legacy OAI building #
 #########################################################
@@ -289,6 +321,22 @@ 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
@@ -1695,6 +1743,17 @@ set (SIMUSRC
 add_library(SIMU STATIC ${SIMUSRC} )
 target_include_directories(SIMU PUBLIC ${OPENAIR1_DIR}/SIMULATION/TOOLS ${OPENAIR1_DIR}/SIMULATION/RF)
 
+option(CHANNEL_SSE "Enable SSE optimizations for channel simulation" OFF)
+if(CHANNEL_SSE)
+    message(STATUS "SSE-optimized channel simulation enabled.")
+    target_compile_definitions(SIMU PRIVATE CHANNEL_SSE)
+endif()
+
+if(CUDA_FOUND)
+  set_property(TARGET SIMU PROPERTY CUDA_ARCHITECTURES ${CMAKE_CUDA_ARCHITECTURES})
+  set_property(TARGET SIMU PROPERTY LINKER_LANGUAGE CXX)
+endif()
+
 include_directories("${NFAPI_DIR}/nfapi/public_inc")
 include_directories("${NFAPI_DIR}/common/public_inc")
 include_directories("${NFAPI_DIR}/pnf/public_inc")
@@ -2092,10 +2151,19 @@ target_link_libraries(nr_srssim PRIVATE
   m pthread ITTI dl nr_ue_phy_meas physim_common
   )
 
+if(CUDA_ENABLE)
+  if (TARGET oai_cuda_lib)
+      target_link_libraries(nr_dlsim PRIVATE oai_cuda_lib)
+      target_link_libraries(nr_ulsim PRIVATE oai_cuda_lib)
+
+      target_compile_definitions(nr_dlsim PRIVATE ENABLE_CUDA)
+      target_compile_definitions(nr_ulsim PRIVATE ENABLE_CUDA)
+  endif()
+endif()
+
 # these simulators do not compile:
 # dlsim_tm7 pbchsim scansim mbmssim pdcchsim pucchsim prachsim syncsim
 foreach(myExe dlsim ulsim)
-
   add_executable(${myExe}
     ${OPENAIR1_DIR}/SIMULATION/LTE_PHY/${myExe}.c
     ${OPENAIR_DIR}/common/utils/lte/ue_power.c
@@ -2205,6 +2273,11 @@ add_subdirectory(openair3)
 add_subdirectory(radio)
 add_subdirectory(tests)
 
+if(TARGET oai_cuda_lib)
+    message(STATUS "CUDA library 'oai_cuda_lib' found, linking to targets...")
+    target_link_libraries(nr_dlsim PRIVATE oai_cuda_lib)
+    target_link_libraries(nr_ulsim PRIVATE oai_cuda_lib)
+endif()
 
 if(PACKAGING_LTE OR PACKAGING_NR OR PACKAGING_COMMON OR PACKAGING_USRP OR PACKAGING_PHYSIM)
   include("tools/packages/packages.cmake")