gpu fixes

[sbryngelson]

No description provided.

[Copilot]

Pull request overview

This PR implements GPU offloading support using OpenMP target directives to accelerate computational kernels in the CFD solver. The changes add GPU execution paths for turbulence models, solver operations, and multigrid Poisson solvers.

Key Changes:

• Added comprehensive GPU execution tests and validation script
• Implemented GPU kernels for turbulence models (baseline mixing length, GEP, and neural network models)
• Added GPU acceleration for convective/diffusive terms and multigrid solver operations

Reviewed changes

Copilot reviewed 10 out of 10 changed files in this pull request and generated 8 comments.

Show a summary per file

File	Description
tests/test_turbulence.cpp	Added GPU upload calls before model updates for MLP and TBNN tests
tests/test_gpu_execution.cpp	New comprehensive test suite verifying GPU execution for MLP, TBNN, and basic GPU computations
test_gpu_offload.sh	New SLURM validation script for end-to-end GPU testing with performance benchmarks
src/turbulence_gep.cpp	Added GPU kernel for GEP turbulence model with inline variant selection
src/turbulence_baseline.cpp	Added GPU kernel for mixing length turbulence model
src/solver.cpp	Added GPU kernels for convective and diffusive term computations
src/poisson_solver_multigrid.cpp	Added GPU kernels for red-black Gauss-Seidel smoothing and residual computation
src/gpu_kernels.cpp	Removed unused variable idx_c from velocity gradient computation
include/turbulence_nn_tbnn.hpp	Added GPU state tracking fields (full_gpu_ready_, cached_n_cells_)
CMakeLists.txt	Added test_gpu_execution test target

---

💡 Add Copilot custom instructions for smarter, more guided reviews. Learn how to get started.