refactor: make solver settings constructible without CUDA

[ramakrishnap-nv]

3 of 4 toward a CUDA-free client library. Stacked on #1802.

This is the subtlest PR of the four — two small changes, each resting on a specific language guarantee. Worth reading closely; that's why it's separated out.

1. Warm-start data behind a shared_ptr

pdlp_solver_settings_t held pdlp_warm_start_data_t by value. That type owns nine rmm::device_uvector, and its default constructor is out-of-line in a CUDA TU — device_uvector has no default ctor (it needs a stream), and constructing even a zero-size one calls cudaGetDevice. So merely constructing settings pulled in libcuopt.

Now held by shared_ptr, allocated lazily.

Why shared_ptr and not unique_ptr: shared_ptr type-erases its deleter into the control block at construction, so a host-only TU can copy and destroy the member without the complete type. unique_ptr bakes the deleter into the type, which would just move the problem from the constructor to the destructor.

The ~88 device-side uses inside set_pdlp_warm_start_data() are untouched — a local reference alias keeps that code reading exactly as before.

2. Compile-time opt-out of the GPU warm-start path

populate_from_data_model_view() inlined both directions in one if/else. The GPU branch is only reachable when handle != nullptr, but the compiler instantiated both into every TU including the header, dragging in convert_to_gpu_warmstart, convert_to_cpu_warmstart and pdlp_warm_start_data_t(view, stream).

Split into apply_warmstart_gpu_target() (declared in the header, defined in optimization_problem.cu) and apply_warmstart_cpu_target() (host-only, inline), selected by a kHostOnly template parameter dispatched with if constexpr:

if constexpr (kHostOnly) {
  apply_warmstart_cpu_target(solver_settings);
} else {
  if (handle != nullptr) { apply_warmstart_gpu_target(solver_settings, handle); }
  else                   { apply_warmstart_cpu_target(solver_settings); }
}

Why if constexpr and not a runtime if: the goal is that a host-only caller never instantiates the GPU branch, so it emits no reference to it. A runtime branch would still instantiate both and leave the undefined symbols. This one change removed 5 of the 6 GPU references from the client.

Also moves the warm-start accessors that need no allocation into solver_settings_accessors.cpp, leaving the CUDA TU with only members that do.

Testing

Full build + 126 test binaries, 0 errors. 111/125 pass; the 14 failures are cudaErrorUnknown from a locally wedged nvidia_uvm, identical on unmodified main.

🤖 Generated with Claude Code

[copy-pr-bot]

Auto-sync is disabled for draft pull requests in this repository. Workflows must be run manually.

Contributors can view more details about this message here.

[ramakrishnap-nv]

/ok to test