Dropping CUDA 11?

[jakirkham]

In RAPIDS 25.08, we are dropping CUDA 11 as noted in RSN 48. This is also tracked in issue:

• Drop CUDA 11 rapidsai/build-planning#184

This means CI support for CUDA 11 is also being dropped

When does cuOpt plan to drop CUDA 11? Does it make sense to also drop in cuOpt 25.08?

[jakirkham]

Here are a few CUDA 11 references that come up:

cuopt/build.sh

Lines 271 to 273 in 9952c26

	if [[ ${CUDA_VERSION} == 11* ]]; then
	CUOPT_CMAKE_CUDA_ARCHITECTURES="70-real;80"
	echo "Building for Volta and Ampere architectures..."

cuopt/cpp/src/linear_programming/utilities/ping_pong_graph.cuh

Lines 48 to 63 in 9952c26

	void end_capture(i_t total_pdlp_iterations)
	{
	if (total_pdlp_iterations % 2 == 0 && !even_initialized) {
	RAFT_CUDA_TRY(cudaStreamEndCapture(stream_view_.value(), &even_graph));
	// Extra NULL NULL 0 mandatory for cuda 11.8
	RAFT_CUDA_TRY(cudaGraphInstantiate(&even_instance, even_graph, nullptr, nullptr, 0));
	even_initialized = true;
	RAFT_CUDA_TRY_NO_THROW(cudaGraphDestroy(even_graph));
	} else if (total_pdlp_iterations % 2 == 1 && !odd_initialized) {
	RAFT_CUDA_TRY(cudaStreamEndCapture(stream_view_.value(), &odd_graph));
	// Extra NULL NULL 0 mandatory for cuda 11.8
	RAFT_CUDA_TRY(cudaGraphInstantiate(&odd_instance, odd_graph, nullptr, nullptr, 0));
	odd_initialized = true;
	RAFT_CUDA_TRY_NO_THROW(cudaGraphDestroy(odd_graph));
	}
	}

cuopt/.github/workflows/pr.yaml

Lines 166 to 176 in 9952c26

	#conda-notebook-tests:
	# needs: [conda-python-build, changed-files]
	# secrets: inherit
	# uses: rapidsai/shared-workflows/.github/workflows/custom-job.yaml@branch-25.08
	# #if: fromJSON(needs.changed-files.outputs.changed_file_groups).test_notebooks
	# with:
	# build_type: pull-request
	# node_type: "gpu-l4-latest-1"
	# arch: "amd64"
	# container_image: "rapidsai/ci-conda:cuda11.8.0-ubuntu22.04-py3.10"
	# run_script: "ci/test_notebooks.sh"

[jakirkham]

From discussion with Ram offline, sounds like CUDA 11 should already be dropped. Ram opened a PR to clean out the last bits of CUDA 11