GPU acceleration for GFN0/1/2 SCF, gradient and AES via a CUDA-C cuSolver/cuBLAS shim#1418

Open

Prasanna163 wants to merge 26 commits into

grimme-lab:mainfrom

Prasanna163:phase2-batched-scf

Commits on Jun 22, 2026

Add GPU batch driver + batched eigensolver (Phase 0/1)

Prasanna163
and
claude
committed
docs: GPU build/validation guide + throughput benchmark

Prasanna163
and
claude
committed
Route GFN0 production energy through the batched solve (Phase 1)

Prasanna163
and
claude
committed

Commits on Jun 24, 2026

Phase 2: GPU-accelerated GFN1/GFN2 SCF, gradient and --opt (cuSolver/cuBLAS)

Prasanna163
and
claude
committed

Commits on Jun 26, 2026

Commits on Jun 27, 2026

Commits on Jul 4, 2026

Add native Windows build support for the CUDA GPU shim

Prasanna163
and
claude
committed

Commits on Jul 6, 2026

Add GPU development helper scripts
Prasanna163
committed