Classes | |
| class | CudaGPUBLAS |
| class | GPUBLAS |
| class | HIPGPUBLAS |
| class | SYCLGPUBLAS |
Functions | |
| std::unique_ptr< GPUBLAS > | create_accelerate_gpu_blas (std::shared_ptr< device::GPUDeviceBackend > device_backend) |
| std::unique_ptr< GPUBLAS > | create_cuda_gpu_blas (std::shared_ptr< device::GPUDeviceBackend > device_backend) |
| std::unique_ptr< GPUBLAS > | create_hip_gpu_blas (std::shared_ptr< device::GPUDeviceBackend > device_backend) |
| std::unique_ptr< GPUBLAS > | create_sycl_gpu_blas (std::shared_ptr< device::GPUDeviceBackend > device_backend) |
Forward declaration — GPUBLAS is defined at: layer_0_core/math/linear_algebra/gpu/blas/init.hpp
main_gpu.cpp — Metal/Accelerate GPU-accelerated entry point for Q3.
Links the Metal device backend with Accelerate-based BLAS/LAPACK factories and delegates the shared entry-point flow to cli::run_gpu_entry_point(). No native GPU FFT grid: the FP64 path uses the PlanewaveBasis FFTW grid.
Usage: q3_gpu <input_file> [–log-level <level>] [–precision double|mixed]
main_gpu_cuda.cpp — CUDA GPU-accelerated entry point for Q3.
Links the CUDA backend factories (device, cuBLAS, cuSOLVER, cuFFT) and delegates the shared entry-point flow to cli::run_gpu_entry_point().
Usage: q3_gpu_cuda <input_file> [–log-level <level>] [–precision double|mixed]
main_gpu_hip.cpp — HIP/ROCm GPU-accelerated entry point for Q3.
Links the HIP backend factories (device, hipBLAS, hipSOLVER, hipFFT) and delegates the shared entry-point flow to cli::run_gpu_entry_point().
Usage: q3_gpu_hip <input_file> [–log-level <level>] [–precision double|mixed]
main_gpu_sycl.cpp — SYCL/oneAPI GPU-accelerated entry point for Q3.
Links the SYCL backend factories (device, oneMKL BLAS/LAPACK, oneMKL DFT) and delegates the shared entry-point flow to cli::run_gpu_entry_point().
Usage: q3_gpu_sycl <input_file> [–log-level <level>] [–precision double|mixed]
| std::unique_ptr< GPUBLAS > q3::execution::gpu::blas::create_accelerate_gpu_blas | ( | std::shared_ptr< device::GPUDeviceBackend > | device_backend | ) |
Referenced by main().
| std::unique_ptr< GPUBLAS > q3::execution::gpu::blas::create_cuda_gpu_blas | ( | std::shared_ptr< device::GPUDeviceBackend > | device_backend | ) |
| std::unique_ptr< GPUBLAS > q3::execution::gpu::blas::create_hip_gpu_blas | ( | std::shared_ptr< device::GPUDeviceBackend > | device_backend | ) |
| std::unique_ptr< GPUBLAS > q3::execution::gpu::blas::create_sycl_gpu_blas | ( | std::shared_ptr< device::GPUDeviceBackend > | device_backend | ) |
1.9.8 