#include <init.hpp>

Inheritance diagram for q3::execution::gpu::GPUExecutionPolicy:

Public Member Functions
	GPUExecutionPolicy (std::shared_ptr< device::GPUDeviceBackend > device_backend)

void	add_scaled_vector (int number_of_elements, ComplexValue scaling_factor, const ComplexValue vector_x, ComplexValue vector_y) const override

void	allgather_within_group (const Real local_data, int local_count, Real full_data, int) const override

const device::GPUDeviceBackend &	device_backend () const

ComplexValue	dot_product (int number_of_elements, const ComplexValue vector_x, const ComplexValue vector_y) const override

int	global_number_of_elements (int local_number_of_elements) const override

int	group_rank_index () const override

int	group_size () const override

bool	is_root () const override

Real	norm (int number_of_elements, const ComplexValue *vector_x) const override

int	number_of_processes () const override

void	reduce_within_group (ComplexValue *, int) const override

void	reduce_within_group (Real *, int) const override

Public Member Functions inherited from q3::execution::ExecutionPolicy
virtual	~ExecutionPolicy ()=default

virtual void	allgather_across_bands (ComplexValue , const int , const int *) const

virtual int	band_distribution_rank_index () const

virtual int	band_distribution_size () const

virtual void	broadcast_across_bands (ComplexValue *, int) const

virtual void	broadcast_across_bands (int *, int) const
	Integer overload, for the counts that decide collective sizes.

virtual void	broadcast_across_bands (Real *, int) const
	Real overload, for eigenvalues – the numbers convergence decisions are made from.

virtual void	broadcast_from_group_root (ComplexValue *, int) const

virtual void	broadcast_from_group_root (Real *, int) const

virtual void	reduce_across_bands (Real *, int) const

virtual void	reduce_group_distributed_result (ComplexValue *, int) const

Protected Attributes
std::shared_ptr< device::GPUDeviceBackend >	device_backend_

Additional Inherited Members
Public Types inherited from q3::execution::ExecutionPolicy
using	ComplexValue = Complex

Detailed Description

Definition at line 35 of file init.hpp.

Constructor & Destructor Documentation

◆ GPUExecutionPolicy()

q3::execution::gpu::GPUExecutionPolicy::GPUExecutionPolicy ( std::shared_ptr< device::GPUDeviceBackend > device_backend )

inlineexplicit

Construct a GPU execution policy.

Parameters

device_backend Shared pointer to the GPU device backend. Must not be null.

Definition at line 43 of file init.hpp.

References device_backend_.

Member Function Documentation

◆ add_scaled_vector()

void q3::execution::gpu::GPUExecutionPolicy::add_scaled_vector	(	int	number_of_elements,
		ComplexValue	scaling_factor,
		const ComplexValue *	vector_x,
		ComplexValue *	vector_y
	)		const

inlineoverridevirtual

Scaled vector addition on GPU-resident data: vector_y += scaling_factor * vector_x.

Intentionally unwired: the GPU compute paths call GPUBLAS directly on device buffers instead of going through this generic policy interface. Throws if reached.

Implements q3::execution::ExecutionPolicy.

Reimplemented in q3::execution::PlanewaveGPUExecutionPolicy.

Definition at line 74 of file init.hpp.

◆ allgather_within_group()

void q3::execution::gpu::GPUExecutionPolicy::allgather_within_group	(	const Real *	local_data,
		int	local_count,
		Real *	full_data,
		int	total_count
	)		const

inlineoverridevirtual

Allgather rank-local slab data into a full-size buffer on every rank.

Gathers each group rank's contiguous local slice into a contiguous full array on every rank (concatenated in rank order). No-op when group_size == 1.

Parameters

local_data	Rank-local slab data.
local_count	Number of real elements in the local slab.
full_data	Output: full-grid data (all slabs concatenated).
total_count	Total number of real elements across all ranks.

Implements q3::execution::ExecutionPolicy.

Definition at line 127 of file init.hpp.

◆ device_backend()

const device::GPUDeviceBackend & q3::execution::gpu::GPUExecutionPolicy::device_backend ( ) const

inline

Access the underlying GPU device backend.

Definition at line 137 of file init.hpp.

References device_backend_.

◆ dot_product()

ComplexValue q3::execution::gpu::GPUExecutionPolicy::dot_product	(	int	number_of_elements,
		const ComplexValue *	vector_x,
		const ComplexValue *	vector_y
	)		const

inlineoverridevirtual

Compute the inner product x† · y on GPU-resident data.

Intentionally unwired: the GPU compute paths (hamiltonian applicator, subspace manager) call GPUBLAS directly on device buffers instead of going through this generic policy interface. Throws if reached.

Implements q3::execution::ExecutionPolicy.

Reimplemented in q3::execution::PlanewaveGPUExecutionPolicy.

Definition at line 57 of file init.hpp.

◆ global_number_of_elements()

int q3::execution::gpu::GPUExecutionPolicy::global_number_of_elements ( int local_number_of_elements ) const

inlineoverridevirtual

Sum a rank-local element count across the process group.

Distributed vector algorithms use this to distinguish the local slab length from the dimension of the complete vector space.

Implements q3::execution::ExecutionPolicy.

Definition at line 119 of file init.hpp.

◆ group_rank_index()

int q3::execution::gpu::GPUExecutionPolicy::group_rank_index ( ) const

inlineoverridevirtual

Index of this process within its process group.

0 in serial mode. NOT necessarily 0 when group_size() == 1: a group whose data distribution was disabled at run time (see MPIExecutionPolicy::disable_group_distribution) reports group_size() == 1 while keeping its pool-local rank index, because the pool identity is still needed. Use group_size() <= 1 to ask "is my data replicated"; use group_rank_index() == 0 only to pick one representative per pool.

Implements q3::execution::ExecutionPolicy.

Definition at line 107 of file init.hpp.

◆ group_size()

int q3::execution::gpu::GPUExecutionPolicy::group_size ( ) const

inlineoverridevirtual

Number of processes in this process group. Always 1 in serial mode or when each group contains a single process.

Implements q3::execution::ExecutionPolicy.

Definition at line 111 of file init.hpp.

◆ is_root()

bool q3::execution::gpu::GPUExecutionPolicy::is_root ( ) const

inlineoverridevirtual

GPU execution is always single-process (no MPI in this policy).

Implements q3::execution::ExecutionPolicy.

Definition at line 103 of file init.hpp.

◆ norm()

Real q3::execution::gpu::GPUExecutionPolicy::norm	(	int	number_of_elements,
		const ComplexValue *	vector_x
	)		const

inlineoverridevirtual

Euclidean norm on GPU-resident data.

Intentionally unwired: the GPU compute paths call GPUBLAS directly on device buffers instead of going through this generic policy interface. Throws if reached.

Implements q3::execution::ExecutionPolicy.

Reimplemented in q3::execution::PlanewaveGPUExecutionPolicy.

Definition at line 93 of file init.hpp.

◆ number_of_processes()

int q3::execution::gpu::GPUExecutionPolicy::number_of_processes ( ) const

inlineoverridevirtual

Number of processes in the complete execution topology.

Implements q3::execution::ExecutionPolicy.

Definition at line 115 of file init.hpp.

◆ reduce_within_group() [1/2]

void q3::execution::gpu::GPUExecutionPolicy::reduce_within_group	(	ComplexValue *	data,
		int	number_of_elements
	)		const

inlineoverridevirtual

Allreduce an array of complex values within the process group.

Used after ZGEMM-based inner products over the distributed vector dimension (e.g., subspace matrices, orthogonalization overlaps) when coefficient vectors are distributed across group ranks. No-op when group_size == 1.

Parameters

data	In-out array of complex values.
number_of_elements	Number of complex elements to reduce.

Implements q3::execution::ExecutionPolicy.

Definition at line 123 of file init.hpp.

◆ reduce_within_group() [2/2]

void q3::execution::gpu::GPUExecutionPolicy::reduce_within_group	(	Real *	data,
		int	number_of_elements
	)		const

inlineoverridevirtual

Allreduce an array of real values within the process group.

Same as the complex variant but for real-valued arrays.

Parameters

data	In-out array of real values.
number_of_elements	Number of real elements to reduce.

Implements q3::execution::ExecutionPolicy.

Definition at line 125 of file init.hpp.

Member Data Documentation

◆ device_backend_

std::shared_ptr<device::GPUDeviceBackend> q3::execution::gpu::GPUExecutionPolicy::device_backend_

protected

Definition at line 142 of file init.hpp.

Referenced by GPUExecutionPolicy(), and device_backend().

The documentation for this class was generated from the following file:

src/layer_0_core/execution/gpu/init.hpp

Public Member Functions

Protected Attributes

Additional Inherited Members

Detailed Description

Constructor & Destructor Documentation

◆ GPUExecutionPolicy()

Member Function Documentation

◆ add_scaled_vector()

◆ allgather_within_group()

◆ device_backend()

◆ dot_product()

◆ global_number_of_elements()

◆ group_rank_index()

◆ group_size()

◆ is_root()

◆ norm()

◆ number_of_processes()

◆ reduce_within_group() [1/2]

◆ reduce_within_group() [2/2]

Member Data Documentation

◆ device_backend_