cuda

package

v1.0.0 Latest Latest Go to latest Published: Jul 15, 2025 License: MIT Imports: 4 Imported by: 0

Details

Valid go.mod file
Redistributable license
Tagged version
Stable version
Learn more about best practices

Repository

github.com/BasedCodeCapital/gocuda

Links

Open Source Insights

Documentation ¶

Index ¶

func BatchVectorAdd(operations []BatchOperation) error
func CleanupCUDA()
func FusedVectorMatrix(op *FusedOperation) error
func GetDeviceCount() int
func InitCUDA() error
func LegacyMatrixMultiply(a, b []float32, width int) ([]float32, error)
func LegacyVectorAdd(a, b []float32) ([]float32, error)
func MatrixMultiply(a, b []float32, width int) ([]float32, error)
func PrintDeviceInfo()
func VectorAdd(a, b []float32) ([]float32, error)
type BatchOperation
type DeviceInfo
- func GetDeviceInfo(deviceID int) (*DeviceInfo, error)
type FusedOperation

Constants ¶

This section is empty.

Variables ¶

This section is empty.

Functions ¶

func BatchVectorAdd ¶

func BatchVectorAdd(operations []BatchOperation) error

BatchVectorAdd performs multiple vector additions in a single GPU call

func CleanupCUDA ¶

func CleanupCUDA()

CleanupCUDA cleans up CUDA resources

func FusedVectorMatrix ¶

func FusedVectorMatrix(op *FusedOperation) error

FusedVectorMatrix performs both vector addition and matrix multiplication in a single kernel

func GetDeviceCount ¶

func GetDeviceCount() int

GetDeviceCount returns the number of CUDA devices

func InitCUDA ¶

func InitCUDA() error

InitCUDA initializes CUDA context and memory pool

func LegacyMatrixMultiply ¶

func LegacyMatrixMultiply(a, b []float32, width int) ([]float32, error)

LegacyMatrixMultiply performs matrix multiplication using the original implementation

func LegacyVectorAdd ¶

func LegacyVectorAdd(a, b []float32) ([]float32, error)

LegacyVectorAdd performs vector addition using the original implementation

func MatrixMultiply ¶

func MatrixMultiply(a, b []float32, width int) ([]float32, error)

MatrixMultiply performs matrix multiplication using CUDA (uses original implementation for best performance)

func PrintDeviceInfo ¶

func PrintDeviceInfo()

PrintDeviceInfo prints information about all CUDA devices

func VectorAdd ¶

func VectorAdd(a, b []float32) ([]float32, error)

VectorAdd performs vector addition using CUDA (uses original implementation for best performance)

Types ¶

type BatchOperation ¶

type BatchOperation struct {
	VectorA []float32
	VectorB []float32
	Result  []float32
}

BatchOperation represents a batch of operations

type DeviceInfo ¶

type DeviceInfo struct {
	Name     string
	MemoryMB int
	DeviceID int
}

DeviceInfo represents CUDA device information

func GetDeviceInfo ¶

func GetDeviceInfo(deviceID int) (*DeviceInfo, error)

GetDeviceInfo returns information about a specific CUDA device

type FusedOperation ¶

type FusedOperation struct {
	VectorA     []float32
	VectorB     []float32
	VectorC     []float32
	MatrixA     []float32
	MatrixB     []float32
	MatrixC     []float32
	MatrixWidth int
}

FusedOperation represents a combined vector and matrix operation

Source Files ¶

View all Source files

cuda.go

?	: This menu
/	: Search site
f or F	: Jump to
y or Y	: Canonical URL