Enum CudaError

Namespace

DotCompute.Backends.CUDA.Types.Native

Assembly

DotCompute.Backends.CUDA.dll

Enumeration of CUDA error codes returned by the CUDA runtime API and driver API. These error codes provide detailed information about the success or failure of CUDA operations.

public enum CudaError

Fields

CapturedEvent = 66

The event was captured in a stream.

CompatNotSupportedOnDevice = 58

The operation is not supported on the current device.

ContextIsDestroyed = 202

The context is being or has been destroyed.

[SuppressMessage("Design", "CA1069:Enums values should not be duplicated", Justification = "NVIDIA CUDA API defines both LaunchFailure and Deinitialized with value 4 for historical reasons")] Deinitialized = 4

The driver API context has been deinitialized.

DeviceAlreadyInUse = 33

The device is already in use.

DevicesUnavailable = 27

All CUDA-capable devices are busy or unavailable.

DuplicateSurfaceName = 26

A surface with the same name already exists.

DuplicateTextureName = 25

A texture with the same name already exists.

DuplicateVariableName = 24

A global with the same name already exists.

EccUncorrectable = 214

An uncorrectable ECC error was detected during execution. This indicates a hardware problem that cannot be automatically corrected.

FileNotFound = 50

The file is not found.

GraphExecUpdateFailure = 69

The graph execution update failed.

HardwareStackError = 215

A hardware stack error occurred. This indicates a problem with the hardware stack management system.

HostMemoryAlreadyRegistered = 218

The host memory is already registered with CUDA.

IllegalAddress = 42

An illegal memory access was encountered.

IllegalInstruction = 216

An illegal instruction was encountered during kernel execution. This indicates a problem with the kernel code or compilation.

IncompatibleDriverContext = 30

The driver context is incompatible with the current context.

InitializationError = 3

The CUDA driver and runtime could not be initialized.

InsufficientDriver = 22

The installed CUDA driver is insufficient for the CUDA runtime version.

InvalidChannelDescriptor = 15

The channel descriptor is invalid.

InvalidConfiguration = 8

The launch configuration is invalid.

InvalidContext = 201

The context handle is invalid or has been destroyed.

InvalidDevice = 9

The device ordinal supplied by the user does not correspond to a valid CUDA device.

InvalidDeviceFunction = 7

The device function being invoked is not defined.

InvalidDevicePointer = 12

The specified device pointer is invalid.

InvalidFilterSetting = 17

The filter setting is invalid.

InvalidGraphicsContext = 46

The device kernel source is invalid.

InvalidHandle = 39

The resource handle is invalid.

InvalidHostPointer = 11

The specified host pointer is invalid.

InvalidKernelImage = 28

The device kernel image is invalid.

InvalidMemcpyChannelDesc = 16

The direction of the memory copy is invalid.

InvalidMemcpyDirection = 10

One or more parameters passed to the API call are not valid memory copy directions.

InvalidNormSetting = 18

The normalization setting is invalid.

InvalidPtx = 45

A PTX compilation failed.

InvalidResourceHandle = 54

The resource handle is invalid.

InvalidSource = 49

The device kernel source is invalid.

InvalidSurface = 23

The surface is not bound.

InvalidTexture = 13

The texture is not bound.

InvalidTextureBinding = 14

The texture binding is invalid.

InvalidValue = 1

One or more parameters passed to the API call are not within an acceptable range of values.

JitCompilerNotFound = 48

PTX JIT compiler is not available.

LaunchFailed = 4

Alias for LaunchFailure - An exception occurred on the device while executing a kernel.

LaunchFailure = 4

An exception occurred on the device while executing a kernel.

LaunchIncompatibleTexturing = 43

A kernel launch is being performed with incompatible texturing mode.

LaunchOutOfResources = 6

The kernel launch failed due to insufficient resources.

LaunchTimeout = 5

The kernel launch timed out.

MemoryAllocation = 2

The memory allocation operation failed due to insufficient memory.

MemoryValueTooLarge = 21

The memory value is too large.

MissingConfiguration = 8

Missing launch configuration.

MixedDeviceExecution = 19

The operation attempted to mix device and non-device execution.

NoDevice = 38

No CUDA-capable device is available.

NoKernelImageForDevice = 29

There is no kernel image available that is suitable for the device.

NotFound = 40

The resource was not found.

NotInitialized = 3

The CUDA driver and runtime could not be initialized (alias for InitializationError).

NotPermitted = 800

The operation is not permitted in this context.

NotReady = 41

The resource is not ready for the requested operation.

NotSupported = 55

The operation is not supported.

NotYetImplemented = 20

The operation is not yet implemented.

NvlinkUncorrectable = 47

A hardware error occurred that cannot be corrected.

OperatingSystem = 53

An OS call failed.

[SuppressMessage("Design", "CA1069:Enums values should not be duplicated", Justification = "NVIDIA CUDA API defines both MemoryAllocation and OutOfMemory with value 2 as equivalent error codes")] OutOfMemory = 2

The system is out of memory.

PeerAccessAlreadyEnabled = 31

Peer access has already been enabled from the current device.

PeerAccessNotEnabled = 32

Peer access has not been enabled from the current device.

PeerAccessUnsupported = 44

Peer access is not supported across the given devices.

ProfilerAlreadyStarted = 36

The profiler has already been started.

ProfilerAlreadyStopped = 37

The profiler has already been stopped.

ProfilerDisabled = 34

The profiler is disabled.

ProfilerNotInitialized = 35

The profiler is not initialized.

SharedObjectInitFailed = 52

Initialization of a shared object failed.

SharedObjectSymbolNotFound = 51

A required symbol was not found in the shared object.

StreamCaptureImplicit = 65

An implicit stream capture error occurred.

StreamCaptureInvalidated = 60

The stream capture was invalidated.

StreamCaptureIsolation = 64

A stream capture isolation error occurred.

StreamCaptureMerge = 61

A stream capture merge operation was attempted.

StreamCaptureUnjoined = 63

The stream capture operation was unjoined.

StreamCaptureUnmatched = 62

The stream capture operation was unmatched.

StreamCaptureUnsupported = 59

Stream capture is not supported.

StreamCaptureWrongThread = 67

Stream capture was called from the wrong thread.

Success = 0

The operation completed successfully.

SystemDriverMismatch = 57

The system driver and CUDA driver are not compatible.

SystemNotReady = 56

The system is not ready.

Timeout = 68

The operation timed out.

TimeoutExpired = 68

The operation has timed out (alias for Timeout).

TooManyPeers = 217

Too many peer-to-peer connections have been established.

Unknown = 999

An unknown error occurred.

Remarks

CUDA error codes are used throughout the CUDA APIs to indicate the status of operations. A return value of Success indicates that the operation completed successfully, while any other value indicates a specific type of error or failure condition.