internal.gpu

Standard library module View source
variable CUDA_SUCCESS 
CUDA_SUCCESS: Literal[int] = 0
variable CUDA_ERROR_NOT_FOUND 
CUDA_ERROR_NOT_FOUND: Literal[int] = 500
variable CUresult 
CUresult = i32
variable CUdevice 
CUdevice = i32
variable CUdeviceptr 
CUdeviceptr = u64
variable CUmodule 
CUmodule = cobj
variable CUcontext 
CUcontext = cobj
variable CUfunction 
CUfunction = cobj
variable modules 
modules: List[CUmodule] = []
variable cuCtxCreate 
cuCtxCreate = Function[([Ptr[cobj], u32, CUdevice], CUresult)](cobj())
variable cuMemAlloc 
cuMemAlloc = Function[([Ptr[CUdeviceptr], i64], CUresult)](cobj())
variable cuMemcpyDtoH 
cuMemcpyDtoH = Function[([cobj, CUdeviceptr, u64], CUresult)](cobj())
variable cuMemcpyHtoD 
cuMemcpyHtoD = Function[([CUdeviceptr, cobj, u64], CUresult)](cobj())
variable cuMemFree 
cuMemFree = Function[([CUdeviceptr], CUresult)](cobj())
variable cuDeviceComputeCapability 
cuDeviceComputeCapability = Function[([Ptr[i32], Ptr[i32], CUdevice], CUresult)](cobj())
variable cuDeviceGet 
cuDeviceGet = Function[([Ptr[CUdevice], i32], CUresult)](cobj())
variable cuDeviceGetCount 
cuDeviceGetCount = Function[([Ptr[i32]], CUresult)](cobj())
variable cuDeviceGetName 
cuDeviceGetName = Function[([Ptr[u8], i32, CUdevice], CUresult)](cobj())
variable cuGetErrorName 
cuGetErrorName = Function[([CUresult, Ptr[cobj]], CUresult)](cobj())
variable cuGetErrorString 
cuGetErrorString = Function[([CUresult, Ptr[cobj]], CUresult)](cobj())
variable cuInit 
cuInit = Function[([u32], CUresult)](cobj())
variable cuLaunchKernel 
cuLaunchKernel = Function[([CUfunction, u32, u32, u32, u32, u32, u32, u32, cobj, cobj, cobj], CUresult)](cobj())
variable cuModuleGetFunction 
cuModuleGetFunction = Function[([Ptr[CUfunction], CUmodule, cobj], CUresult)](cobj())
variable cuModuleLoadData 
cuModuleLoadData = Function[([Ptr[CUmodule], cobj], CUresult)](cobj())
class Device
@tuple 
class Device
View complete class API →
Constructors 1
__new__ 
__new__(idx: int)
Properties 1
compute_capability
@property 
compute_capability(self)
Methods 1
count
@staticmethod 
count()
Special methods 3
__str__ 
__str__(self)
__index__ 
__index__(self)
__bool__ 
__bool__(self)
function cuda_init_handles 
cuda_init_handles(cuda_handle: cobj)
function cuda_check 
cuda_check(err)
function nvptx_init 
nvptx_init()
function nvptx_load_module 
nvptx_load_module()
function cuda_init 
cuda_init(debug: bool = False)
class Memory
@tuple 
class Memory[T]
View complete class API →
Fields 1
T type
Constructors 1
__new__ 
__new__(ptr: cobj)
Properties 1
ptr
@property 
ptr(self)
Methods 1
to_ptr
@pure @llvm 
to_ptr(p: CUdeviceptr)

%0 = inttoptr i64 %p to ptr
ret ptr %0
function syncthreads
@llvm 
syncthreads()

declare void @llvm.nvvm.barrier0()
call void @llvm.nvvm.barrier0()
ret {} {}
class Dim3
@tuple 
class Dim3
View complete class API →
Constructors 1
__new__ 
__new__(x: int, y: int, z: int)
Properties 3
x
@property 
x(self)
y
@property 
y(self)
z
@property 
z(self)
class Block
@tuple 
class Block
View complete class API →
Properties 4
x
@property 
x(self)
y
@property 
y(self)
z
@property 
z(self)
dim
@property 
dim(self)
variable thread 
thread = Thread()
variable block 
block = Block()
variable grid 
grid = Grid()
variable warp 
warp = Warp()
function nvptx_function 
nvptx_function(name: Literal[str])
function canonical_dim 
canonical_dim(dim)
function offsets 
offsets(t)
function kernel_wrapper 
kernel_wrapper(*args, grid, block, fn)
function kernel 
kernel(fn)
class Pointer
@tuple 
class Pointer[T]
View complete class API →
Fields 1
T type
Special methods 3
__to_gpu__ 
__to_gpu__(self, cache: AllocCache)
__from_gpu__ 
__from_gpu__(self, other: Ptr[T])
__from_gpu_new__ 
__from_gpu_new__(other: Ptr[T])
function raw 
raw(v)
class Ptr
@extend 
class Ptr[T]
Extends Ptr · View complete class API →
Special methods 3
__to_gpu__ 
__to_gpu__(self, cache: AllocCache)
__from_gpu__ 
__from_gpu__(self, other: Ptr[T])
__from_gpu_new__ 
__from_gpu_new__(other: Ptr[T])
class NoneType
@extend 
class NoneType
Extends NoneType · View complete class API →
Special methods 3
__to_gpu__ 
__to_gpu__(self, cache: AllocCache)
__from_gpu__ 
__from_gpu__(self, other: NoneType)
__from_gpu_new__ 
__from_gpu_new__(other: NoneType)
class float
@extend 
class float
Extends float · View complete class API →
Special methods 3
__to_gpu__ 
__to_gpu__(self, cache: AllocCache)
__from_gpu__ 
__from_gpu__(self, other: float)
__from_gpu_new__ 
__from_gpu_new__(other: float)
class float32
@extend 
class float32
Extends float32 · View complete class API →
Special methods 3
__to_gpu__ 
__to_gpu__(self, cache: AllocCache)
__from_gpu__ 
__from_gpu__(self, other: float32)
__from_gpu_new__ 
__from_gpu_new__(other: float32)
class float16
@extend 
class float16
Extends float16 · View complete class API →
Special methods 3
__to_gpu__ 
__to_gpu__(self, cache: AllocCache)
__from_gpu__ 
__from_gpu__(self, other: float16)
__from_gpu_new__ 
__from_gpu_new__(other: float16)
class bfloat16
@extend 
class bfloat16
Extends bfloat16 · View complete class API →
Special methods 3
__to_gpu__ 
__to_gpu__(self, cache: AllocCache)
__from_gpu__ 
__from_gpu__(self, other: bfloat16)
__from_gpu_new__ 
__from_gpu_new__(other: bfloat16)
class bool
@extend 
class bool
Extends bool · View complete class API →
Special methods 3
__to_gpu__ 
__to_gpu__(self, cache: AllocCache)
__from_gpu__ 
__from_gpu__(self, other: bool)
__from_gpu_new__ 
__from_gpu_new__(other: bool)
class Int
@extend 
class Int[N]
Extends Int · View complete class API →
Special methods 3
__to_gpu__ 
__to_gpu__(self, cache: AllocCache)
__from_gpu__ 
__from_gpu__(self, other: Int[N])
__from_gpu_new__ 
__from_gpu_new__(other: Int[N])
class UInt
@extend 
class UInt[N]
Extends UInt · View complete class API →
Special methods 3
__to_gpu__ 
__to_gpu__(self, cache: AllocCache)
__from_gpu__ 
__from_gpu__(self, other: UInt[N])
__from_gpu_new__ 
__from_gpu_new__(other: UInt[N])
class str
@extend 
class str
Extends str · View complete class API →
Special methods 3
__to_gpu__ 
__to_gpu__(self, cache: AllocCache)
__from_gpu__ 
__from_gpu__(self, other: str)
__from_gpu_new__ 
__from_gpu_new__(other: str)
class bytes
@extend 
class bytes
Extends bytes · View complete class API →
Special methods 3
__to_gpu__ 
__to_gpu__(self, cache: AllocCache)
__from_gpu__ 
__from_gpu__(self, other: bytes)
__from_gpu_new__ 
__from_gpu_new__(other: bytes)
class List
@extend 
class List[T]
Extends List · View complete class API →
Special methods 3
__to_gpu__
@inline 
__to_gpu__(self, cache: AllocCache)
__from_gpu__
@inline 
__from_gpu__(self, other: List[T])
__from_gpu_new__
@inline 
__from_gpu_new__(other: List[T])
class DynamicTuple
@extend 
class DynamicTuple[T]
Extends DynamicTuple · View complete class API →
Special methods 3
__to_gpu__
@inline 
__to_gpu__(self, cache: AllocCache)
__from_gpu__
@inline 
__from_gpu__(self, other: DynamicTuple[T])
__from_gpu_new__
@inline 
__from_gpu_new__(other: DynamicTuple[T])
class OrderedDict
@extend 
class OrderedDict[K, V]
Extends OrderedDict · View complete class API →
Special methods 3
__to_gpu__ 
__to_gpu__(self, cache: AllocCache)
__from_gpu__ 
__from_gpu__(self, other: OrderedDict[K, V])
__from_gpu_new__ 
__from_gpu_new__(other: OrderedDict[K, V])
class Set
@extend 
class Set[K]
Extends Set · View complete class API →
Special methods 3
__to_gpu__ 
__to_gpu__(self, cache: AllocCache)
__from_gpu__ 
__from_gpu__(self, other: Set[K])
__from_gpu_new__ 
__from_gpu_new__(other: Set[K])
class Optional
@extend 
class Optional[T]
Extends Optional · View complete class API →
Special methods 3
__to_gpu__ 
__to_gpu__(self, cache: AllocCache)
__from_gpu__ 
__from_gpu__(self, other: Optional[T])
__from_gpu_new__ 
__from_gpu_new__(other: Optional[T])