MCPcopy Index your code

hub / github.com/NVlabs/cutile-rs / types & classes

Types & classes444 in github.com/NVlabs/cutile-rs

⨍ Functions 3,402◇ Types & classes 444
↓ 89 callersInterfaceScalar
Marker trait for scalar values that can be broadcast to tiles. Auto-implemented for every `ElementType`.
cutile/src/_core.rs:397
↓ 86 callersClassTile
cutile/src/_core.rs:512
↓ 64 callersClassArray
cutile/src/_core.rs:1034
↓ 39 callersEnumType
cutile-ir/src/ir/types.rs:170
↓ 33 callersClassTensorView
A borrowed, reshaped view of a tensor. Created by [`Tensor::view`]. The view borrows the base tensor's device memory with different shape/strides met
cutile/src/tensor.rs:926
↓ 31 callersClassConfig
cutile-benchmarks/paper/sec5_exp1_safety_overhead/diagnostics/gemm_python_swizzle.py:38
↓ 28 callersInterfaceInteger
cutile-compiler/src/compiler/shared_utils.rs:211
↓ 25 callersInterfacePointer
Marker for GPU pointer types. Impl'd for `*mut E` where `E: ElementType`.
cutile/src/_core.rs:421
↓ 10 callersClassDenseElements
cutile-ir/src/ir/attr.rs:171
↓ 9 callersClassValue
cutile-ir/src/ir/value.rs:19
↓ 5 callersInterfaceElementType
Marker trait for valid `Tile<E, …>` / `Tensor<E, …>` element types.
cutile/src/_core.rs:289
↓ 5 callersClassPersistentConfig
cutile-benchmarks/paper/sec5_exp1_safety_overhead/paper/gemm_python_persistent.py:84
↓ 4 callersClassDriverError
cuda-core/src/error.rs:17
↓ 4 callersClassSameElements
cutile-ir/src/ir/attr.rs:109
↓ 4 callersClassf8e4m3fn
cuda-core/src/dtype.rs:30
↓ 3 callersClassBounded
cutile-ir/src/ir/attr.rs:115
↓ 3 callersInterfaceFloat
cutile-compiler/src/compiler/shared_utils.rs:118
↓ 3 callersClassTensor
cutile/src/_core.rs:628
↓ 3 callersClassf4e2m1fn
cuda-core/src/dtype.rs:58
↓ 2 callersClassBlock
cutile-ir/src/ir/block.rs:16
↓ 2 callersClassBlockId
cutile-ir/src/ir/value.rs:51
↓ 2 callersClassDivBy
cutile-ir/src/ir/attr.rs:101
↓ 2 callersClassInferVarId
cutile-compiler/src/passes/type_inference.rs:355
↓ 2 callersClassKernelLaunchError
cutile/src/error.rs:31
↓ 2 callersClassNodeId
cutile-compiler/src/passes/node_ids.rs:26
↓ 2 callersClassOptimizationHints
Collection of optimization hints for kernel compilation, keyed by SM architecture.
cutile-compiler/src/hints.rs:113
↓ 2 callersClassf8e8m0fnu
cuda-core/src/dtype.rs:47
↓ 1 callersClassDevice
cutile/src/_core.rs:136
↓ 1 callersClassOpId
cutile-ir/src/ir/value.rs:47
↓ 1 callersClassRegionId
cutile-ir/src/ir/value.rs:55
↓ 1 callersClassTensorError
cutile/src/error.rs:18
↓ 1 callersClassf8e5m2
cuda-core/src/dtype.rs:38
↓ 1 callersClassi4
cuda-core/src/dtype.rs:120
↓ 1 callersClasstf32
cuda-core/src/dtype.rs:23
ClassAcqRel
cutile/src/_core.rs:108
ClassAcquire
cutile/src/_core.rs:106
ClassAdd
cutile/src/_core.rs:71
ClassAddF
cutile/src/_core.rs:72
InterfaceAddF
cutile/tests/trait_dispatch_probe.rs:107
ClassAliasSubstitution
cutile-compiler/src/type_aliases.rs:23
ClassAnd
cutile/src/_core.rs:73
ClassAndThen
cuda-async/src/device_operation.rs:615
ClassAndThenWithContext
cuda-async/src/device_operation.rs:1350
ClassApiSpec
cuda-bindings/build.rs:14
ClassApprox
cutile/src/_core.rs:49
InterfaceArcKernelArgument
A kernel argument that can be pushed from an `Arc` reference.
cuda-async/src/launch.rs:117
EnumArrayLikeExpr
cutile-compiler/src/passes/type_inference.rs:485
ClassAssocConstHost
cutile/tests/type_inference_sanity.rs:42
ClassAsyncDeviceContext
Per-device state: GPU device, scheduling policy, and compiled kernel cache. Each GPU device has one `AsyncDeviceContext` stored in a thread-local map
cuda-async/src/device_context.rs:84
ClassAsyncDeviceContexts
cuda-async/src/device_context.rs:96
ClassAsyncKernelLaunch
cuda-async/src/launch.rs:23
EnumAtomicMode
Supported atomic read-modify-write modes.
cutile-compiler/src/compiler/shared_utils.rs:40
InterfaceAtomicMode
`atomic_*_tko` ops: `Relaxed`, `Acquire`, `Release`, `AcqRel`.
cutile/src/_core.rs:102
EnumAtomicRMWMode
cutile-ir/src/ir/attr.rs:64
EnumAttribute
cutile-ir/src/ir/attr.rs:135
EnumAttributeTag
cutile-ir/src/bytecode/enums.rs:123
InterfaceAutorefTrait
cutile/tests/type_inference_sanity.rs:23
EnumBlockTerminator
cutile-compiler/src/compiler/shared_types.rs:41
ClassBoundedPartition
cutile/src/_core.rs:905
ClassBounds
cutile-compiler/src/bounds.rs:107
ClassBoxedDeviceOp
A type-erased [`DeviceOp`] that boxes the execution closure. Created via [`DeviceOp::boxed()`]. Useful when you need to store operations with differe
cuda-async/src/device_operation.rs:418
EnumBranchTerm
cutile-compiler/src/passes/type_inference.rs:372
InterfaceBroadcastScalar
cutile/src/_core.rs:403
InterfaceByteElement
Rust byte marker for packed Tile IR sub-byte values. `u8` is the low-level interop path for raw bytes. `f4e2m1fnx2` is the typed packed storage helpe
cutile/src/_core.rs:297
ClassBytecodeVersion
cutile-ir/src/bytecode/enums.rs:147
ClassCGAParameter
A const generic array parameter with a concrete length.
cutile-compiler/src/syn_utils.rs:424
ClassCUDATileFunctionCompiler
Compiles a single Rust function into Tile IR bytecode.
cutile-compiler/src/compiler/_function.rs:43
ClassCUDATileModules
Aggregated index of all DSL modules, types, impls, and functions. Internally delegates to [`NameResolver`] for all item storage and name resolution.
cutile-compiler/src/compiler/_module.rs:33
EnumCUctx_st
cutile-examples/examples/cudarc_interop.rs:41
EnumCUstream_st
cutile-examples/examples/cudarc_interop.rs:42
ClassCallFinder
cutile/tests/compile_error_quality.rs:166
ClassCallSiteSpanSetter
Rewrites every span in a syn AST node to a fixed target span. When inlining library/core functions, the callee body's spans point into the core modul
cutile-compiler/src/compiler/compile_inline.rs:64
ClassCgaInfo
cutile-macro/src/shadow_dispatch.rs:106
EnumCgaRole
cutile-macro/src/shadow_dispatch.rs:1620
ClassClosureInfo
cutile-compiler/src/syn_utils.rs:804
ClassClosureParam
cutile-compiler/src/syn_utils.rs:797
ClassClosureSignature
cutile-compiler/src/passes/type_inference.rs:216
EnumCmd
cutile-benchmarks/paper/app_exp1_work_distribution/bimodal_gemms_rust/src/main.rs:286
EnumComparisonOrdering
cutile-ir/src/ir/attr.rs:46
EnumComparisonPredicate
cutile-ir/src/ir/attr.rs:53
ClassCompileArtifacts
Compiled kernel artifacts: IR and bytecode. Produced by [`KernelCompiler::compile`]. All methods are pure Rust and do not require a GPU or CUDA drive
cutile-compiler/src/compile_api.rs:36
ClassCompileOptions
cutile-compiler/src/hints.rs:85
ClassCompilerContext
cutile-compiler/src/compiler/_value.rs:497
ClassConfig
cutile-examples/examples/cuda_graphs.rs:99
ClassConfig
cutile-examples/examples/cuda_graphs_deviceop.rs:100
ClassConstParameter
A scalar const generic parameter (e.g. `const N: i32`).
cutile-compiler/src/syn_utils.rs:432
ClassConstPathNormalizer
cutile-compiler/src/type_aliases.rs:150
ClassConstantManager
cutile-ir/src/bytecode/writer.rs:451
EnumContiguousLayout
cutile-kernels/src/experimental/kvbm.rs:370
ClassCoord2
cutile/src/_core.rs:774
ClassCopyDeviceToDevice
Device operation for copying a tensor within GPU memory. This internal type implements the async copy operation that allocates new GPU memory and cop
cutile/src/api.rs:155
ClassCopyDeviceToHostVec
Device operation for copying a tensor from GPU to CPU as a Vec. This internal type implements the async copy operation that transfers data from GPU m
cutile/src/api.rs:318
ClassCopyHostVecToDevice
cutile/src/api.rs:378
ClassCounter
cutile-compiler/src/passes/node_ids.rs:258
ClassCudaGraph
A captured and instantiated CUDA graph, ready for replay. Created via [`CudaGraph::capture`], which executes a [`DeviceOp`] once on a capture stream,
cuda-async/src/cuda_graph.rs:45
ClassCutileModuleEntry
Registry entry for a single cuTile module. Every `#[cutile::module]` emits a `static` of this type into [`CUTILE_MODULES`]. The JIT iterates the slic
cutile-compiler/src/registry.rs:38
ClassDICompileUnit
cutile-ir/src/ir/location.rs:51
ClassDIFile
cutile-ir/src/ir/location.rs:45
ClassDILexicalBlock
cutile-ir/src/ir/location.rs:66
ClassDISubprogram
cutile-ir/src/ir/location.rs:56
next →1–100 of 444, ranked by callers