CodegenSplitReduction

Helpers & Utilities · AiDotNet.Tensors.Engines.Compilation.Codegen.Ir

No summary documentation available yet.

Methods

Method Summary
CodegenSplitPlan TryPlan(CodegenKernelSpec spec, int multiprocessors, int blockThreads)
CodegenSplitPlan TryPlanChunked(CodegenKernelSpec spec, int splitFactor, int multiprocessors, int blockThreads)
CodegenSplitPlan TryPlanChunkedForTiles(CodegenKernelSpec spec, int splitFactor)
IReadOnlyList<int> ChooseAxes(CodegenKernelSpec spec, int multiprocessors, int blockThreads)
ValueTuple<CodegenKernelSpec, CodegenKernelSpec> Split(CodegenKernelSpec spec, IReadOnlyList<int> reductionAxes)
ValueTuple<CodegenKernelSpec, CodegenKernelSpec> Split(CodegenKernelSpec spec, int reductionAxis)
ValueTuple<CodegenKernelSpec, CodegenKernelSpec> SplitChunked(CodegenKernelSpec spec, int reductionAxis, int splitFactor)
bool IsSplittable(CodegenKernelSpec spec)
bool IsSplittable(CodegenReduceKind reduce)
int ChooseAxis(CodegenKernelSpec spec, int multiprocessors, int blockThreads)
int[] CombineOperandSources(CodegenKernelSpec spec)
int[] PartialOperandSources(CodegenKernelSpec spec)