xla::HloInstruction Class Reference

#include <hlo_instruction.h>

Collaboration diagram for xla::HloInstruction:

Public Types
enum	FusionKind { , FusionKind::kInput, FusionKind::kOutput, FusionKind::kTransposeDot, FusionKind::kCustom }

Static Public Member Functions
static std::unique_ptr< HloInstruction >	CreateFusion (const Shape &shape, FusionKind fusion_kind, HloInstruction *fused_root)

Detailed Description

HLO instructions are the IR used by the high-level compiler.

Member Enumeration Documentation

FusionKind

enum xla::HloInstruction::FusionKind

strong

FusionKind

Enumerator
kInput	Fused into a loop
kOutput	Op's input is fused into the op itself.
kTransposeDot	Op's output is fused into the op itself. REQUIRES: At least one operand buffer must be able to alias the output buffer.
kCustom	Fused into a dot with transposed operands

Member Function Documentation

CreateFusion()

std::unique_ptr< HloInstruction > xla::HloInstruction::CreateFusion ( const Shape &  shape, FusionKind  fusion_kind, HloInstruction *  fused_root  )

static

Google docs:

Creates a fusion instruction. A fusion instruction contains one or more fused instructions forming an expression with a single root "fused_root". Additional instructions can be added to the fusion instruction with the method xla::HloInstruction::FuseInstruction.

Here is the caller graph for this function:

---

The documentation for this class was generated from the following files:

• tensorflow/tensorflow/compiler/xla/service/hlo_instruction.h
• tensorflow/tensorflow/compiler/xla/service/hlo_instruction.cc