MLIR Language Reference
MLIR (Multi-Level IR) is a compiler intermediate representation with similarities to traditional three-address SSA representations (like LLVM IR or SIL), but which introduces notions from polyhedral loop optimization as first-class concepts. This hybrid design is optimized to represent, analyze, and transform high level dataflow graphs as well as target-specific code generated for high performance data parallel systems. Beyond its representational capabilities, its single continuous design provides a framework to lower from dataflow graphs to high-performance target-specific code.
This document defines and describes the key concepts in MLIR, and is intended to be a dry reference document - the rationale documentation, glossary, and other content are hosted elsewhere.
MLIR is designed to be used in three different forms: a human-readable textual form suitable for debugging, an in-memory form suitable for programmatic transformations and analysis, and a compact serialized form suitable for storage and transport. The different forms all describe the same semantic content. This document describes the human-readable textual form.
High-Level Structure
MLIR is fundamentally based on a graph-like data structure of nodes, called Operations, and edges, called Values. Each Value is the result of exactly one Operation or Block Argument, and has a Value Type defined by the type system. Operations are contained in Blocks and Blocks are contained in Regions. Operations are also ordered within their containing block and Blocks are ordered in their containing region, although this order may or may not be semantically meaningful in a given kind of region). Operations may also contain regions, enabling hierarchical structures to be represented.
Operations can represent many different concepts, from higher-level concepts like function definitions, function calls, buffer allocations, view or slices of buffers, and process creation, to lower-level concepts like target-independent arithmetic, target-specific instructions, configuration registers, and logic gates. These different concepts are represented by different operations in MLIR and the set of operations usable in MLIR can be arbitrarily extended.
MLIR also provides an extensible framework for transformations on operations, using familiar concepts of compiler Passes. Enabling an arbitrary set of passes on an arbitrary set of operations results in a significant scaling challenge, since each transformation must potentially take into account the semantics of any operation. MLIR addresses this complexity by allowing operation semantics to be described abstractly using Traits and Interfaces, enabling transformations to operate on operations more generically. Traits often describe verification constraints on valid IR, enabling complex invariants to be captured and checked. (see Op vs Operation)
One obvious application of MLIR is to represent an SSA-based IR, like the LLVM core IR, with appropriate choice of Operation Types to define Modules, Functions, Branches, Allocations, and verification constraints to ensure the SSA Dominance property. MLIR includes a ‘standard’ dialect which defines just such structures. However, MLIR is intended to be general enough to represent other compiler-like data structures, such as Abstract Syntax Trees in a language frontend, generated instructions in a target-specific backend, or circuits in a High-Level Synthesis tool.
Here's an example of an MLIR module:
// Compute A*B using an implementation of multiply kernel and print the // result using a TensorFlow op. The dimensions of A and B are partially // known. The shapes are assumed to match. func @mul(%A: tensor<100x?xf32>, %B: tensor<?x50xf32>) -> (tensor<100x50xf32>) { // Compute the inner dimension of %A using the dim operation. %n = dim %A, 1 : tensor<100x?xf32> // Allocate addressable "buffers" and copy tensors %A and %B into them. %A_m = alloc(%n) : memref<100x?xf32> tensor_store %A to %A_m : memref<100x?xf32> %B_m = alloc(%n) : memref<?x50xf32> tensor_store %B to %B_m : memref<?x50xf32> // Call function @multiply passing memrefs as arguments, // and getting returned the result of the multiplication. %C_m = call @multiply(%A_m, %B_m) : (memref<100x?xf32>, memref<?x50xf32>) -> (memref<100x50xf32>) dealloc %A_m : memref<100x?xf32> dealloc %B_m : memref<?x50xf32> // Load the buffer data into a higher level "tensor" value. %C = tensor_load %C_m : memref<100x50xf32> dealloc %C_m : memref<100x50xf32> // Call TensorFlow built-in function to print the result tensor. "tf.Print"(%C){message: "mul result"} : (tensor<100x50xf32) -> (tensor<100x50xf32>) return %C : tensor<100x50xf32> } // A function that multiplies two memrefs and returns the result. func @multiply(%A: memref<100x?xf32>, %B: memref<?x50xf32>) -> (memref<100x50xf32>) { // Compute the inner dimension of %A. %n = dim %A, 1 : memref<100x?xf32> // Allocate memory for the multiplication result. %C = alloc() : memref<100x50xf32> // Multiplication loop nest. affine.for %i = 0 to 100 { affine.for %j = 0 to 50 { store 0 to %C[%i, %j] : memref<100x50xf32> affine.for %k = 0 to %n { %a_v = load %A[%i, %k] : memref<100x?xf32> %b_v = load %B[%k, %j] : memref<?x50xf32> %prod = mulf %a_v, %b_v : f32 %c_v = load %C[%i, %j] : memref<100x50xf32> %sum = addf %c_v, %prod : f32 store %sum, %C[%i, %j] : memref<100x50xf32> } } } return %C : memref<100x50xf32> }
Notation
MLIR has a simple and unambiguous grammar, allowing it to reliably round-trip through a textual form. This is important for development of the compiler - e.g. for understanding the state of code as it is being transformed and writing test cases.
This document describes the grammar using Extended Backus-Naur Form (EBNF).
This is the EBNF grammar used in this document, presented in yellow boxes.
alternation ::= expr0 | expr1 | expr2 // Either expr0 or expr1 or expr2. sequence ::= expr0 expr1 expr2 // Sequence of expr0 expr1 expr2. repetition0 ::= expr* // 0 or more occurrences. repetition1 ::= expr+ // 1 or more occurrences. optionality ::= expr? // 0 or 1 occurrence. grouping ::= (expr) // Everything inside parens is grouped together. literal ::= `abcd` // Matches the literal `abcd`.
Code examples are presented in blue boxes.
// This is an example use of the grammar above: // This matches things like: ba, bana, boma, banana, banoma, bomana... example ::= `b` (`an` | `om`)* `a`
Common syntax
The following core grammar productions are used in this document:
// TODO: Clarify the split between lexing (tokens) and parsing (grammar). digit ::= [0-9] hex_digit ::= [0-9a-fA-F] letter ::= [a-zA-Z] id-punct ::= [$._-] integer-literal ::= decimal-literal | hexadecimal-literal decimal-literal ::= digit+ hexadecimal-literal ::= `0x` hex_digit+ float-literal ::= [-+]?[0-9]+[.][0-9]*([eE][-+]?[0-9]+)? string-literal ::= `"` [^"\n\f\v\r]* `"` TODO: define escaping rules
Not listed here, but MLIR does support comments. They use standard BCPL syntax, starting with a // and going until the end of the line.
Identifiers and keywords
Syntax:
// Identifiers bare-id ::= (letter|[_]) (letter|digit|[_$.])* bare-id-list ::= bare-id (`,` bare-id)* value-id ::= `%` suffix-id suffix-id ::= (digit+ | ((letter|id-punct) (letter|id-punct|digit)*)) symbol-ref-id ::= `@` (suffix-id | string-literal) value-id-list ::= value-id (`,` value-id)* // Uses of value, e.g. in an operand list to an operation. value-use ::= value-id value-use-list ::= value-use (`,` value-use)*
Identifiers name entities such as values, types and functions, and are chosen by the writer of MLIR code. Identifiers may be descriptive (e.g. %batch_size, @matmul), or may be non-descriptive when they are auto-generated (e.g. %23, @func42). Identifier names for values may be used in an MLIR text file but are not persisted as part of the IR - the printer will give them anonymous names like %42.
MLIR guarantees identifiers never collide with keywords by prefixing identifiers with a sigil (e.g. %, #, @, ^, !). In certain unambiguous contexts (e.g. affine expressions), identifiers are not prefixed, for brevity. New keywords may be added to future versions of MLIR without danger of collision with existing identifiers.
Value identifiers are only in scope for the (nested) region in which they are defined and cannot be accessed or referenced outside of that region. Argument identifiers in mapping functions are in scope for the mapping body. Particular operations may further limit which identifiers are in scope in their regions. For instance, the scope of values in a region with SSA control flow semantics is constrained according to the standard definition of SSA dominance. Another example is the IsolatedFromAbove trait, which restricts directly accessing values defined in containing regions.
Function identifiers and mapping identifiers are associated with Symbols and have scoping rules dependent on symbol attributes.
Dialects
Dialects are the mechanism by which to engage with and extend the MLIR ecosystem. They allow for defining new operations, as well as attributes and types. Each dialect is given a unique namespace that is prefixed to each defined attribute/operation/type. For example, the Affine dialect defines the namespace: affine.
MLIR allows for multiple dialects, even those outside of the main tree, to co-exist together within one module. Dialects are produced and consumed by certain passes. MLIR provides a framework to convert between, and within, different dialects.
A few of the dialects supported by MLIR:
Target specific operations
Dialects provide a modular way in which targets can expose target-specific operations directly through to MLIR. As an example, some targets go through LLVM. LLVM has a rich set of intrinsics for certain target-independent operations (e.g. addition with overflow check) as well as providing access to target-specific operations for the targets it supports (e.g. vector permutation operations). LLVM intrinsics in MLIR are represented via operations that start with an “llvm.” name.
Example:
// LLVM: %x = call {i16, i1} @llvm.sadd.with.overflow.i16(i16 %a, i16 %b) %x:2 = "llvm.sadd.with.overflow.i16"(%a, %b) : (i16, i16) -> (i16, i1)
These operations only work when targeting LLVM as a backend (e.g. for CPUs and GPUs), and are required to align with the LLVM definition of these intrinsics.
Operations
Syntax:
operation ::= op-result-list? (generic-operation | custom-operation)
trailing-location?
generic-operation ::= string-literal `(` value-use-list? `)` successor-list?
(`(` region-list `)`)? attribute-dict? `:` function-type
custom-operation ::= bare-id custom-operation-format
op-result-list ::= op-result (`,` op-result)* `=`
op-result ::= value-id (`:` integer-literal)
successor-list ::= successor (`,` successor)*
successor ::= caret-id (`:` bb-arg-list)?
region-list ::= region (`,` region)*
trailing-location ::= (`loc` `(` location `)`)?
MLIR introduces a uniform concept called operations to enable describing many different levels of abstractions and computations. Operations in MLIR are fully extensible (there is no fixed list of operations) and have application-specific semantics. For example, MLIR supports target-independent operations, affine operations, and target-specific machine operations.
The internal representation of an operation is simple: an operation is identified by a unique string (e.g. dim, tf.Conv2d, x86.repmovsb, ppc.eieio, etc), can return zero or more results, take zero or more operands, may have zero or more attributes, may have zero or more successors, and zero or more enclosed regions. The generic printing form includes all these elements literally, with a function type to indicate the types of the results and operands.
Example:
// An operation that produces two results. // The results of %result can be accessed via the <name> `#` <opNo> syntax. %result:2 = "foo_div"() : () -> (f32, i32) // Pretty form that defines a unique name for each result. %foo, %bar = "foo_div"() : () -> (f32, i32) // Invoke a TensorFlow function called tf.scramble with two inputs // and an attribute "fruit". %2 = "tf.scramble"(%result#0, %bar) {fruit: "banana"} : (f32, i32) -> f32
In addition to the basic syntax above, dialects may register known operations. This allows those dialects to support custom assembly form for parsing and printing operations. In the operation sets listed below, we show both forms.
Terminator Operations
These are a special category of operations that must terminate a block, e.g. branches. These operations may also have a list of successors (blocks and their arguments).
Example:
// Branch to ^bb1 or ^bb2 depending on the condition %cond. // Pass value %v to ^bb2, but not to ^bb1. "cond_br"(%cond)[^bb1, ^bb2(%v : index)] : (i1) -> ()
Module
module ::= `module` symbol-ref-id? (`attributes` attribute-dict)? region
An MLIR Module represents a top-level container operation. It contains a single SSACFG region containing a single block which can contain any operations. Operations within this region cannot implicitly capture values defined outside the module, i.e. Modules are IsolatedFromAbove. Modules have an optional symbol name which can be used to refer to them in operations.
Functions
An MLIR Function is an operation with a name containing a single SSACFG region. Operations within this region cannot implicitly capture values defined outside of the function, i.e. Functions are IsolatedFromAbove. All external references must use function arguments or attributes that establish a symbolic connection (e.g. symbols referenced by name via a string attribute like SymbolRefAttr):
function ::= `func` function-signature function-attributes? function-body?
function-signature ::= symbol-ref-id `(` argument-list `)`
(`->` function-result-list)?
argument-list ::= (named-argument (`,` named-argument)*) | /*empty*/
argument-list ::= (type attribute-dict? (`,` type attribute-dict?)*) | /*empty*/
named-argument ::= value-id `:` type attribute-dict?
function-result-list ::= function-result-list-parens
| non-function-type
function-result-list-parens ::= `(` `)`
| `(` function-result-list-no-parens `)`
function-result-list-no-parens ::= function-result (`,` function-result)*
function-result ::= type attribute-dict?
function-attributes ::= `attributes` attribute-dict
function-body ::= region
An external function declaration (used when referring to a function declared in some other module) has no body. While the MLIR textual form provides a nice inline syntax for function arguments, they are internally represented as “block arguments” to the first block in the region.
Only dialect attribute names may be specified in the attribute dictionaries for function arguments, results, or the function itself.
Examples:
// External function definitions. func @abort() func @scribble(i32, i64, memref<? x 128 x f32, #layout_map0>) -> f64 // A function that returns its argument twice: func @count(%x: i64) -> (i64, i64) attributes {fruit: "banana"} { return %x, %x: i64, i64 } // A function with an argument attribute func @example_fn_arg(%x: i32 {swift.self = unit}) // A function with a result attribute func @example_fn_result() -> (f64 {dialectName.attrName = 0 : i64}) // A function with an attribute func @example_fn_attr() attributes {dialectName.attrName = false}
Blocks
Syntax:
block ::= block-label operation+ block-label ::= block-id block-arg-list? `:` block-id ::= caret-id caret-id ::= `^` suffix-id value-id-and-type ::= value-id `:` type // Non-empty list of names and types. value-id-and-type-list ::= value-id-and-type (`,` value-id-and-type)* block-arg-list ::= `(` value-id-and-type-list? `)`
A Block is an ordered list of operations, concluding with a single terminator operation. In SSACFG regions, each block represents a compiler basic block where instructions inside the block are executed in order and terminator operations implement control flow branches between basic blocks.
Blocks in MLIR take a list of block arguments, notated in a function-like way. Block arguments are bound to values specified by the semantics of individual operations. Block arguments of the entry block of a region are also arguments to the region and the values bound to these arguments are determined by the semantics of the containing operation. Block arguments of other blocks are determined by the semantics of terminator operations, e.g. Branches, which have the block as a successor. In regions with control flow, MLIR leverages this structure to implicitly represent the passage of control-flow dependent values without the complex nuances of PHI nodes in traditional SSA representations. Note that values which are not control-flow dependent can be referenced directly and do not need to be passed through block arguments.
Here is a simple example function showing branches, returns, and block arguments:
func @simple(i64, i1) -> i64 { ^bb0(%a: i64, %cond: i1): // Code dominated by ^bb0 may refer to %a cond_br %cond, ^bb1, ^bb2 ^bb1: br ^bb3(%a: i64) // Branch passes %a as the argument ^bb2: %b = addi %a, %a : i64 br ^bb3(%b: i64) // Branch passes %b as the argument // ^bb3 receives an argument, named %c, from predecessors // and passes it on to bb4 along with %a. %a is referenced // directly from its defining operation and is not passed through // an argument of ^bb3. ^bb3(%c: i64): br ^bb4(%c, %a : i64, i64) ^bb4(%d : i64, %e : i64): %0 = addi %d, %e : i64 return %0 : i64 // Return is also a terminator. }
Context: The “block argument” representation eliminates a number of special cases from the IR compared to traditional “PHI nodes are operations” SSA IRs (like LLVM). For example, the parallel copy semantics of SSA is immediately apparent, and function arguments are no longer a special case: they become arguments to the entry block [more rationale]. Blocks are also a fundamental concept that cannot be represented by operations because values defined in an operation cannot be accessed outside the operation.
Regions
Definition
A region is an ordered list of MLIR Blocks. The semantics within a region is not imposed by the IR. Instead, the containing operation defines the semantics of the regions it contains. MLIR currently defines two kinds of regions: SSACFG regions, which describe control flow between blocks, and Graph regions, which do not require control flow between block. The kinds of regions within an operation are described using the RegionKindInterface.
Regions do not have a name or an address, only the blocks contained in a region do. Regions must be contained within operations and have no type or attributes. The first block in the region is a special block called the ‘entry block’. The arguments to the entry block are also the arguments of the region itself. The entry block cannot be listed as a successor of any other block. The syntax for a region is as follows:
region ::= `{` block* `}`
A function body is an example of a region: it consists of a CFG of blocks and has additional semantic restrictions that other types of regions may not have. For example, in a function body, block terminators must either branch to a different block, or return from a function where the types of the return arguments must match the result types of the function signature. Similarly, the function arguments must match the types and count of the region arguments. In general, operations with regions can define these correspondances arbitrarily.
Value Scoping
Regions provide hierarchical encapsulation of programs: it is impossible to reference, i.e. branch to, a block which is not in the same region as the source of the reference, i.e. a terminator operation. Similarly, regions provides a natural scoping for value visibility: values defined in a region don't escape to the enclosing region, if any. By default, operations inside a region can reference values defined outside of the region whenever it would have been legal for operands of the enclosing operation to reference those values, but this can be restricted using traits, such as OpTrait::IsolatedFromAbove, or a custom verifier.
Example:
"any_op"(%a) ({ // if %a is in-scope in the containing region... // then %a is in-scope here too. %new_value = "another_op"(%a) : (i64) -> (i64) }) : (i64) -> (i64)
MLIR defines a generalized ‘hierarchical dominance’ concept that operates across hierarchy and defines whether a value is ‘in scope’ and can be used by a particular operation. Whether a value can be used by another operation in the same region is defined by the kind of region. A value defined in a region can be used by an operation which has a parent in the same region, if and only if the parent could use the value. A value defined by an argument to a region can always be used by any operation deeply contained in the region. A value defined in a region can never be used outside of the region.
Control Flow and SSACFG Regions
In MLIR, control flow semantics of a region is indicated by RegionKind::SSACFG. Informally, these regions support semantics where operations in a region ‘execute sequentially’. Before an operation executes, its operands have well-defined values. After an operation executes, the operands have the same values and results also have well-defined values. After an operation executes, the next operation in the block executes until the operation is the terminator operation at the end of a block, in which case some other operation will execute. The determination of the next instruction to execute is the ‘passing of control flow’.
In general, when control flow is passed to an operation, MLIR does not restrict when control flow enters or exits the regions contained in that operation. However, when control flow enters a region, it always begins in the first block of the region, called the entry block. Terminator operations ending each block represent control flow by explicitly specifying the successor blocks of the block. Control flow can only pass to one of the specified successor blocks as in a branch operation, or back to the containing operation as in a return operation. Terminator operations without successors can only pass control back to the containing operation. Within these restrictions, the particular semantics of terminator operations is determined by the specific dialect operations involved. Blocks (other than the entry block) that are not listed as a successor of a terminator operation are defined to be unreachable and can be removed without affecting the semantics of the containing operation.
Although control flow always enters a region through the entry block, control flow may exit a region through any block with an appropriate terminator. The standard dialect leverages this capability to define operations with Single-Entry-Multiple-Exit (SEME) regions, possibly flowing through different blocks in the region and exiting through any block with a return operation. This behavior is similar to that of a function body in most programming languages. In addition, control flow may also not reach the end of a block or region, for example if a function call does not return.
Example:
func @accelerator_compute(i64, i1) -> i64 { // An SSACFG region ^bb0(%a: i64, %cond: i1): // Code dominated by ^bb0 may refer to %a cond_br %cond, ^bb1, ^bb2 ^bb1: // This def for %value does not dominate ^bb2 %value = "op.convert"(%a) : (i64) -> i64 br ^bb3(%a: i64) // Branch passes %a as the argument ^bb2: accelerator.launch() { // An SSACFG region ^bb0: // Region of code nested under "accelerator.launch", it can reference %a but // not %value. %new_value = "accelerator.do_something"(%a) : (i64) -> () } // %new_value cannot be referenced outside of the region ^bb3: ... }
Operations with Multiple Regions
An operation containing multiple regions also completely determines the semantics of those regions. In particular, when control flow is passed to an operation, it may transfer control flow to any contained region. When control flow exits a region and is returned to the containing operation, the containing operation may pass control flow to any region in the same operation. An operation may also pass control flow to multiple contained regions concurrently. An operation may also pass control flow into regions that were specified in other operations, in particular those that defined the values or symbols the given operation uses as in a call operation. This passage of control is generally independent of passage of control flow through the basic blocks of the containing region.
Closure
Regions allow defining an operation that creates a closure, for example by “boxing” the body of the region into a value they produce. It remains up to the operation to define its semantics. Note that if an operation triggers asynchronous execution of the region, it is under the responsibility of the operation caller to wait for the region to be executed guaranteeing that any directly used values remain live.
Graph Regions
In MLIR, graph-like semantics in a region is indicated by RegionKind::Graph. Graph regions are appropriate for concurrent semantics without control flow, or for modeling generic directed graph data structures. Graph regions are appropriate for representing cyclic relationships between coupled values where there is no fundamental order to the relationships. For instance, operations in a graph region may represent independent threads of control with values representing streams of data. As usual in MLIR, the particular semantics of a region is completely determined by its containing operation. Graph regions may only contain a single basic block (the entry block).
Rationale: Currently graph regions are arbitrarily limited to a single basic block, although there is no particular semantic reason for this limitation. This limitation has been added to make it easier to stabilize the pass infrastructure and commonly used passes for processing graph regions to properly handle feedback loops. Multi-block regions may be allowed in the future if use cases that require it arise.
In graph regions, MLIR operations naturally represent nodes, while each MLIR value represents a multi-edge connecting a single source node and multiple destination nodes. All values defined in the region as results of operations are in scope within the region and can be accessed by any other operation in the region. In graph regions, the order of operations within a block and the order of blocks in a region is not semantically meaningful and non-terminator operations may be freely reordered, for instance, by canonicalization. Other kinds of graphs, such as graphs with multiple source nodes and multiple destination nodes, can also be represented by representing graph edges as MLIR operations.
Note that cycles can occur within a single block in a graph region, or between basic blocks.
"test.graph_region"() ({ // A Graph region %1 = "op1"(%1, %3) : (i32, i32) -> (i32) // OK: %1, %3 allowed here %2 = "test.ssacfg_region"() ({ %5 = "op2"(%1, %2, %3, %4) : (i32, i32, i32, i32) -> (i32) // OK: %1, %2, %3, %4 all defined in the containing region }) : () -> (i32) %3 = "op2"(%1, %4) : (i32, i32) -> (i32) // OK: %4 allowed here %4 = "op3"(%1) : (i32) -> (i32) }) : () -> ()
Arguments and Results
The arguments of the first block of a region are treated as arguments of the region. The source of these arguments is defined by the semantics of the parent operation. They may correspond to some of the values the operation itself uses.
Regions produce a (possibly empty) list of values. The operation semantics defines the relation between the region results and the operation results.
Type System
Each value in MLIR has a type defined by the type system below. There are a number of primitive types (like integers) and also aggregate types for tensors and memory buffers. MLIR builtin types do not include structures, arrays, or dictionaries.
MLIR has an open type system (i.e. there is no fixed list of types), and types may have application-specific semantics. For example, MLIR supports a set of dialect types.
type ::= type-alias | dialect-type | builtin-type
type-list-no-parens ::= type (`,` type)*
type-list-parens ::= `(` `)`
| `(` type-list-no-parens `)`
// This is a common way to refer to a value with a specified type.
ssa-use-and-type ::= ssa-use `:` type
// Non-empty list of names and types.
ssa-use-and-type-list ::= ssa-use-and-type (`,` ssa-use-and-type)*
Type Aliases
type-alias-def ::= '!' alias-name '=' 'type' type type-alias ::= '!' alias-name
MLIR supports defining named aliases for types. A type alias is an identifier that can be used in the place of the type that it defines. These aliases must be defined before their uses. Alias names may not contain a ‘.’, since those names are reserved for dialect types.
Example:
!avx_m128 = type vector<4 x f32> // Using the original type. "foo"(%x) : vector<4 x f32> -> () // Using the type alias. "foo"(%x) : !avx_m128 -> ()
Dialect Types
Similarly to operations, dialects may define custom extensions to the type system.
dialect-namespace ::= bare-id
opaque-dialect-item ::= dialect-namespace '<' string-literal '>'
pretty-dialect-item ::= dialect-namespace '.' pretty-dialect-item-lead-ident
pretty-dialect-item-body?
pretty-dialect-item-lead-ident ::= '[A-Za-z][A-Za-z0-9._]*'
pretty-dialect-item-body ::= '<' pretty-dialect-item-contents+ '>'
pretty-dialect-item-contents ::= pretty-dialect-item-body
| '(' pretty-dialect-item-contents+ ')'
| '[' pretty-dialect-item-contents+ ']'
| '{' pretty-dialect-item-contents+ '}'
| '[^[<({>\])}\0]+'
dialect-type ::= '!' opaque-dialect-item
dialect-type ::= '!' pretty-dialect-item
Dialect types can be specified in a verbose form, e.g. like this:
// LLVM type that wraps around llvm IR types. !llvm<"i32*"> // Tensor flow string type. !tf.string // Complex type !foo<"something<abcd>"> // Even more complex type !foo<"something<a%%123^^^>>>">
Dialect types that are simple enough can use the pretty format, which is a lighter weight syntax that is equivalent to the above forms:
// Tensor flow string type. !tf.string // Complex type !foo.something<abcd>
Sufficiently complex dialect types are required to use the verbose form for generality. For example, the more complex type shown above wouldn't be valid in the lighter syntax: !foo.something<a%%123^^^>>> because it contains characters that are not allowed in the lighter syntax, as well as unbalanced <> characters.
See here to learn how to define dialect types.
Builtin Types
Builtin types are a core set of dialect types that are defined in a builtin dialect and thus available to all users of MLIR.
builtin-type ::= complex-type
| float-type
| function-type
| index-type
| integer-type
| memref-type
| none-type
| tensor-type
| tuple-type
| vector-type
Complex Type
Syntax:
complex-type ::= `complex` `<` type `>`
The value of complex type represents a complex number with a parameterized element type, which is composed of a real and imaginary value of that element type. The element must be a floating point or integer scalar type.
Examples:
complex<f32> complex<i32>
Floating Point Types
Syntax:
// Floating point. float-type ::= `f16` | `bf16` | `f32` | `f64` | `f80` | `f128`
MLIR supports float types of certain widths that are widely used as indicated above.
Function Type
Syntax:
// MLIR functions can return multiple values.
function-result-type ::= type-list-parens
| non-function-type
function-type ::= type-list-parens `->` function-result-type
MLIR supports first-class functions: for example, the constant operation produces the address of a function as a value. This value may be passed to and returned from functions, merged across control flow boundaries with block arguments, and called with the call_indirect operation.
Function types are also used to indicate the arguments and results of operations.
Index Type
Syntax:
// Target word-sized integer. index-type ::= `index`
The index type is a signless integer whose size is equal to the natural machine word of the target (rationale) and is used by the affine constructs in MLIR. Unlike fixed-size integers, it cannot be used as an element of vector (rationale).
Rationale: integers of platform-specific bit widths are practical to express sizes, dimensionalities and subscripts.
Integer Type
Syntax:
// Sized integers like i1, i4, i8, i16, i32.
signed-integer-type ::= `si` [1-9][0-9]*
unsigned-integer-type ::= `ui` [1-9][0-9]*
signless-integer-type ::= `i` [1-9][0-9]*
integer-type ::= signed-integer-type |
unsigned-integer-type |
signless-integer-type
MLIR supports arbitrary precision integer types. Integer types have a designated width and may have signedness semantics.
Rationale: low precision integers (like i2, i4 etc) are useful for low-precision inference chips, and arbitrary precision integers are useful for hardware synthesis (where a 13 bit multiplier is a lot cheaper/smaller than a 16 bit one).
TODO: Need to decide on a representation for quantized integers (initial thoughts).
Memref Type
Syntax:
memref-type ::= ranked-memref-type | unranked-memref-type
ranked-memref-type ::= `memref` `<` dimension-list-ranked type
(`,` layout-specification)? (`,` memory-space)? `>`
unranked-memref-type ::= `memref` `<*x` type (`,` memory-space)? `>`
stride-list ::= `[` (dimension (`,` dimension)*)? `]`
strided-layout ::= `offset:` dimension `,` `strides: ` stride-list
semi-affine-map-composition ::= (semi-affine-map `,` )* semi-affine-map
layout-specification ::= semi-affine-map-composition | strided-layout
memory-space ::= integer-literal /* | TODO: address-space-id */
A memref type is a reference to a region of memory (similar to a buffer pointer, but more powerful). The buffer pointed to by a memref can be allocated, aliased and deallocated. A memref can be used to read and write data from/to the memory region which it references. Memref types use the same shape specifier as tensor types. Note that memref<f32>, memref<0 x f32>, memref<1 x 0 x f32>, and memref<0 x 1 x f32> are all different types.
A memref is allowed to have an unknown rank (e.g. memref<*xf32>). The purpose of unranked memrefs is to allow external library functions to receive memref arguments of any rank without versioning the functions based on the rank. Other uses of this type are disallowed or will have undefined behavior.
Codegen of Unranked Memref
Using unranked memref in codegen besides the case mentioned above is highly discouraged. Codegen is concerned with generating loop nests and specialized instructions for high-performance, unranked memref is concerned with hiding the rank and thus, the number of enclosing loops required to iterate over the data. However, if there is a need to code-gen unranked memref, one possible path is to cast into a static ranked type based on the dynamic rank. Another possible path is to emit a single while loop conditioned on a linear index and perform delinearization of the linear index to a dynamic array containing the (unranked) indices. While this is possible, it is expected to not be a good idea to perform this during codegen as the cost of the translations is expected to be prohibitive and optimizations at this level are not expected to be worthwhile. If expressiveness is the main concern, irrespective of performance, passing unranked memrefs to an external C++ library and implementing rank-agnostic logic there is expected to be significantly simpler.
Unranked memrefs may provide expressiveness gains in the future and help bridge the gap with unranked tensors. Unranked memrefs will not be expected to be exposed to codegen but one may query the rank of an unranked memref (a special op will be needed for this purpose) and perform a switch and cast to a ranked memref as a prerequisite to codegen.
Example:
// With static ranks, we need a function for each possible argument type %A = alloc() : memref<16x32xf32> %B = alloc() : memref<16x32x64xf32> call @helper_2D(%A) : (memref<16x32xf32>)->() call @helper_3D(%B) : (memref<16x32x64xf32>)->() // With unknown rank, the functions can be unified under one unranked type %A = alloc() : memref<16x32xf32> %B = alloc() : memref<16x32x64xf32> // Remove rank info %A_u = memref_cast %A : memref<16x32xf32> -> memref<*xf32> %B_u = memref_cast %B : memref<16x32x64xf32> -> memref<*xf32> // call same function with dynamic ranks call @helper(%A_u) : (memref<*xf32>)->() call @helper(%B_u) : (memref<*xf32>)->()
The core syntax and representation of a layout specification is a semi-affine map. Additionally, syntactic sugar is supported to make certain layout specifications more intuitive to read. For the moment, a memref supports parsing a strided form which is converted to a semi-affine map automatically.
The memory space of a memref is specified by a target-specific integer index. If no memory space is specified, then the default memory space (0) is used. The default space is target specific but always at index 0.
TODO: MLIR will eventually have target-dialects which allow symbolic use of memory hierarchy names (e.g. L3, L2, L1, ...) but we have not spec'd the details of that mechanism yet. Until then, this document pretends that it is valid to refer to these memories by bare-id.
The notionally dynamic value of a memref value includes the address of the buffer allocated, as well as the symbols referred to by the shape, layout map, and index maps.
Examples of memref static type
// Identity index/layout map #identity = affine_map<(d0, d1) -> (d0, d1)> // Column major layout. #col_major = affine_map<(d0, d1, d2) -> (d2, d1, d0)> // A 2-d tiled layout with tiles of size 128 x 256. #tiled_2d_128x256 = affine_map<(d0, d1) -> (d0 div 128, d1 div 256, d0 mod 128, d1 mod 256)> // A tiled data layout with non-constant tile sizes. #tiled_dynamic = affine_map<(d0, d1)[s0, s1] -> (d0 floordiv s0, d1 floordiv s1, d0 mod s0, d1 mod s1)> // A layout that yields a padding on two at either end of the minor dimension. #padded = affine_map<(d0, d1) -> (d0, (d1 + 2) floordiv 2, (d1 + 2) mod 2)> // The dimension list "16x32" defines the following 2D index space: // // { (i, j) : 0 <= i < 16, 0 <= j < 32 } // memref<16x32xf32, #identity> // The dimension list "16x4x?" defines the following 3D index space: // // { (i, j, k) : 0 <= i < 16, 0 <= j < 4, 0 <= k < N } // // where N is a symbol which represents the runtime value of the size of // the third dimension. // // %N here binds to the size of the third dimension. %A = alloc(%N) : memref<16x4x?xf32, #col_major> // A 2-d dynamic shaped memref that also has a dynamically sized tiled layout. // The memref index space is of size %M x %N, while %B1 and %B2 bind to the // symbols s0, s1 respectively of the layout map #tiled_dynamic. Data tiles of // size %B1 x %B2 in the logical space will be stored contiguously in memory. // The allocation size will be (%M ceildiv %B1) * %B1 * (%N ceildiv %B2) * %B2 // f32 elements. %T = alloc(%M, %N) [%B1, %B2] : memref<?x?xf32, #tiled_dynamic> // A memref that has a two-element padding at either end. The allocation size // will fit 16 * 64 float elements of data. %P = alloc() : memref<16x64xf32, #padded> // Affine map with symbol 's0' used as offset for the first dimension. #imapS = affine_map<(d0, d1) [s0] -> (d0 + s0, d1)> // Allocate memref and bind the following symbols: // '%n' is bound to the dynamic second dimension of the memref type. // '%o' is bound to the symbol 's0' in the affine map of the memref type. %n = ... %o = ... %A = alloc (%n)[%o] : <16x?xf32, #imapS>
Index Space
A memref dimension list defines an index space within which the memref can be indexed to access data.
Index
Data is accessed through a memref type using a multidimensional index into the multidimensional index space defined by the memref's dimension list.
Examples
// Allocates a memref with 2D index space: // { (i, j) : 0 <= i < 16, 0 <= j < 32 } %A = alloc() : memref<16x32xf32, #imapA> // Loads data from memref '%A' using a 2D index: (%i, %j) %v = load %A[%i, %j] : memref<16x32xf32, #imapA>
Index Map
An index map is a one-to-one semi-affine map that transforms a multidimensional index from one index space to another. For example, the following figure shows an index map which maps a 2-dimensional index from a 2x2 index space to a 3x3 index space, using symbols S0 and S1 as offsets.
The number of domain dimensions and range dimensions of an index map can be different, but must match the number of dimensions of the input and output index spaces on which the map operates. The index space is always non-negative and integral. In addition, an index map must specify the size of each of its range dimensions onto which it maps. Index map symbols must be listed in order with symbols for dynamic dimension sizes first, followed by other required symbols.
Layout Map
A layout map is a semi-affine map which encodes logical to physical index space mapping, by mapping input dimensions to their ordering from most-major (slowest varying) to most-minor (fastest varying). Therefore, an identity layout map corresponds to a row-major layout. Identity layout maps do not contribute to the MemRef type identification and are discarded on construction. That is, a type with an explicit identity map is memref<?x?xf32, (i,j)->(i,j)> is strictly the same as the one without layout maps, memref<?x?xf32>.
Layout map examples:
// MxN matrix stored in row major layout in memory: #layout_map_row_major = (i, j) -> (i, j) // MxN matrix stored in column major layout in memory: #layout_map_col_major = (i, j) -> (j, i) // MxN matrix stored in a 2-d blocked/tiled layout with 64x64 tiles. #layout_tiled = (i, j) -> (i floordiv 64, j floordiv 64, i mod 64, j mod 64)
Affine Map Composition
A memref specifies a semi-affine map composition as part of its type. A semi-affine map composition is a composition of semi-affine maps beginning with zero or more index maps, and ending with a layout map. The composition must be conformant: the number of dimensions of the range of one map, must match the number of dimensions of the domain of the next map in the composition.
The semi-affine map composition specified in the memref type, maps from accesses used to index the memref in load/store operations to other index spaces (i.e. logical to physical index mapping). Each of the semi-affine maps and thus its composition is required to be one-to-one.
The semi-affine map composition can be used in dependence analysis, memory access pattern analysis, and for performance optimizations like vectorization, copy elision and in-place updates. If an affine map composition is not specified for the memref, the identity affine map is assumed.
Strided MemRef
A memref may specify strides as part of its type. A stride specification is a list of integer values that are either static or ? (dynamic case). Strides encode the distance, in number of elements, in (linear) memory between successive entries along a particular dimension. A stride specification is syntactic sugar for an equivalent strided memref representation using semi-affine maps. For example, memref<42x16xf32, offset: 33, strides: [1, 64]> specifies a non-contiguous memory region of 42 by 16 f32 elements such that:
- the minimal size of the enclosing memory region must be
33 + 42 * 1 + 16 * 64 = 1066elements; - the address calculation for accessing element
(i, j)computes33 + i + 64 * j - the distance between two consecutive elements along the outer dimension is
1element and the distance between two consecutive elements along the outer dimension is64elements.
This corresponds to a column major view of the memory region and is internally represented as the type memref<42x16xf32, (i, j) -> (33 + i + 64 * j)>.
The specification of strides must not alias: given an n-D strided memref, indices (i1, ..., in) and (j1, ..., jn) may not refer to the same memory address unless i1 == j1, ..., in == jn.
Strided memrefs represent a view abstraction over preallocated data. They are constructed with special ops, yet to be introduced. Strided memrefs are a special subclass of memrefs with generic semi-affine map and correspond to a normalized memref descriptor when lowering to LLVM.
None Type
Syntax:
none-type ::= `none`
The none type is a unit type, i.e. a type with exactly one possible value, where its value does not have a defined dynamic representation.
Tensor Type
Syntax:
tensor-type ::= `tensor` `<` dimension-list type `>` dimension-list ::= dimension-list-ranked | (`*` `x`) dimension-list-ranked ::= (dimension `x`)* dimension ::= `?` | decimal-literal
Values with tensor type represents aggregate N-dimensional data values, and have a known element type. It may have an unknown rank (indicated by *) or may have a fixed rank with a list of dimensions. Each dimension may be a static non-negative decimal constant or be dynamically determined (indicated by ?).
The runtime representation of the MLIR tensor type is intentionally abstracted - you cannot control layout or get a pointer to the data. For low level buffer access, MLIR has a memref type. This abstracted runtime representation holds both the tensor data values as well as information about the (potentially dynamic) shape of the tensor. The dim operation returns the size of a dimension from a value of tensor type.
Note: hexadecimal integer literals are not allowed in tensor type declarations to avoid confusion between 0xf32 and 0 x f32. Zero sizes are allowed in tensors and treated as other sizes, e.g., tensor<0 x 1 x i32> and tensor<1 x 0 x i32> are different types. Since zero sizes are not allowed in some other types, such tensors should be optimized away before lowering tensors to vectors.
Examples:
// Tensor with unknown rank. tensor<* x f32> // Known rank but unknown dimensions. tensor<? x ? x ? x ? x f32> // Partially known dimensions. tensor<? x ? x 13 x ? x f32> // Full static shape. tensor<17 x 4 x 13 x 4 x f32> // Tensor with rank zero. Represents a scalar. tensor<f32> // Zero-element dimensions are allowed. tensor<0 x 42 x f32> // Zero-element tensor of f32 type (hexadecimal literals not allowed here). tensor<0xf32>
Tuple Type
Syntax:
tuple-type ::= `tuple` `<` (type ( `,` type)*)? `>`
The value of tuple type represents a fixed-size collection of elements, where each element may be of a different type.
Rationale: Though this type is first class in the type system, MLIR provides no standard operations for operating on tuple types (rationale).
Examples:
// Empty tuple. tuple<> // Single element tuple<f32> // Many elements. tuple<i32, f32, tensor<i1>, i5>
Vector Type
Syntax:
vector-type ::= `vector` `<` static-dimension-list vector-element-type `>` vector-element-type ::= float-type | integer-type static-dimension-list ::= (decimal-literal `x`)+
The vector type represents a SIMD style vector, used by target-specific operation sets like AVX. While the most common use is for 1D vectors (e.g. vector<16 x f32>) we also support multidimensional registers on targets that support them (like TPUs).
Vector shapes must be positive decimal integers.
Note: hexadecimal integer literals are not allowed in vector type declarations, vector<0x42xi32> is invalid because it is interpreted as a 2D vector with shape (0, 42) and zero shapes are not allowed.
Attributes
Syntax:
attribute-dict ::= `{` `}`
| `{` attribute-entry (`,` attribute-entry)* `}`
attribute-entry ::= dialect-attribute-entry | dependent-attribute-entry
dialect-attribute-entry ::= dialect-namespace `.` bare-id `=` attribute-value
dependent-attribute-entry ::= dependent-attribute-name `=` attribute-value
dependent-attribute-name ::= ((letter|[_]) (letter|digit|[_$])*)
| string-literal
Attributes are the mechanism for specifying constant data on operations in places where a variable is never allowed - e.g. the index of a dim operation, or the stride of a convolution. They consist of a name and a concrete attribute value. The set of expected attributes, their structure, and their interpretation are all contextually dependent on what they are attached to.
There are two main classes of attributes: dependent and dialect. Dependent attributes derive their structure and meaning from what they are attached to; e.g., the meaning of the index attribute on a dim operation is defined by the dim operation. Dialect attributes, on the other hand, derive their context and meaning from a specific dialect. An example of a dialect attribute may be a swift.self function argument attribute that indicates an argument is the self/context parameter. The context of this attribute is defined by the swift dialect and not the function argument.
Attribute values are represented by the following forms:
attribute-value ::= attribute-alias | dialect-attribute | builtin-attribute
Attribute Value Aliases
attribute-alias ::= '#' alias-name '=' attribute-value attribute-alias ::= '#' alias-name
MLIR supports defining named aliases for attribute values. An attribute alias is an identifier that can be used in the place of the attribute that it defines. These aliases must be defined before their uses. Alias names may not contain a ‘.’, since those names are reserved for dialect attributes.
Example:
#map = affine_map<(d0) -> (d0 + 10)> // Using the original attribute. %b = affine.apply affine_map<(d0) -> (d0 + 10)> (%a) // Using the attribute alias. %b = affine.apply #map(%a)
Dialect Attribute Values
Similarly to operations, dialects may define custom attribute values. The syntactic structure of these values is identical to custom dialect type values, except that dialect attribute values are distinguished with a leading ‘#’, while dialect types are distinguished with a leading ‘!’.
dialect-attribute-value ::= '#' opaque-dialect-item dialect-attribute-value ::= '#' pretty-dialect-item
Dialect attribute values can be specified in a verbose form, e.g. like this:
// Complex attribute value. #foo<"something<abcd>"> // Even more complex attribute value. #foo<"something<a%%123^^^>>>">
Dialect attribute values that are simple enough can use the pretty format, which is a lighter weight syntax that is equivalent to the above forms:
// Complex attribute #foo.something<abcd>
Sufficiently complex dialect attribute values are required to use the verbose form for generality. For example, the more complex type shown above would not be valid in the lighter syntax: #foo.something<a%%123^^^>>> because it contains characters that are not allowed in the lighter syntax, as well as unbalanced <> characters.
See here on how to define dialect attribute values.
Builtin Attribute Values
Builtin attributes are a core set of dialect attributes that are defined in a builtin dialect and thus available to all users of MLIR.
builtin-attribute ::= affine-map-attribute
| array-attribute
| bool-attribute
| dictionary-attribute
| elements-attribute
| float-attribute
| integer-attribute
| integer-set-attribute
| string-attribute
| symbol-ref-attribute
| type-attribute
| unit-attribute
AffineMap Attribute
Syntax:
affine-map-attribute ::= `affine_map` `<` affine-map `>`
An affine-map attribute is an attribute that represents an affine-map object.
Array Attribute
Syntax:
array-attribute ::= `[` (attribute-value (`,` attribute-value)*)? `]`
An array attribute is an attribute that represents a collection of attribute values.
Boolean Attribute
Syntax:
bool-attribute ::= bool-literal
A boolean attribute is a literal attribute that represents a one-bit boolean value, true or false.
Dictionary Attribute
Syntax:
dictionary-attribute ::= `{` (attribute-entry (`,` attribute-entry)*)? `}`
A dictionary attribute is an attribute that represents a sorted collection of named attribute values. The elements are sorted by name, and each name must be unique within the collection.
Elements Attributes
Syntax:
elements-attribute ::= dense-elements-attribute
| opaque-elements-attribute
| sparse-elements-attribute
An elements attribute is a literal attribute that represents a constant vector or tensor value.
Dense Elements Attribute
Syntax:
dense-elements-attribute ::= `dense` `<` attribute-value `>` `:`
( tensor-type | vector-type )
A dense elements attribute is an elements attribute where the storage for the constant vector or tensor value has been densely packed. The attribute supports storing integer or floating point elements, with integer/index/floating element types. It also support storing string elements with a custom dialect string element type.
Opaque Elements Attribute
Syntax:
opaque-elements-attribute ::= `opaque` `<` dialect-namespace `,`
hex-string-literal `>` `:`
( tensor-type | vector-type )
An opaque elements attribute is an elements attribute where the content of the value is opaque. The representation of the constant stored by this elements attribute is only understood, and thus decodable, by the dialect that created it.
Note: The parsed string literal must be in hexadecimal form.
Sparse Elements Attribute
Syntax:
sparse-elements-attribute ::= `sparse` `<` attribute-value `,` attribute-value
`>` `:` ( tensor-type | vector-type )
A sparse elements attribute is an elements attribute that represents a sparse vector or tensor object. This is where very few of the elements are non-zero.
The attribute uses COO (coordinate list) encoding to represent the sparse elements of the elements attribute. The indices are stored via a 2-D tensor of 64-bit integer elements with shape [N, ndims], which specifies the indices of the elements in the sparse tensor that contains non-zero values. The element values are stored via a 1-D tensor with shape [N], that supplies the corresponding values for the indices.
Example:
sparse<[[0, 0], [1, 2]], [1, 5]> : tensor<3x4xi32> // This represents the following tensor: /// [[1, 0, 0, 0], /// [0, 0, 5, 0], /// [0, 0, 0, 0]]
Float Attribute
Syntax:
float-attribute ::= (float-literal (`:` float-type)?)
| (hexadecimal-literal `:` float-type)
A float attribute is a literal attribute that represents a floating point value of the specified float type. It can be represented in the hexadecimal form where the hexadecimal value is interpreted as bits of the underlying binary representation. This form is useful for representing infinity and NaN floating point values. To avoid confusion with integer attributes, hexadecimal literals must be followed by a float type to define a float attribute.
Examples:
42.0 // float attribute defaults to f64 type 42.0 : f32 // float attribute of f32 type 0x7C00 : f16 // positive infinity 0x7CFF : f16 // NaN (one of possible values) 42 : f32 // Error: expected integer type
Integer Attribute
Syntax:
integer-attribute ::= integer-literal ( `:` (index-type | integer-type) )?
An integer attribute is a literal attribute that represents an integral value of the specified integer or index type. The default type for this attribute, if one is not specified, is a 64-bit integer.
Integer Set Attribute
Syntax:
integer-set-attribute ::= `affine_set` `<` integer-set `>`
An integer-set attribute is an attribute that represents an integer-set object.
String Attribute
Syntax:
string-attribute ::= string-literal (`:` type)?
A string attribute is an attribute that represents a string literal value.
Symbol Reference Attribute
Syntax:
symbol-ref-attribute ::= symbol-ref-id (`::` symbol-ref-id)*
A symbol reference attribute is a literal attribute that represents a named reference to an operation that is nested within an operation with the OpTrait::SymbolTable trait. As such, this reference is given meaning by the nearest parent operation containing the OpTrait::SymbolTable trait. It may optionally contain a set of nested references that further resolve to a symbol nested within a different symbol table.
This attribute can only be held internally by array attributes and dictionary attributes(including the top-level operation attribute dictionary), i.e. no other attribute kinds such as Locations or extended attribute kinds.
Rationale: Identifying accesses to global data is critical to enabling efficient multi-threaded compilation. Restricting global data access to occur through symbols and limiting the places that can legally hold a symbol reference simplifies reasoning about these data accesses.
See Symbols And SymbolTables for more information.
Type Attribute
Syntax:
type-attribute ::= type
A type attribute is an attribute that represents a type object.
Unit Attribute
unit-attribute ::= `unit`
A unit attribute is an attribute that represents a value of unit type. The unit type allows only one value forming a singleton set. This attribute value is used to represent attributes that only have meaning from their existence.
One example of such an attribute could be the swift.self attribute. This attribute indicates that a function parameter is the self/context parameter. It could be represented as a boolean attribute(true or false), but a value of false doesn‘t really bring any value. The parameter either is the self/context or it isn’t.
// A unit attribute defined with the `unit` value specifier. func @verbose_form(i1) attributes {dialectName.unitAttr = unit} // A unit attribute can also be defined without the value specifier. func @simple_form(i1) attributes {dialectName.unitAttr}