CINXE.COM

<!doctype html><html lang=en-us><head><meta charset=utf-8><meta http-equiv=x-ua-compatible content="IE=edge"><meta name=viewport content="width=device-width,initial-scale=1,maximum-scale=1,user-scalable=no"><title>PDLL - PDL Language - MLIR</title><meta name=description content="Multi-Level IR Compiler Framework"><meta name=generator content="Hugo 0.119.0"><link href=https://mlir.llvm.org/index.xml rel=alternate type=application/rss+xml><link rel=canonical href=https://mlir.llvm.org/docs/PDLL/><link rel=stylesheet href=https://mlir.llvm.org/css/theme.css><script src=https://use.fontawesome.com/releases/v5.0.6/js/all.js></script> <link rel=stylesheet href=https://mlir.llvm.org/css/chroma.min.css><script src=https://cdn.jsdelivr.net/npm/jquery@3.3.1/dist/jquery.min.js></script> <script src=https://cdn.jsdelivr.net/npm/jquery.easing@1.4.1/jquery.easing.min.js></script> <script src=https://mlir.llvm.org/js/bundle.js></script> <script type=text/javascript src="https://cdnjs.cloudflare.com/ajax/libs/mathjax/2.7.1/MathJax.js?config=TeX-AMS-MML_HTMLorMML"></script> <script type=text/x-mathjax-config> MathJax.Hub.Config({ tex2jax: { inlineMath: [['$', '$'] ], displayMath: [ ['$$','$$'], ["\\[","\\]"] ] } }); </script><link rel=apple-touch-icon sizes=180x180 href="/apple-touch-icon.png?v=1"><link rel=icon type=image/png sizes=32x32 href="/favicon-32x32.png?v=1"><link rel=icon type=image/png sizes=16x16 href="/favicon-16x16.png?v=1"><link rel=manifest href="/site.webmanifest?v=1"><link rel=mask-icon href="/safari-pinned-tab.svg?v=1" color=#3775e0><link rel="shortcut icon" href="/favicon.ico?v=1"><meta name=msapplication-TileColor content="#2d89ef"><meta name=theme-color content="#ffffff"><link rel=icon href=/favicon.svg type=image/svg+xml sizes=any><style>:root{}</style></head><body><div class=container><header><h1><div><img src=https://mlir.llvm.org//mlir-logo.png width=40px align=absmiddle> MLIR</div></h1>Multi-Level IR Compiler Framework</header><div class=global-menu><nav><ul><li class=parent><a href>Community</a><ul class=sub-menu><li class=child><a href=https://llvm.discourse.group/c/mlir/31>Forums</a></li><li class=child><a href=https://discord.gg/xS7Z362>Chat</a></li></ul></li><li><a href=/getting_started/Debugging/>Debugging Tips</a></li><li><a href=/getting_started/Faq/>FAQ</a></li><li class=parent><a href=https://github.com/llvm/llvm-project/tree/main/mlir>Source</a><ul class=sub-menu><li class=child><a href=/doxygen/>Doxygen</a></li><li class=child><a href=https://github.com/llvm/llvm-project/tree/main/mlir>GitHub</a></li></ul></li><li><a href="https://bugs.llvm.org/buglist.cgi?bug_status=__open__&list_id=177877&order=changeddate%20DESC%2Cpriority%2Cbug_severity&product=MLIR&query_format=specific">Bugs</a></li><li><a href=https://github.com/llvm/mlir-www/tree/main/website/static/LogoAssets>Logo Assets</a></li><li><a href=https://www.youtube.com/MLIRCompiler>Youtube Channel</a></li></ul></nav></div><div class=content-container><main><h1>PDLL - PDL Language</h1>This document details the PDL Language (PDLL), a custom frontend language for writing pattern rewrites targeting MLIR.Note: This document assumes a familiarity with MLIR concepts; more specifically the concepts detailed within the <a href=/docs/PatternRewriter/>MLIR Pattern Rewriting</a> and <a href=/docs/DefiningDialects/Operations/>Operation Definition Specification (ODS)</a> documentation.<nav id=TableOfContents><ul><li><a href=#introduction>Introduction</a></li><li><a href=#rationale>Rationale</a><ul><li><a href=#why-build-a-new-language-instead-of-improving-tablegen-drr>Why build a new language instead of improving TableGen DRR?</a></li><li><a href=#why-not-build-a-dsl-in-x>Why not build a DSL in “X”?</a></li></ul></li><li><a href=#language-specification>Language Specification</a><ul><li><a href=#includes>Includes</a></li><li><a href=#patterns>Patterns</a></li><li><a href=#variables>Variables</a></li><li><a href=#operation-expression>Operation Expression</a></li><li><a href=#attribute-expression>Attribute Expression</a></li><li><a href=#type-expression>Type Expression</a></li><li><a href=#tuples>Tuples</a></li><li><a href=#constraints-1>Constraints</a></li><li><a href=#rewriters>Rewriters</a></li></ul></li></ul></nav><h2 id=introduction>Introduction <a class=headline-hash href=#introduction>¶</a></h2>Pattern matching is an extremely important component within MLIR, as it encompasses many different facets of the compiler. From canonicalization, to optimization, to conversion; every MLIR based compiler will heavily rely on the pattern matching infrastructure in some capacity.The PDL Language (PDLL) provides a declarative pattern language designed from the ground up for representing MLIR pattern rewrites. PDLL is designed to natively support writing matchers on all of MLIRs constructs via an intuitive interface that may be used for both ahead-of-time (AOT) and just-in-time (JIT) pattern compilation.<h2 id=rationale>Rationale <a class=headline-hash href=#rationale>¶</a></h2>This section provides details on various design decisions, their rationale, and alternatives considered when designing PDLL. Given the nature of software development, this section may include references to areas of the MLIR compiler that no longer exist.<h3 id=why-build-a-new-language-instead-of-improving-tablegen-drr>Why build a new language instead of improving TableGen DRR? <a class=headline-hash href=#why-build-a-new-language-instead-of-improving-tablegen-drr>¶</a></h3>Note: This section assumes familiarity with <a href=/docs/DeclarativeRewrites/>TDRR</a>, please refer the relevant documentation before continuing.Tablegen DRR (TDRR), i.e. <a href=/docs/DeclarativeRewrites/>Table-driven Declarative Rewrite Rules</a>, is a declarative DSL for defining MLIR pattern rewrites within the <a href=https://llvm.org/docs/TableGen/index.html>TableGen</a> language. This infrastructure is currently the main way in which patterns may be defined declaratively within MLIR. TDRR utilizes TableGen’s <code>dag</code> support to enable defining MLIR patterns that fit nicely within a DAG structure; in a similar way in which tablegen has been used to defined patterns for LLVM’s backend infrastructure (SelectionDAG/Global Isel/etc.). Unfortunately however, the TableGen language is not as amenable to the structure of MLIR patterns as it has been for LLVM.The issues with TDRR largely stem from the use of TableGen as the host language for the DSL. These issues have risen from a mismatch in the structure of TableGen compared to the structure of MLIR, and from TableGen having different motivational goals than MLIR. A majority (or all depending on how stubborn you are) of the issues that we’ve come across with TDRR have been addressable in some form; the sticking point here is that the solutions to these problems have often been more “creative” than we’d like. This is a problem, and why we decided not to invest a larger effort into improving TDRR; users generally don’t want “creative” APIs, they want something that is intuitive to read/write.To highlight some of these issues, below we will take a tour through some of the problems that have arisen, and how we “fixed” them.<h4 id=multi-result-operations>Multi-result operations <a class=headline-hash href=#multi-result-operations>¶</a></h4>MLIR natively supports a variable number of operation results. For the DAG based structure of TDRR, any form of multiple results (operations in this instance) creates a problem. This is because the DAG wants a single root node, and does not have nice facilities for indexing or naming the multiple results. Let’s take a look at a quick example to see how this manifests:<div class=highlight><pre tabindex=0 class=chroma><code class=language-tablegen data-lang=tablegen>// Suppose we have a three result operation, defined as seen below. def ThreeResultOp : Op<"three_result_op"> { let arguments = (ins ...); let results = (outs AnyTensor:$output1, AnyTensor:$output2, AnyTensor:$output3 ); } // To bind the results of `ThreeResultOp` in a TDRR pattern, we bind all results // to a single name and use a special naming convention: `__N`, where `N` is the // N-th result. def : Pattern<(ThreeResultOp:$results ...), [(... $results__0), ..., (... $results__2), ...]>; </code></pre></div>In TDRR, we “solved” the problem of accessing multiple results, but this isn’t a very intuitive interface for users. Magical naming conventions obfuscate the code and can easily introduce bugs and other errors. There are various things that we could try to improve this situation, but there is a fundamental limit to what we can do given the limits of the TableGen dag structure. In PDLL, however, we have the freedom and flexibility to provide a proper interface into operations, regardless of their structure:<pre tabindex=0><code class=language-pdll data-lang=pdll>// Import our definition of `ThreeResultOp`. #include "ops.td" Pattern { ... // In PDLL, we can directly reference the results of an operation variable. // This provides a closer mental model to what the user expects. let threeResultOp = op<my_dialect.three_result_op>; let userOp = op<my_dialect.user_op>(threeResultOp.output1, ..., threeResultOp.output3); ... } </code></pre><h4 id=constraints>Constraints <a class=headline-hash href=#constraints>¶</a></h4>In TDRR, the match dag defines the general structure of the input IR to match. Any non-structural/non-type constraints on the input are generally relegated to a list of constraints specified after the rewrite dag. For very simple patterns this may suffice, but with larger patterns it becomes quite problematic as it separates the constraint from the entity it constrains and negatively impacts the readability of the pattern. As an example, let’s look at a simple pattern that adds additional constraints to its inputs:<div class=highlight><pre tabindex=0 class=chroma><code class=language-tablegen data-lang=tablegen>// Suppose we have a two result operation, defined as seen below. def TwoResultOp : Op<"two_result_op"> { let arguments = (ins ...); let results = (outs AnyTensor:$output1, AnyTensor:$output2 ); } // A simple constraint to check if a value is use_empty. def HasNoUseOf: Constraint<CPred<"$_self.use_empty()">, "has no use">; // Check if two values have a ShapedType with the same element type. def HasSameElementType : Constraint< CPred<"$0.getType().cast<ShapedType>().getElementType() == " "$1.getType().cast<ShapedType>().getElementType()">, "values have same element type">; def : Pattern<(TwoResultOp:$results $input), [(...), (...)], [(HasNoUseOf:$results__1), (HasSameElementType $results__0, $input)]>; </code></pre></div>Above, when observing the constraints we need to search through the input dag for the inputs (also keeping in mind the magic naming convention for multiple results). For this simple pattern it may be just a few lines above, but complex patterns often grow to 10s of lines long. In PDLL, these constraints can be applied directly on or next to the entities they apply to:<pre tabindex=0><code class=language-pdll data-lang=pdll>// The same constraints that we defined above: Constraint HasNoUseOf(value: Value) [{ return success(value.use_empty()); }]; Constraint HasSameElementType(value1: Value, value2: Value) [{ return success(value1.getType().cast<ShapedType>().getElementType() == value2.getType().cast<ShapedType>().getElementType()); }]; Pattern { // In PDLL, we can apply the constraint as early (or as late) as we want. This // enables better structuring of the matcher code, and improves the // readability/maintainability of the pattern. let op = op<my_dialect.two_result_op>(input: Value); HasNoUseOf(op.output2); HasSameElementType(input, op.output2); // ... } </code></pre><h4 id=replacing-multiple-operations>Replacing Multiple Operations <a class=headline-hash href=#replacing-multiple-operations>¶</a></h4>Often times a pattern will transform N number of input operations into N number of result operations. In PDLL, replacing multiple operations is as simple as adding two <a href=#replace-statement><code>replace</code> statements</a>. In TDRR, the situation is a bit more nuanced. Given the single root structure of the TableGen dag, replacing a non-root operation is not nicely supported. It currently isn’t natively possible, and instead requires using multiple patterns. We could potentially add another special rewrite directive, or extend <code>replaceWithValue</code>, but this simply highlights how even a basic IR transformation is muddled by the complexity of the host language.<h3 id=why-not-build-a-dsl-in-x>Why not build a DSL in “X”? <a class=headline-hash href=#why-not-build-a-dsl-in-x>¶</a></h3>Yes! Well yes and no. To understand why, we have to consider what types of users we are trying to serve and what constraints we enforce upon them. The goal of PDLL is to provide a default and effective pattern language for MLIR that all users of MLIR can interact with immediately, regardless of their host environment. This language is available with no extra dependencies and comes “free” along with MLIR. If we were to use an existing host language to build our new DSL, we would need to make compromises along with it depending on the language. For some, there are questions of how to enforce matching environments (python2 or python3?, which version?), performance considerations, integration, etc. As an LLVM project, this could also mean enforcing a new language dependency on the users of MLIR (many of which may not want/need such a dependency otherwise). Another issue that comes along with any DSL that is embeded in another language: mitigating the user impedance mismatch between what the user expects from the host language and what our “backend” supports. For example, the PDL IR abstraction only contains limited support for control flow. If we were to build a DSL in python, we would need to ensure that complex control flow is either handled completely or effectively errors out. Even with ideal error handling, not having the expected features available creates user frustration. In addition to the environment constraints, there is also the issue of language tooling. With PDLL we intend to build a very robust and modern toolset that is designed to cater the needs of pattern developers, including code completion, signature help, and many more features that are specific to the problem we are solving. Integrating custom language tooling into existing languages can be difficult, and in some cases impossible (as our DSL would merely be a small subset of the existing language).These various points have led us to the initial conclusion that the most effective tool we can provide for our users is a custom tool designed for the problem at hand. With all of that being said, we understand that not all users have the same constraints that we have placed upon ourselves. We absolutely encourage and support the existence of various PDL frontends defined in different languages. This is one of the original motivating factors around building the PDL IR abstraction in the first place; to enable innovation and flexibility for our users (and in turn their users). For some, such as those in research and the Machine Learning space, they may already have a certain language (such as Python) heavily integrated into their workflow. For these users, a PDL DSL in their language may be ideal and we will remain committed to supporting and endorsing that from an infrastructure point-of-view.<h2 id=language-specification>Language Specification <a class=headline-hash href=#language-specification>¶</a></h2>Note: PDLL is still under active development, and the designs discussed below are not necessarily final and may be subject to change.The design of PDLL is heavily influenced and centered around the <a href=https://mlir.llvm.org/docs/Dialects/PDLOps/>PDL IR abstraction</a>, which in turn is designed as an abstract model of the core MLIR structures. This leads to a design and structure that feels very similar to if you were directly writing the IR you want to match.<h3 id=includes>Includes <a class=headline-hash href=#includes>¶</a></h3>PDLL supports an <code>include</code> directive to import content defined within other source files. There are two types of files that may be included: <code>.pdll</code> and <code>.td</code> files.<h4 id=pdll-includes><code>.pdll</code> includes <a class=headline-hash href=#pdll-includes>¶</a></h4>When including a <code>.pdll</code> file, the contents of that file are copied directly into the current file being processed. This means that any patterns, constraints, rewrites, etc., defined within that file are processed along with those within the current file.<h4 id=td-includes><code>.td</code> includes <a class=headline-hash href=#td-includes>¶</a></h4>When including a <code>.td</code> file, PDLL will automatically import any pertinent <a href=/docs/DefiningDialects/Operations/>ODS</a> information within that file. This includes any defined operations, constraints, interfaces, and more, making them implicitly accessible within PDLL. This is important, as ODS information allows for certain PDLL constructs, such as the <a href=#operation><code>operation</code> expression</a>, to become much more powerful.<h3 id=patterns>Patterns <a class=headline-hash href=#patterns>¶</a></h3>In any pattern descriptor language, pattern definition is at the core. In PDLL, patterns start with <code>Pattern</code> optionally followed by a name and a set of pattern metadata, and finally terminated by a pattern body. A few simple examples are shown below:<pre tabindex=0><code class=language-pdll data-lang=pdll>// Here we have defined an anonymous pattern: Pattern { // Pattern bodies are separated into two components: // * Match Section // - Describes the input IR. let root = op<toy.reshape>(op<toy.reshape>(arg: Value)); // * Rewrite Section // - Describes how to transform the IR. // - Last statement starts the rewrite. replace root with op<toy.reshape>(arg); } // Here we have defined a pattern named `ReshapeReshapeOptPattern` with a // benefit of 10: Pattern ReshapeReshapeOptPattern with benefit(10) { replace op<toy.reshape>(op<toy.reshape>(arg: Value)) with op<toy.reshape>(arg); } </code></pre>After the definition of the pattern metadata, we specify the pattern body. The structure of a pattern body is comprised of two main sections, the <code>match</code> section and the <code>rewrite</code> section. The <code>match</code> section of a pattern describes the expected input IR, whereas the <code>rewrite</code> section describes how to transform that IR. This distinction is an important one to make, as PDLL handles certain variables and expressions differently within the different sections. When relevant in each of the sections below, we shall explicitly call out any behavioral differences.The general layout of the <code>match</code> and <code>rewrite</code> section is as follows: the last statement of the pattern body is required to be a <a href=#operation-rewrite-statements><code>operation rewrite statement</code></a>, and denotes the <code>rewrite</code> section; every statement before denotes the <code>match</code> section.<h4 id=pattern-metadata>Pattern metadata <a class=headline-hash href=#pattern-metadata>¶</a></h4>Rewrite patterns in MLIR have a set of metadata that allow for controlling certain behaviors, and providing information to the rewrite driver applying the pattern. In PDLL, a pattern can provide a non-default value for this metadata after the pattern name. Below, examples are shown for the different types of metadata supported:<h5 id=benefit>Benefit <a class=headline-hash href=#benefit>¶</a></h5>The benefit of a Pattern is an integer value that represents the “benefit” of matching that pattern. It is used by pattern drivers to determine the relative priorities of patterns during application; a pattern with a higher benefit is generally applied before one with a lower benefit.In PDLL, a pattern has a default benefit set to the number of input operations, i.e. the number of distinct <code>Op</code> expressions/variables, in the match section. This rule is driven by an observation that larger matches are more beneficial than smaller ones, and if a smaller one is applied first the larger one may not apply anymore. Patterns can override this behavior by specifying the benefit in the metadata section of the pattern:<pre tabindex=0><code class=language-pdll data-lang=pdll>// Here we specify that this pattern has a benefit of `10`, overriding the // default behavior. Pattern with benefit(10) { ... } </code></pre><h5 id=bounded-rewrite-recursion>Bounded Rewrite Recursion <a class=headline-hash href=#bounded-rewrite-recursion>¶</a></h5>During pattern application, there are situations in which a pattern may be applicable to the result of a previous application of that same pattern. If the pattern does not properly handle this recusive application, the pattern driver could become stuck in an infinite loop of application. To prevent this, patterns by-default are assumed to not have proper recursive bounding and will not be recursively applied. A pattern can signal that it does have proper handling for recursion by specifying the <code>recusion</code> flag in the pattern metadata section:<pre tabindex=0><code class=language-pdll data-lang=pdll>// Here we signal that this pattern properly bounds recursive application. Pattern with recusion { ... } </code></pre><h4 id=single-line-lambda-body>Single Line “Lambda” Body <a class=headline-hash href=#single-line-lambda-body>¶</a></h4>Patterns generally define their body using a compound block of statements, as shown below:<pre tabindex=0><code class=language-pdll data-lang=pdll>Pattern { replace op<my_dialect.foo>(operands: ValueRange) with operands; } </code></pre>Patterns also support a lambda-like syntax for specifying simple single line bodies. The lambda body of a Pattern expects a single <a href=#operation-rewrite-statements>operation rewrite statement</a>:<pre tabindex=0><code class=language-pdll data-lang=pdll>Pattern => replace op<my_dialect.foo>(operands: ValueRange) with operands; </code></pre><h3 id=variables>Variables <a class=headline-hash href=#variables>¶</a></h3>Variables in PDLL represent specific instances of IR entities, such as <code>Value</code>s, <code>Operation</code>s, <code>Type</code>s, etc. Consider the simple pattern below:<pre tabindex=0><code class=language-pdll data-lang=pdll>Pattern { let value: Value; let root = op<mydialect.foo>(value); replace root with value; } </code></pre>In this pattern we define two variables, <code>value</code> and <code>root</code>, using the <code>let</code> statement. The <code>let</code> statement allows for defining variables and constraining them. Every variable in PDLL is of a certain type, which defines the type of IR entity the variable represents. The type of a variable may be determined via either a constraint, or an initializer expression.<h4 id=variable-binding>Variable “Binding” <a class=headline-hash href=#variable-binding>¶</a></h4>In addition to having a type, variables must also be “bound”, either via an initializer expression or to a non-native constraint or rewrite use within the <code>match</code> section of the pattern. “Binding” a variable contextually identifies that variable within either the input (i.e. <code>match</code> section) or output (i.e. <code>rewrite</code> section) IR. In the <code>match</code> section, this allows for building the match tree from the pattern’s root operation, which must be “bound” to the <a href=#operation-rewrite-statements>operation rewrite statement</a> that denotes the <code>rewrite</code> section of the pattern. All non-root variables within the <code>match</code> section must be bound in some way to the “root” operation. To help illustrate the concept, let’s take a look at a quick example. Consider the <code>.mlir</code> snippet below:<div class=highlight><pre tabindex=0 class=chroma><code class=language-mlir data-lang=mlir>func.func @baz(%arg: i32) { %result = my_dialect.foo %arg, %arg -> i32 } </code></pre></div>Say that we want to write a pattern that matches <code>my_dialect.foo</code> and replaces it with its unique input argument. A naive way to write this pattern in PDLL is shown below:<pre tabindex=0><code class=language-pdll data-lang=pdll>Pattern { // ** match section ** // let arg: Value; let root = op<my_dialect.foo>(arg, arg); // ** rewrite section ** // replace root with arg; } </code></pre>In the above pattern, the <code>arg</code> variable is “bound” to the first and second operands of the <code>root</code> operation. Every use of <code>arg</code> is constrained to be the same <code>Value</code>, i.e. the first and second operands of <code>root</code> will be constrained to refer to the same input Value. The same is true for the <code>root</code> operation, it is bound to the “root” operation of the pattern as it is used in input of the top-level <a href=#replace-statement><code>replace</code> statement</a> of the <code>rewrite</code> section of the pattern. Writing this pattern using the C++ API, the concept of “binding” becomes more clear:<div class=highlight><pre tabindex=0 class=chroma><code class=language-c++ data-lang=c++>struct Pattern : public OpRewritePattern<my_dialect::FooOp> { LogicalResult matchAndRewrite(my_dialect::FooOp root, PatternRewriter &rewriter) { Value arg = root->getOperand(0); if (arg != root->getOperand(1)) return failure(); rewriter.replaceOp(root, arg); return success(); } }; </code></pre></div>If a variable is not “bound” properly, PDLL won’t be able to identify what value it would correspond to in the IR. As a final example, let’s consider a variable that hasn’t been bound:<pre tabindex=0><code class=language-pdll data-lang=pdll>Pattern { // ** match section ** // let arg: Value; let root = op<my_dialect.foo> // ** rewrite section ** // replace root with arg; } </code></pre>If we were to write this exact pattern in C++, we would end up with:<div class=highlight><pre tabindex=0 class=chroma><code class=language-c++ data-lang=c++>struct Pattern : public OpRewritePattern<my_dialect::FooOp> { LogicalResult matchAndRewrite(my_dialect::FooOp root, PatternRewriter &rewriter) { // `arg` was never bound, so we don't know what input Value it was meant to // correspond to. Value arg; rewriter.replaceOp(root, arg); return success(); } }; </code></pre></div><h4 id=variable-constraints>Variable Constraints <a class=headline-hash href=#variable-constraints>¶</a></h4><pre tabindex=0><code class=language-pdll data-lang=pdll>// This statement defines a variable `value` that is constrained to be a `Value`. let value: Value; // This statement defines a variable `value` that is constrained to be a `Value` // *and* constrained to have a single use. let value: [Value, HasOneUse]; </code></pre>Any number of single entity constraints may be attached directly to a variable upon declaration. Within the <code>matcher</code> section, these constraints may add additional checks on the input IR. Within the <code>rewriter</code> section, constraints are only used to define the type of the variable. There are a number of builtin constraints that correlate to the core MLIR constructs: <code>Attr</code>, <code>Op</code>, <code>Type</code>, <code>TypeRange</code>, <code>Value</code>, <code>ValueRange</code>. Along with these, users may define custom constraints that are implemented within PDLL, or natively (i.e. outside of PDLL). See the general <a href=#constraints>Constraints</a> section for more detailed information.<h4 id=inline-variable-definition>Inline Variable Definition <a class=headline-hash href=#inline-variable-definition>¶</a></h4>Along with the <code>let</code> statement, variables may also be defined inline by specifying the constraint list along with the desired variable name in the first place that the variable would be used. After definition, the variable is visible from all points forward. See below for an example:<pre tabindex=0><code class=language-pdll data-lang=pdll>// `value` is used as an operand to the operation `root`: let value: Value; let root = op<my_dialect.foo>(value); replace root with value; // `value` could also be defined "inline": let root = op<my_dialect.foo>(value: Value); replace root with value; </code></pre>Note that the point of definition of an inline variable is the point of reference, meaning that an inline variable can be used immediately in the same parent expression within which it was defined:<pre tabindex=0><code class=language-pdll data-lang=pdll>let root = op<my_dialect.foo>(value: Value, _: Value, value); replace root with value; </code></pre><h5 id=wildcard-variable-definition>Wildcard Variable Definition <a class=headline-hash href=#wildcard-variable-definition>¶</a></h5>Often times when defining a variable inline, the variable isn’t intended to be used anywhere else in the pattern. For example, this may happen if you want to attach constraints to a variable but have no other use for it. In these situations, the “wildcard” variable can be used to remove the need to provide a name, as “wildcard” variables are not visible outside of the point of definition. An example is shown below:<pre tabindex=0><code class=language-pdll data-lang=pdll>Pattern { let root = op<my_dialect.foo>(arg: Value, _: Value, _: [Value, I64Value], arg); replace root with arg; } </code></pre>In the above example, the second operand isn’t needed for the pattern but we need to provide it to signal that a second operand does exist (we just don’t care what it is in this pattern).<h3 id=operation-expression>Operation Expression <a class=headline-hash href=#operation-expression>¶</a></h3>An operation expression in PDLL represents an MLIR operation. In the <code>match</code> section of the pattern, this expression models one of the input operations to the pattern. In the <code>rewrite</code> section of the pattern, this expression models one of the operations to create. The general structure of the operation expression is very similar to that of the “generic form” of textual MLIR assembly:<pre tabindex=0><code class=language-pdll data-lang=pdll>let root = op<my_dialect.foo>(operands: ValueRange) {attr = attr: Attr} -> (resultTypes: TypeRange); </code></pre>Let’s walk through each of the different components of the expression:<h4 id=operation-name>Operation name <a class=headline-hash href=#operation-name>¶</a></h4>The operation name signifies which type of MLIR Op this operation corresponds to. In the <code>match</code> section of the pattern, the name may be elided. This would cause this pattern to match any operation type that satifies the rest of the constraints of the operation. In the <code>rewrite</code> section, the name is required.<pre tabindex=0><code class=language-pdll data-lang=pdll>// `root` corresponds to an instance of a `my_dialect.foo` operation. let root = op<my_dialect.foo>; // `root` could be an instance of any operation type. let root = op<>; </code></pre><h4 id=operands>Operands <a class=headline-hash href=#operands>¶</a></h4>The operands section corresponds to the operands of the operation. This section of an operation expression may be elided, which within a <code>match</code> section means that the operands are not constrained in any way. If elided within a <code>rewrite</code> section, the operation is treated as having no operands. When present, the operands of an operation expression are interpreted in the following ways:<ol><li>A single instance of type <code>ValueRange</code>:</li></ol>In this case, the single range is treated as all of the operands of the operation:<pre tabindex=0><code class=language-pdll data-lang=pdll>// Define an instance with single range of operands. let root = op<my_dialect.foo>(allOperands: ValueRange); </code></pre><ol start=2><li>A variadic number of either <code>Value</code> or <code>ValueRange</code>:</li></ol>In this case, the inputs are expected to correspond with the operand groups as defined on the operation in ODS.Given the following operation definition in ODS:<div class=highlight><pre tabindex=0 class=chroma><code class=language-tablegen data-lang=tablegen>def MyIndirectCallOp { let arguments = (ins FunctionType:$call, Variadic<AnyType>:$args); } </code></pre></div>We can match the operands as so:<pre tabindex=0><code class=language-pdll data-lang=pdll>let root = op<my_dialect.indirect_call>(call: Value, args: ValueRange); </code></pre><h4 id=results>Results <a class=headline-hash href=#results>¶</a></h4>The results section corresponds to the result types of the operation. This section of an operation expression may be elided, which within a <code>match</code> section means that the result types are not constrained in any way. If elided within a <code>rewrite</code> section, the results of the operation are <a href=#inferred-results>inferred</a>. When present, the result types of an operation expression are interpreted in the following ways:<ol><li>A single instance of type <code>TypeRange</code>:</li></ol>In this case, the single range is treated as all of the result types of the operation:<pre tabindex=0><code class=language-pdll data-lang=pdll>// Define an instance with single range of types. let root = op<my_dialect.foo> -> (allResultTypes: TypeRange); </code></pre><ol start=2><li>A variadic number of either <code>Type</code> or <code>TypeRange</code>:</li></ol>In this case, the inputs are expected to correspond with the result groups as defined on the operation in ODS.Given the following operation definition in ODS:<div class=highlight><pre tabindex=0 class=chroma><code class=language-tablegen data-lang=tablegen>def MyOp { let results = (outs SomeType:$result, Variadic<SomeType>:$otherResults); } </code></pre></div>We can match the result types as so:<pre tabindex=0><code class=language-pdll data-lang=pdll>let root = op<my_dialect.op> -> (result: Type, otherResults: TypeRange); </code></pre><h4 id=inferred-results>Inferred Results <a class=headline-hash href=#inferred-results>¶</a></h4>Within the <code>rewrite</code> section of a pattern, the result types of an operation are inferred if they are elided or otherwise not previously bound. The <a href=#variable-binding>“variable binding”</a> section above discusses the concept of “binding” in more detail. Below are various examples that build upon this to help showcase how a result type may be “bound”:<ul><li>Binding to a <a href=#type-expression>constant</a>:</li></ul><pre tabindex=0><code class=language-pdll data-lang=pdll>op<my_dialect.op> -> (type<"i32">); </code></pre><ul><li>Binding to types within the <code>match</code> section:</li></ul><pre tabindex=0><code class=language-pdll data-lang=pdll>Pattern { replace op<dialect.inputOp> -> (resultTypes: TypeRange) with op<dialect.outputOp> -> (resultTypes); } </code></pre><ul><li>Binding to previously inferred types:</li></ul><pre tabindex=0><code class=language-pdll data-lang=pdll>Pattern { rewrite root: Op with { // `resultTypes` here is *not* yet bound, and will be inferred when // creating `dialect.op`. Any uses of `resultTypes` after this expression, // will use the types inferred when creating this operation. op<dialect.op> -> (resultTypes: TypeRange); // `resultTypes` here is bound to the types inferred when creating `dialect.op`. op<dialect.bar> -> (resultTypes); }; } </code></pre><ul><li>Binding to a <a href=#native-rewriters><code>Native Rewrite</code></a> method result:</li></ul><pre tabindex=0><code class=language-pdll data-lang=pdll>Rewrite BuildTypes() -> TypeRange; Pattern { rewrite root: Op with { op<dialect.op> -> (BuildTypes()); }; } </code></pre>Below are the set of contexts in which result type inferrence is supported:<h5 id=inferred-results-of-replacement-operation>Inferred Results of Replacement Operation <a class=headline-hash href=#inferred-results-of-replacement-operation>¶</a></h5>Replacements have the invariant that the types of the replacement values must match the result types of the input operation. This means that when replacing one operation with another, the result types of the replacement operation may be inferred from the result types of the operation being replaced. For example, consider the following pattern:<pre tabindex=0><code class=language-pdll data-lang=pdll>Pattern => replace op<dialect.inputOp> with op<dialect.outputOp>; </code></pre>This pattern could be written in a more explicit way as:<pre tabindex=0><code class=language-pdll data-lang=pdll>Pattern { replace op<dialect.inputOp> -> (resultTypes: TypeRange) with op<dialect.outputOp> -> (resultTypes); } </code></pre><h5 id=inferred-results-with-infertypeopinterface>Inferred Results with InferTypeOpInterface <a class=headline-hash href=#inferred-results-with-infertypeopinterface>¶</a></h5><code>InferTypeOpInterface</code> is an interface that enables operations to infer its result types from its input attributes, operands, regions, etc. When the result types of an operation cannot be inferred from any other context, this interface is invoked to infer the result types of the operation.<h4 id=attributes>Attributes <a class=headline-hash href=#attributes>¶</a></h4>The attributes section of the operation expression corresponds to the attribute dictionary of the operation. This section of an operation expression may be elided, in which case the attributes are not constrained in any way. The composition of this component maps exactly to how attribute dictionaries are structured in the MLIR textual assembly format:<pre tabindex=0><code class=language-pdll data-lang=pdll>let root = op<my_dialect.foo> {attr1 = attrValue: Attr, attr2 = attrValue2: Attr}; </code></pre>Within the <code>{}</code> attribute entries are specified by an identifier or string name, corresponding to the attribute name, followed by an assignment to the attribute value. If the attribute value is elided, the value of the attribute is implicitly defined as a <a href=https://mlir.llvm.org/docs/Dialects/Builtin/#unitattr><code>UnitAttr</code></a>.<pre tabindex=0><code class=language-pdll data-lang=pdll>let unitConstant = op<my_dialect.constant> {value}; </code></pre><h5 id=accessing-operation-results>Accessing Operation Results <a class=headline-hash href=#accessing-operation-results>¶</a></h5>In multi-operation patterns, the result of one operation often feeds as an input into another. The result groups of an operation may be accessed by name or by index via the <code>.</code> operator:Note: Remember to import the definition of your operation via <a href=#%60.td%60_includes>include</a> to ensure it is visible to PDLL.Given the following operation definition in ODS:<div class=highlight><pre tabindex=0 class=chroma><code class=language-tablegen data-lang=tablegen>def MyResultOp { let results = (outs SomeType:$result); } def MyInputOp { let arguments = (ins SomeType:$input, SomeType:$input); } </code></pre></div>We can write a pattern where <code>MyResultOp</code> feeds into <code>MyInputOp</code> as so:<pre tabindex=0><code class=language-pdll data-lang=pdll>// In this example, we use both `result`(the name) and `0`(the index) to refer to // the first result group of `resultOp`. // Note: If we elide the result types section within the match section, it means // they aren't constrained, not that the operation has no results. let resultOp = op<my_dialect.result_op>; let inputOp = op<my_dialect.input_op>(resultOp.result, resultOp.0); </code></pre>Along with result name access, variables of <code>Op</code> type may implicitly convert to <code>Value</code> or <code>ValueRange</code>. If these variables are registered (has ODS entry), they are converted to <code>Value</code> when they are known to only have one result, otherwise they will be converted to <code>ValueRange</code>:<pre tabindex=0><code class=language-pdll data-lang=pdll>// `resultOp` may also convert implicitly to a Value for use in `inputOp`: let resultOp = op<my_dialect.result_op>; let inputOp = op<my_dialect.input_op>(resultOp); // We could also inline `resultOp` directly: let inputOp = op<my_dialect.input_op>(op<my_dialect.result_op>); </code></pre><h4 id=unregistered-operations>Unregistered Operations <a class=headline-hash href=#unregistered-operations>¶</a></h4>A variable of unregistered op is still available for numeric result indexing. Given that we don’t have knowledge of its result groups, numeric indexing returns a Value corresponding to the individual result at the given index.<pre tabindex=0><code class=language-pdll data-lang=pdll>// Use the index `0` to refer to the first result value of the unregistered op. let inputOp = op<my_dialect.input_op>(op<my_dialect.unregistered_op>.0); </code></pre><h3 id=attribute-expression>Attribute Expression <a class=headline-hash href=#attribute-expression>¶</a></h3>An attribute expression represents a literal MLIR attribute. It allows for statically specifying an MLIR attribute to use, by specifying the textual form of that attribute.<pre tabindex=0><code class=language-pdll data-lang=pdll>let trueConstant = op<arith.constant> {value = attr<"true">}; let applyResult = op<affine.apply>(args: ValueRange) {map = attr<"affine_map<(d0, d1) -> (d1 - 3)>">} </code></pre><h3 id=type-expression>Type Expression <a class=headline-hash href=#type-expression>¶</a></h3>A type expression represents a literal MLIR type. It allows for statically specifying an MLIR type to use, by specifying the textual form of that type.<pre tabindex=0><code class=language-pdll data-lang=pdll>let i32Constant = op<arith.constant> -> (type<"i32">); </code></pre><h3 id=tuples>Tuples <a class=headline-hash href=#tuples>¶</a></h3>PDLL provides native support for tuples, which are used to group multiple elements into a single compound value. The values in a tuple can be of any type, and do not need to be of the same type. There is also no limit to the number of elements held by a tuple. The elements of a tuple can be accessed by index:<pre tabindex=0><code class=language-pdll data-lang=pdll>let tupleValue = (op<my_dialect.foo>, attr<"10 : i32">, type<"i32">); let opValue = tupleValue.0; let attrValue = tupleValue.1; let typeValue = tupleValue.2; </code></pre>You can also name the elements of a tuple and use those names to refer to the values of the individual elements. An element name consists of an identifier followed immediately by an equal (=).<pre tabindex=0><code class=language-pdll data-lang=pdll>let tupleValue = ( opValue = op<my_dialect.foo>, attr<"10 : i32">, typeValue = type<"i32"> ); let opValue = tupleValue.opValue; let attrValue = tupleValue.1; let typeValue = tupleValue.typeValue; </code></pre>Tuples are used to represent multiple results from a <a href=#constraints-with-multiple-results>constraint</a> or <a href=#rewrites-with-multiple-results>rewrite</a>.<h3 id=constraints-1>Constraints <a class=headline-hash href=#constraints-1>¶</a></h3>Constraints provide the ability to inject additional checks on the input IR within the <code>match</code> section of a pattern. Constraints can be applied anywhere within the <code>match</code> section, and depending on the type can either be applied via the constraint list of a <a href=#variables>variable</a> or via the call operator (e.g. <code>MyConstraint(...)</code>). There are three main categories of constraints:<h4 id=core-constraints>Core Constraints <a class=headline-hash href=#core-constraints>¶</a></h4>PDLL defines a number of core constraints that constrain the type of the IR entity. These constraints can only be applied via the <a href=#variable-constraints>constraint list</a> of a variable.<ul><li><code>Attr</code> (<code><</code> type <code>></code>)?</li></ul>A single entity constraint that corresponds to an <code>mlir::Attribute</code>. This constraint optionally takes a type component that constrains the result type of the attribute.<pre tabindex=0><code class=language-pdll data-lang=pdll>// Define a simple variable using the `Attr` constraint. let attr: Attr; let constant = op<arith.constant> {value = attr}; // Define a simple variable using the `Attr` constraint, that has its type // constrained as well. let attrType: Type; let attr: Attr<attrType>; let constant = op<arith.constant> {value = attr}; </code></pre><ul><li><code>Op</code> (<code><</code> op-name <code>></code>)?</li></ul>A single entity constraint that corresponds to an <code>mlir::Operation *</code>.<pre tabindex=0><code class=language-pdll data-lang=pdll>// Match only when the input is from another operation. let inputOp: Op; let root = op<my_dialect.foo>(inputOp); // Match only when the input is from another `my_dialect.foo` operation. let inputOp: Op<my_dialect.foo>; let root = op<my_dialect.foo>(inputOp); </code></pre><ul><li><code>Type</code></li></ul>A single entity constraint that corresponds to an <code>mlir::Type</code>.<pre tabindex=0><code class=language-pdll data-lang=pdll>// Define a simple variable using the `Type` constraint. let resultType: Type; let root = op<my_dialect.foo> -> (resultType); </code></pre><ul><li><code>TypeRange</code></li></ul>A single entity constraint that corresponds to a <code>mlir::TypeRange</code>.<pre tabindex=0><code class=language-pdll data-lang=pdll>// Define a simple variable using the `TypeRange` constraint. let resultTypes: TypeRange; let root = op<my_dialect.foo> -> (resultTypes); </code></pre><ul><li><code>Value</code> (<code><</code> type-expr <code>></code>)?</li></ul>A single entity constraint that corresponds to an <code>mlir::Value</code>. This constraint optionally takes a type component that constrains the result type of the value.<pre tabindex=0><code class=language-pdll data-lang=pdll>// Define a simple variable using the `Value` constraint. let value: Value; let root = op<my_dialect.foo>(value); // Define a variable using the `Value` constraint, that has its type constrained // to be same as the result type of the `root` op. let valueType: Type; let input: Value<valueType>; let root = op<my_dialect.foo>(input) -> (valueType); </code></pre><ul><li><code>ValueRange</code> (<code><</code> type-expr <code>></code>)?</li></ul>A single entity constraint that corresponds to a <code>mlir::ValueRange</code>. This constraint optionally takes a type component that constrains the result types of the value range.<pre tabindex=0><code class=language-pdll data-lang=pdll>// Define a simple variable using the `ValueRange` constraint. let inputs: ValueRange; let root = op<my_dialect.foo>(inputs); // Define a variable using the `ValueRange` constraint, that has its types // constrained to be same as the result types of the `root` op. let valueTypes: TypeRange; let inputs: ValueRange<valueTypes>; let root = op<my_dialect.foo>(inputs) -> (valueTypes); </code></pre><h4 id=defining-constraints-in-pdll>Defining Constraints in PDLL <a class=headline-hash href=#defining-constraints-in-pdll>¶</a></h4>Aside from the core constraints, additional constraints can also be defined within PDLL. This allows for building matcher fragments that can be composed across many different patterns. A constraint in PDLL is defined similarly to a function in traditional programming languages; it contains a name, a set of input arguments, a set of result types, and a body. Results of a constraint are returned via a <code>return</code> statement. A few examples are shown below:<pre tabindex=0><code class=language-pdll data-lang=pdll>/// A constraint that takes an input and constrains the use to an operation of /// a given type. Constraint UsedByFooOp(value: Value) { op<my_dialect.foo>(value); } /// A constraint that returns a result of an existing operation. Constraint ExtractResult(op: Op<my_dialect.foo>) -> Value { return op.result; } Pattern { let value = ExtractResult(op<my_dialect.foo>); UsedByFooOp(value); } </code></pre><h5 id=constraints-with-multiple-results>Constraints with multiple results <a class=headline-hash href=#constraints-with-multiple-results>¶</a></h5>Constraints can return multiple results by returning a tuple of values. When returning multiple results, each result can also be assigned a name to use when indexing that tuple element. Tuple elements can be referenced by their index number, or by name if they were assigned one.<pre tabindex=0><code class=language-pdll data-lang=pdll>// A constraint that returns multiple results, with some of the results assigned // a more readable name. Constraint ExtractMultipleResults(op: Op<my_dialect.foo>) -> (Value, result1: Value) { return (op.result1, op.result2); } Pattern { // Return a tuple of values. let result = ExtractMultipleResults(op: op<my_dialect.foo>); // Index the tuple elements by index, or by name. replace op<my_dialect.foo> with (result.0, result.1, result.result1); } </code></pre><h5 id=constraint-result-type-inference>Constraint result type inference <a class=headline-hash href=#constraint-result-type-inference>¶</a></h5>In addition to explicitly specifying the results of the constraint via the constraint signature, PDLL defined constraints also support inferring the result type from the return statement. Result type inference is active whenever the constraint is defined with no result constraints:<pre tabindex=0><code class=language-pdll data-lang=pdll>// This constraint returns a derived operation. Constraint ReturnSelf(op: Op<my_dialect.foo>) { return op; } // This constraint returns a tuple of two Values. Constraint ExtractMultipleResults(op: Op<my_dialect.foo>) { return (result1 = op.result1, result2 = op.result2); } Pattern { let values = ExtractMultipleResults(op<my_dialect.foo>); replace op<my_dialect.foo> with (values.result1, values.result2); } </code></pre><h5 id=single-line-lambda-body-1>Single Line “Lambda” Body <a class=headline-hash href=#single-line-lambda-body-1>¶</a></h5>Constraints generally define their body using a compound block of statements, as shown below:<pre tabindex=0><code class=language-pdll data-lang=pdll>Constraint ReturnSelf(op: Op<my_dialect.foo>) { return op; } Constraint ExtractMultipleResults(op: Op<my_dialect.foo>) { return (result1 = op.result1, result2 = op.result2); } </code></pre>Constraints also support a lambda-like syntax for specifying simple single line bodies. The lambda body of a Constraint expects a single expression, which is implicitly returned:<pre tabindex=0><code class=language-pdll data-lang=pdll>Constraint ReturnSelf(op: Op<my_dialect.foo>) => op; Constraint ExtractMultipleResults(op: Op<my_dialect.foo>) => (result1 = op.result1, result2 = op.result2); </code></pre><h4 id=native-constraints>Native Constraints <a class=headline-hash href=#native-constraints>¶</a></h4>Constraints may also be defined outside of PDLL, and registered natively within the C++ API.<h5 id=importing-existing-native-constraints>Importing existing Native Constraints <a class=headline-hash href=#importing-existing-native-constraints>¶</a></h5>Constraints defined externally can be imported into PDLL by specifying a constraint “declaration”. This is similar to the PDLL form of defining a constraint but omits the body. Importing the declaration in this form allows for PDLL to statically know the expected input and output types.<pre tabindex=0><code class=language-pdll data-lang=pdll>// Import a single entity value native constraint that checks if the value has a // single use. This constraint must be registered by the consumer of the // compiled PDL. Constraint HasOneUse(value: Value); // Import a multi-entity type constraint that checks if two values have the same // element type. Constraint HasSameElementType(value1: Value, value2: Value); Pattern { // A single entity constraint can be applied via the variable argument list. let value: HasOneUse; // Otherwise, constraints can be applied via the call operator: let value: Value = ...; let value2: Value = ...; HasOneUse(value); HasSameElementType(value, value2); } </code></pre>External constraints are those registered explicitly with the <code>RewritePatternSet</code> via the C++ PDL API. For example, the constraints above may be registered as:<div class=highlight><pre tabindex=0 class=chroma><code class=language-c++ data-lang=c++>static LogicalResult hasOneUseImpl(PatternRewriter &rewriter, Value value) { return success(value.hasOneUse()); } static LogicalResult hasSameElementTypeImpl(PatternRewriter &rewriter, Value value1, Value Value2) { return success(value1.getType().cast<ShapedType>().getElementType() == value2.getType().cast<ShapedType>().getElementType()); } void registerNativeConstraints(RewritePatternSet &patterns) { patternList.getPDLPatterns().registerConstraintFunction( "HasOneUse", hasOneUseImpl); patternList.getPDLPatterns().registerConstraintFunction( "HasSameElementType", hasSameElementTypeImpl); } </code></pre></div><h5 id=defining-native-constraints-in-pdll>Defining Native Constraints in PDLL <a class=headline-hash href=#defining-native-constraints-in-pdll>¶</a></h5>In addition to importing native constraints, PDLL also supports defining native constraints directly when compiling ahead-of-time (AOT) for C++. These constraints can be defined by specifying a string code block after the constraint declaration:<pre tabindex=0><code class=language-pdll data-lang=pdll>Constraint HasOneUse(value: Value) [{ return success(value.hasOneUse()); }]; Constraint HasSameElementType(value1: Value, value2: Value) [{ return success(value1.getType().cast<ShapedType>().getElementType() == value2.getType().cast<ShapedType>().getElementType()); }]; Pattern { // A single entity constraint can be applied via the variable argument list. let value: HasOneUse; // Otherwise, constraints can be applied via the call operator: let value: Value = ...; let value2: Value = ...; HasOneUse(value); HasSameElementType(value, value2); } </code></pre>The arguments of the constraint are accessible within the code block via the same name. See the <a href=#native-constraint-type-translations>“type translation”</a> below for detailed information on how PDLL types are converted to native types. In addition to the PDLL arguments, the code block may also access the current <code>PatternRewriter</code> using <code>rewriter</code>. The result type of the native constraint function is implicitly defined as a <code>::llvm::LogicalResult</code>.Taking the constraints defined above as an example, these function would roughly be translated into:<div class=highlight><pre tabindex=0 class=chroma><code class=language-c++ data-lang=c++>LogicalResult HasOneUse(PatternRewriter &rewriter, Value value) { return success(value.hasOneUse()); } LogicalResult HasSameElementType(Value value1, Value value2) { return success(value1.getType().cast<ShapedType>().getElementType() == value2.getType().cast<ShapedType>().getElementType()); } </code></pre></div>TODO: Native constraints should also be allowed to return values in certain cases.<h6 id=native-constraint-type-translations>Native Constraint Type Translations <a class=headline-hash href=#native-constraint-type-translations>¶</a></h6>The types of argument and result variables are generally mapped to the corresponding MLIR type of the <a href=#constraints>constraint</a> used. Below is a detailed description of how the mapped type of a variable is determined for the various different types of constraints.<ul><li>Attr, Op, Type, TypeRange, Value, ValueRange:</li></ul>These are all core constraints, and are mapped directly to the MLIR equivalent (that their names suggest), namely:<ul><li><code>Attr</code> -> “::mlir::Attribute”</li><li><code>Op</code> -> “::mlir::Operation *”</li><li><code>Type</code> -> “::mlir::Type”</li><li><code>TypeRange</code> -> “::mlir::TypeRange”</li><li><code>Value</code> -> “::mlir::Value”</li><li><code>ValueRange</code> -> “::mlir::ValueRange”</li><li>Op<dialect.name></li></ul>A named operation constraint has a unique translation. If the ODS registration of the referenced operation has been included, the qualified C++ is used. If the ODS information is not available, this constraint maps to “::mlir::Operation *”, similarly to the unnamed variant. For example, given the following:<pre tabindex=0><code class=language-pdll data-lang=pdll>// `my_ops.td` provides the ODS definition of the `my_dialect` operations, such as // `my_dialect.bar` used below. #include "my_ops.td" Constraint Cst(op: Op<my_dialect.bar>) [{ return success(op ... ); }]; </code></pre>The native type used for <code>op</code> may be of the form <code>my_dialect::BarOp</code>, as opposed to the default <code>::mlir::Operation *</code>. Below is a sample translation of the above constraint:<div class=highlight><pre tabindex=0 class=chroma><code class=language-c++ data-lang=c++>LogicalResult Cst(my_dialect::BarOp op) { return success(op ... ); } </code></pre></div><ul><li>Imported ODS Constraints</li></ul>Aside from the core constraints, certain constraints imported from ODS may use a unique native type. How to enable this unique type depends on the ODS constraint construct that was imported:<ul><li><code>Attr</code> constraints<ul><li>Imported <code>Attr</code> constraints utilize the <code>storageType</code> field for native type translation.</li></ul></li><li><code>Type</code> constraints<ul><li>Imported <code>Type</code> constraints utilize the <code>cppClassName</code> field for native type translation.</li></ul></li><li><code>AttrInterface</code>/<code>OpInterface</code>/<code>TypeInterface</code> constraints<ul><li>Imported interfaces utilize the <code>cppInterfaceName</code> field for native type translation.</li></ul></li></ul><h4 id=defining-constraints-inline>Defining Constraints Inline <a class=headline-hash href=#defining-constraints-inline>¶</a></h4>In addition to global scope, PDLL Constraints and Native Constraints defined in PDLL may be specified inline at any level of nesting. This means that they may be defined in Patterns, other Constraints, Rewrites, etc:<pre tabindex=0><code class=language-pdll data-lang=pdll>Constraint GlobalConstraint() { Constraint LocalConstraint(value: Value) { ... }; Constraint LocalNativeConstraint(value: Value) [{ ... }]; let someValue: [LocalConstraint, LocalNativeConstraint] = ...; } </code></pre>Constraints that are defined inline may also elide the name when used directly:<pre tabindex=0><code class=language-pdll data-lang=pdll>Constraint GlobalConstraint(inputValue: Value) { Constraint(value: Value) { ... }(inputValue); Constraint(value: Value) [{ ... }](inputValue); } </code></pre>When defined inline, PDLL constraints may reference any previously defined variable:<pre tabindex=0><code class=language-pdll data-lang=pdll>Constraint GlobalConstraint(op: Op<my_dialect.foo>) { Constraint LocalConstraint() { let results = op.results; }; } </code></pre><h3 id=rewriters>Rewriters <a class=headline-hash href=#rewriters>¶</a></h3>Rewriters define the set of transformations to be performed within the <code>rewrite</code> section of a pattern, and, more specifically, how to transform the input IR after a successful pattern match. All PDLL rewrites must be defined within the <code>rewrite</code> section of the pattern. The <code>rewrite</code> section is denoted by the last statement within the body of the <code>Pattern</code>, which is required to be an <a href=#operation-rewrite-statements>operation rewrite statement</a>. There are two main categories of rewrites in PDLL: operation rewrite statements, and user defined rewrites.<h4 id=operation-rewrite-statements>Operation Rewrite statements <a class=headline-hash href=#operation-rewrite-statements>¶</a></h4>Operation rewrite statements are builtin PDLL statements that perform an IR transformation given a root operation. These statements are the only ones able to start the <code>rewrite</code> section of a pattern, as they allow for properly <a href=#variable-binding>“binding”</a> the root operation of the pattern.<h5 id=erase-statement><code>erase</code> statement <a class=headline-hash href=#erase-statement>¶</a></h5><pre tabindex=0><code class=language-pdll data-lang=pdll>// A pattern that erases all `my_dialect.foo` operations. Pattern => erase op<my_dialect.foo>; </code></pre>The <code>erase</code> statement erases a given operation.<h5 id=replace-statement><code>replace</code> statement <a class=headline-hash href=#replace-statement>¶</a></h5><pre tabindex=0><code class=language-pdll data-lang=pdll>// A pattern that replaces the root operation with its input value. Pattern { let root = op<my_dialect.foo>(input: Value); replace root with input; } // A pattern that replaces the root operation with multiple input values. Pattern { let root = op<my_dialect.foo>(input: Value, _: Value, input2: Value); replace root with (input, input2); } // A pattern that replaces the root operation with another operation. // Note that when an operation is used as the replacement, we can infer its // result types from the input operation. In these cases, the result // types of replacement operation may be elided. Pattern { // Note: In this pattern we also inlined the `root` expression. replace op<my_dialect.foo> with op<my_dialect.bar>; } </code></pre>The <code>replace</code> statement allows for replacing a given root operation with either another operation, or a set of input <code>Value</code> and <code>ValueRange</code> values. When an operation is used as the replacement, we allow infering the result types from the input operation. In these cases, the result types of replacement operation may be elided. Note that no other components aside from the result types will be inferred from the input operation during the replacement.<h5 id=rewrite-statement><code>rewrite</code> statement <a class=headline-hash href=#rewrite-statement>¶</a></h5><pre tabindex=0><code class=language-pdll data-lang=pdll>// A simple pattern that replaces the root operation with its input value. Pattern { let root = op<my_dialect.foo>(input: Value); rewrite root with { ... replace root with input; }; } </code></pre>The <code>rewrite</code> statement allows for rewriting a given root operation with a block of nested rewriters. The root operation is not implicitly erased or replaced, and any transformations to it must be expressed within the nested rewrite block. The inner body may contain any number of other rewrite statements, variables, or expressions.<h4 id=defining-rewriters-in-pdll>Defining Rewriters in PDLL <a class=headline-hash href=#defining-rewriters-in-pdll>¶</a></h4>Additional rewrites can also be defined within PDLL, which allows for building rewrite fragments that can be composed across many different patterns. A rewriter in PDLL is defined similarly to a function in traditional programming languages; it contains a name, a set of input arguments, a set of result types, and a body. Results of a rewrite are returned via a <code>return</code> statement. A few examples are shown below:<pre tabindex=0><code class=language-pdll data-lang=pdll>// A rewrite that constructs and returns a new operation, given an input value. Rewrite BuildFooOp(value: Value) -> Op { return op<my_dialect.foo>(value); } Pattern { // We invoke the rewrite in the same way as functions in traditional // languages. replace op<my_dialect.old_op>(input: Value) with BuildFooOp(input); } </code></pre><h5 id=rewrites-with-multiple-results>Rewrites with multiple results <a class=headline-hash href=#rewrites-with-multiple-results>¶</a></h5>Rewrites can return multiple results by returning a tuple of values. When returning multiple results, each result can also be assigned a name to use when indexing that tuple element. Tuple elements can be referenced by their index number, or by name if they were assigned one.<pre tabindex=0><code class=language-pdll data-lang=pdll>// A rewrite that returns multiple results, with some of the results assigned // a more readable name. Rewrite CreateRewriteOps() -> (Op, result1: ValueRange) { return (op<my_dialect.bar>, op<my_dialect.foo>); } Pattern { rewrite root: Op<my_dialect.foo> with { // Invoke the rewrite, which returns a tuple of values. let result = CreateRewriteOps(); // Index the tuple elements by index, or by name. replace root with (result.0, result.1, result.result1); } } </code></pre><h5 id=rewrite-result-type-inference>Rewrite result type inference <a class=headline-hash href=#rewrite-result-type-inference>¶</a></h5>In addition to explicitly specifying the results of the rewrite via the rewrite signature, PDLL defined rewrites also support inferring the result type from the return statement. Result type inference is active whenever the rewrite is defined with no result constraints:<pre tabindex=0><code class=language-pdll data-lang=pdll>// This rewrite returns a derived operation. Rewrite ReturnSelf(op: Op<my_dialect.foo>) => op; // This rewrite returns a tuple of two Values. Rewrite ExtractMultipleResults(op: Op<my_dialect.foo>) { return (result1 = op.result1, result2 = op.result2); } Pattern { rewrite root: Op<my_dialect.foo> with { let values = ExtractMultipleResults(op<my_dialect.foo>); replace root with (values.result1, values.result2); } } </code></pre><h5 id=single-line-lambda-body-2>Single Line “Lambda” Body <a class=headline-hash href=#single-line-lambda-body-2>¶</a></h5>Rewrites generally define their body using a compound block of statements, as shown below:<pre tabindex=0><code class=language-pdll data-lang=pdll>Rewrite ReturnSelf(op: Op<my_dialect.foo>) { return op; } Rewrite EraseOp(op: Op) { erase op; } </code></pre>Rewrites also support a lambda-like syntax for specifying simple single line bodies. The lambda body of a Rewrite expects a single expression, which is implicitly returned, or a single <a href=#operation-rewrite-statements>operation rewrite statement</a>:<pre tabindex=0><code class=language-pdll data-lang=pdll>Rewrite ReturnSelf(op: Op<my_dialect.foo>) => op; Rewrite EraseOp(op: Op) => erase op; </code></pre><h4 id=native-rewriters>Native Rewriters <a class=headline-hash href=#native-rewriters>¶</a></h4>Rewriters may also be defined outside of PDLL, and registered natively within the C++ API.<h5 id=importing-existing-native-rewrites>Importing existing Native Rewrites <a class=headline-hash href=#importing-existing-native-rewrites>¶</a></h5>Rewrites defined externally can be imported into PDLL by specifying a rewrite “declaration”. This is similar to the PDLL form of defining a rewrite but omits the body. Importing the declaration in this form allows for PDLL to statically know the expected input and output types.<pre tabindex=0><code class=language-pdll data-lang=pdll>// Import a single input native rewrite that returns a new operation. This // rewrite must be registered by the consumer of the compiled PDL. Rewrite BuildOp(value: Value) -> Op; Pattern { replace op<my_dialect.old_op>(input: Value) with BuildOp(input); } </code></pre>External rewrites are those registered explicitly with the <code>RewritePatternSet</code> via the C++ PDL API. For example, the rewrite above may be registered as:<div class=highlight><pre tabindex=0 class=chroma><code class=language-c++ data-lang=c++>static Operation *buildOpImpl(PDLResultList &results, Value value) { // insert special rewrite logic here. Operation *resultOp = ...; return resultOp; } void registerNativeRewrite(RewritePatternSet &patterns) { patterns.getPDLPatterns().registerRewriteFunction("BuildOp", buildOpImpl); } </code></pre></div><h5 id=defining-native-rewrites-in-pdll>Defining Native Rewrites in PDLL <a class=headline-hash href=#defining-native-rewrites-in-pdll>¶</a></h5>In addition to importing native rewrites, PDLL also supports defining native rewrites directly when compiling ahead-of-time (AOT) for C++. These rewrites can be defined by specifying a string code block after the rewrite declaration:<pre tabindex=0><code class=language-pdll data-lang=pdll>Rewrite BuildOp(value: Value) -> (foo: Op<my_dialect.foo>, bar: Op<my_dialect.bar>) [{ return {rewriter.create<my_dialect::FooOp>(value), rewriter.create<my_dialect::BarOp>()}; }]; Pattern { let root = op<my_dialect.foo>(input: Value); rewrite root with { // Invoke the native rewrite and use the results when replacing the root. let results = BuildOp(input); replace root with (results.foo, results.bar); } } </code></pre>The arguments of the rewrite are accessible within the code block via the same name. See the <a href=#native-rewrite-type-translations>“type translation”</a> below for detailed information on how PDLL types are converted to native types. In addition to the PDLL arguments, the code block may also access the current <code>PatternRewriter</code> using <code>rewriter</code>. See the <a href=#native-rewrite-result-translation>“result translation”</a> section for detailed information on how the result type of the native function is determined.Taking the rewrite defined above as an example, this function would roughly be translated into:<div class=highlight><pre tabindex=0 class=chroma><code class=language-c++ data-lang=c++>std::tuple<my_dialect::FooOp, my_dialect::BarOp> BuildOp(Value value) { return {rewriter.create<my_dialect::FooOp>(value), rewriter.create<my_dialect::BarOp>()}; } </code></pre></div><h6 id=native-rewrite-type-translations>Native Rewrite Type Translations <a class=headline-hash href=#native-rewrite-type-translations>¶</a></h6>The types of argument and result variables are generally mapped to the corresponding MLIR type of the <a href=#constraints>constraint</a> used. The rules of native <code>Rewrite</code> type translation are identical to those of native <code>Constraint</code>s, please view the corresponding <a href=#native-constraint-type-translations>native <code>Constraint</code> type translation</a> section for a detailed description of how the mapped type of a variable is determined.<h6 id=native-rewrite-result-translation>Native Rewrite Result Translation <a class=headline-hash href=#native-rewrite-result-translation>¶</a></h6>The results of a native rewrite are directly translated to the results of the native function, using the type translation rules <a href=#native-rewrite-type-translations>described above</a>. The section below describes the various result translation scenarios:<ul><li>Zero Result</li></ul><pre tabindex=0><code class=language-pdll data-lang=pdll>Rewrite createOp() [{ rewriter.create<my_dialect::FooOp>(); }]; </code></pre>In the case where a native <code>Rewrite</code> has no results, the native function returns <code>void</code>:<div class=highlight><pre tabindex=0 class=chroma><code class=language-c++ data-lang=c++>void createOp(PatternRewriter &rewriter) { rewriter.create<my_dialect::FooOp>(); } </code></pre></div><ul><li>Single Result</li></ul><pre tabindex=0><code class=language-pdll data-lang=pdll>Rewrite createOp() -> Op<my_dialect.foo> [{ return rewriter.create<my_dialect::FooOp>(); }]; </code></pre>In the case where a native <code>Rewrite</code> has a single result, the native function returns the corresponding native type for that single result:<div class=highlight><pre tabindex=0 class=chroma><code class=language-c++ data-lang=c++>my_dialect::FooOp createOp(PatternRewriter &rewriter) { return rewriter.create<my_dialect::FooOp>(); } </code></pre></div><ul><li>Multi Result</li></ul><pre tabindex=0><code class=language-pdll data-lang=pdll>Rewrite complexRewrite(value: Value) -> (Op<my_dialect.foo>, FunctionOpInterface) [{ ... }]; </code></pre>In the case where a native <code>Rewrite</code> has multiple results, the native function returns a <code>std::tuple<...></code> containing the corresponding native types for each of the results:<div class=highlight><pre tabindex=0 class=chroma><code class=language-c++ data-lang=c++>std::tuple<my_dialect::FooOp, FunctionOpInterface> complexRewrite(PatternRewriter &rewriter, Value value) { ... } </code></pre></div><h4 id=defining-rewrites-inline>Defining Rewrites Inline <a class=headline-hash href=#defining-rewrites-inline>¶</a></h4>In addition to global scope, PDLL Rewrites and Native Rewrites defined in PDLL may be specified inline at any level of nesting. This means that they may be defined in Patterns, other Rewrites, etc:<pre tabindex=0><code class=language-pdll data-lang=pdll>Rewrite GlobalRewrite(inputValue: Value) { Rewrite localRewrite(value: Value) { ... }; Rewrite localNativeRewrite(value: Value) [{ ... }]; localRewrite(inputValue); localNativeRewrite(inputValue); } </code></pre>Rewrites that are defined inline may also elide the name when used directly:<pre tabindex=0><code class=language-pdll data-lang=pdll>Rewrite GlobalRewrite(inputValue: Value) { Rewrite(value: Value) { ... }(inputValue); Rewrite(value: Value) [{ ... }](inputValue); } </code></pre>When defined inline, PDLL rewrites may reference any previously defined variable:<pre tabindex=0><code class=language-pdll data-lang=pdll>Rewrite GlobalRewrite(op: Op<my_dialect.foo>) { Rewrite localRewrite() { let results = op.results; }; } </code></pre><div class=edit-meta> </div><nav class=pagination><a class="nav nav-prev" href=https://mlir.llvm.org/docs/PatternRewriter/ title="Pattern Rewriting : Generic DAG-to-DAG Rewriting"> Prev - Pattern Rewriting : Generic DAG-to-DAG Rewriting</a> <a class="nav nav-next" href=https://mlir.llvm.org/docs/Quantization/ title=Quantization>Next - Quantization </a></nav><footer>Powered by <a href=https://gohugo.io>Hugo</a>. Theme by <a href=https://themes.gohugo.io/hugo-theme-techdoc/>TechDoc</a>. Designed by <a href=https://github.com/thingsym/hugo-theme-techdoc>Thingsym</a>.</footer></main><div class=sidebar><nav class=slide-menu><ul><li><a href=https://mlir.llvm.org/>Home</a></li><li><a href=https://mlir.llvm.org/users/>Users of MLIR</a></li><li><a href=https://mlir.llvm.org/pubs/>MLIR Related Publications</a></li><li><a href=https://mlir.llvm.org/talks/>Talks</a></li><li><a href=https://mlir.llvm.org/deprecation/>Deprecations & Current Refactoring</a></li><li class=has-sub-menu><a href=https://mlir.llvm.org/getting_started/>Getting Started+</a><ul class=sub-menu><li><a href=https://mlir.llvm.org/getting_started/ReportingIssues/>Reporting Issues</a></li><li><a href=https://mlir.llvm.org/getting_started/Debugging/>Debugging Tips</a></li><li><a href=https://mlir.llvm.org/getting_started/Faq/>FAQ</a></li><li><a href=https://mlir.llvm.org/getting_started/Contributing/>How to Contribute</a></li><li><a href=https://mlir.llvm.org/getting_started/DeveloperGuide/>Developer Guide</a></li><li><a href=https://mlir.llvm.org/getting_started/openprojects/>Open Projects</a></li><li><a href=https://mlir.llvm.org/getting_started/Glossary/>Glossary</a></li><li><a href=https://mlir.llvm.org/getting_started/TestingGuide/>Testing Guide</a></li></ul></li><li class="parent has-sub-menu"><a href=https://mlir.llvm.org/docs/>Code Documentation-</a><ul class=sub-menu><li class=has-sub-menu><a href=https://mlir.llvm.org/docs/Bindings/>Bindings+</a><ul class=sub-menu><li><a href=https://mlir.llvm.org/docs/Bindings/Python/>MLIR Python Bindings</a></li></ul></li><li class=has-sub-menu><a href=https://mlir.llvm.org/docs/Tools/>Tools+</a><ul class=sub-menu><li><a href=https://mlir.llvm.org/docs/Tools/MLIRLSP/>MLIR : Language Server Protocol</a></li><li><a href=https://mlir.llvm.org/docs/Tools/mlir-reduce/>MLIR Reduce</a></li><li><a href=https://mlir.llvm.org/docs/Tools/mlir-rewrite/>mlir-rewrite</a></li></ul></li><li><a href=https://mlir.llvm.org/docs/QuantPasses/></a></li><li><a href=https://mlir.llvm.org/docs/ActionTracing/>Action: Tracing and Debugging MLIR-based Compilers</a></li><li><a href=https://mlir.llvm.org/docs/Bufferization/>Bufferization</a></li><li><a href=https://mlir.llvm.org/docs/DataLayout/>Data Layout Modeling</a></li><li class=has-sub-menu><a href=https://mlir.llvm.org/docs/DefiningDialects/>Defining Dialects+</a><ul class=sub-menu><li><a href=https://mlir.llvm.org/docs/DefiningDialects/Constraints/>Constraints</a></li><li><a href=https://mlir.llvm.org/docs/DefiningDialects/Assembly/>Customizing Assembly Behavior</a></li><li><a href=https://mlir.llvm.org/docs/DefiningDialects/AttributesAndTypes/>Defining Dialect Attributes and Types</a></li><li><a href=https://mlir.llvm.org/docs/DefiningDialects/Operations/>Operation Definition Specification (ODS)</a></li></ul></li><li><a href=https://mlir.llvm.org/docs/Diagnostics/>Diagnostic Infrastructure</a></li><li><a href=https://mlir.llvm.org/docs/DialectConversion/>Dialect Conversion</a></li><li class=has-sub-menu><a href=https://mlir.llvm.org/docs/Dialects/>Dialects+</a><ul class=sub-menu><li><a href=https://mlir.llvm.org/docs/Dialects/OpenACCDialect/>'acc' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/Affine/>'affine' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/AMDGPU/>'amdgpu' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/AMX/>'amx' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/ArithOps/>'arith' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/ArmNeon/>'arm_neon' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/ArmSVE/>'arm_sve' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/ArmSME/>'ArmSME' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/AsyncDialect/>'async' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/BufferizationOps/>'bufferization' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/ControlFlowDialect/>'cf' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/ComplexOps/>'complex' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/DLTIDialect/>'dlti' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/EmitC/>'emitc' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/Func/>'func' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/GPU/>'gpu' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/IndexOps/>'index' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/IRDL/>'irdl' Dialect</a></li><li class=has-sub-menu><a href=https://mlir.llvm.org/docs/Dialects/Linalg/>'linalg' Dialect+</a><ul class=sub-menu><li><a href=https://mlir.llvm.org/docs/Dialects/Linalg/OpDSL/>Linalg OpDSL</a></li></ul></li><li><a href=https://mlir.llvm.org/docs/Dialects/LLVM/>'llvm' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/MathOps/>'math' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/MemRef/>'memref' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/Mesh/>'mesh' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/MLProgramOps/>'ml_program' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/MPI/>'mpi' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/NVGPU/>'nvgpu' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/NVVMDialect/>'nvvm' Dialect</a></li><li class=has-sub-menu><a href=https://mlir.llvm.org/docs/Dialects/OpenMPDialect/>'omp' Dialect+</a><ul class=sub-menu><li><a href=https://mlir.llvm.org/docs/Dialects/OpenMPDialect/ODS/>ODS Documentation</a></li></ul></li><li><a href=https://mlir.llvm.org/docs/Dialects/PDLInterpOps/>'pdl_interp' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/PDLOps/>'pdl' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/PolynomialDialect/>'polynomial' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/PtrOps/>'ptr' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/QuantDialect/>'quant' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/ROCDLDialect/>'rocdl' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/SCFDialect/>'scf' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/ShapeDialect/>'shape' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/SparseTensorOps/>'sparse_tensor' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/TensorOps/>'tensor' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/UBOps/>'ub' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/VCIXDialect/>'vcix' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/Vector/>'vector' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/X86Vector/>'x86vector' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/XeGPU/>'xegpu' Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/Builtin/>Builtin Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/MatchOpInterfaces/>OpInterface definitions</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/SPIR-V/>SPIR-V Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/TOSA/>Tensor Operator Set Architecture (TOSA) Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Dialects/Transform/>Transform Dialect</a></li></ul></li><li><a href=https://mlir.llvm.org/docs/Interfaces/>Interfaces</a></li><li><a href=https://mlir.llvm.org/docs/TargetLLVMIR/>LLVM IR Target</a></li><li><a href=https://mlir.llvm.org/docs/BytecodeFormat/>MLIR Bytecode Format</a></li><li><a href=https://mlir.llvm.org/docs/CAPI/>MLIR C API</a></li><li><a href=https://mlir.llvm.org/docs/LangRef/>MLIR Language Reference</a></li><li><a href=https://mlir.llvm.org/docs/ReleaseNotes/>MLIR Release Notes</a></li><li><a href=https://mlir.llvm.org/docs/Canonicalization/>Operation Canonicalization</a></li><li><a href=https://mlir.llvm.org/docs/OwnershipBasedBufferDeallocation/>Ownership-based Buffer Deallocation</a></li><li><a href=https://mlir.llvm.org/docs/PassManagement/>Pass Infrastructure</a></li><li><a href=https://mlir.llvm.org/docs/Passes/>Passes</a></li><li><a href=https://mlir.llvm.org/docs/PatternRewriter/>Pattern Rewriting : Generic DAG-to-DAG Rewriting</a></li><li class=active><a href=https://mlir.llvm.org/docs/PDLL/>PDLL - PDL Language</a></li><li><a href=https://mlir.llvm.org/docs/Quantization/>Quantization</a></li><li class=has-sub-menu><a href=https://mlir.llvm.org/docs/Rationale/>Rationale+</a><ul class=sub-menu><li><a href=https://mlir.llvm.org/docs/Rationale/RationaleGenericDAGRewriter/>Generic DAG Rewriter Infrastructure Rationale</a></li><li><a href=https://mlir.llvm.org/docs/Rationale/RationaleLinalgDialect/>Linalg Dialect Rationale: The Case For Compiler-Friendly Custom Operations</a></li><li><a href=https://mlir.llvm.org/docs/Rationale/Rationale/>MLIR Rationale</a></li><li><a href=https://mlir.llvm.org/docs/Rationale/MLIRForGraphAlgorithms/>MLIR: Incremental Application to Graph Algorithms in ML Frameworks</a></li><li><a href=https://mlir.llvm.org/docs/Rationale/RationaleSimplifiedPolyhedralForm/>MLIR: The case for a simplified polyhedral form</a></li><li><a href=https://mlir.llvm.org/docs/Rationale/SideEffectsAndSpeculation/>Side Effects & Speculation</a></li><li><a href=https://mlir.llvm.org/docs/Rationale/UsageOfConst/>Usage of 'const' in MLIR, for core IR types</a></li></ul></li><li><a href=https://mlir.llvm.org/docs/ShapeInference/>Shape Inference</a></li><li><a href=https://mlir.llvm.org/docs/SPIRVToLLVMDialectConversion/>SPIR-V Dialect to LLVM Dialect conversion manual</a></li><li><a href=https://mlir.llvm.org/docs/SymbolsAndSymbolTables/>Symbols and Symbol Tables</a></li><li><a href=https://mlir.llvm.org/docs/DeclarativeRewrites/>Table-driven Declarative Rewrite Rule (DRR)</a></li><li class=has-sub-menu><a href=https://mlir.llvm.org/docs/Traits/>Traits+</a><ul class=sub-menu><li><a href=https://mlir.llvm.org/docs/Traits/Broadcastable/>The `Broadcastable` Trait</a></li></ul></li><li class=has-sub-menu><a href=https://mlir.llvm.org/docs/Tutorials/>Tutorials+</a><ul class=sub-menu><li><a href=https://mlir.llvm.org/docs/Tutorials/CreatingADialect/>Creating a Dialect</a></li><li><a href=https://mlir.llvm.org/docs/Tutorials/QuickstartRewrites/>Quickstart tutorial to adding MLIR graph rewrite</a></li><li class=has-sub-menu><a href=https://mlir.llvm.org/docs/Tutorials/Toy/>Toy Tutorial+</a><ul class=sub-menu><li><a href=https://mlir.llvm.org/docs/Tutorials/Toy/Ch-1/>Chapter 1: Toy Language and AST</a></li><li><a href=https://mlir.llvm.org/docs/Tutorials/Toy/Ch-2/>Chapter 2: Emitting Basic MLIR</a></li><li><a href=https://mlir.llvm.org/docs/Tutorials/Toy/Ch-3/>Chapter 3: High-level Language-Specific Analysis and Transformation</a></li><li><a href=https://mlir.llvm.org/docs/Tutorials/Toy/Ch-4/>Chapter 4: Enabling Generic Transformation with Interfaces</a></li><li><a href=https://mlir.llvm.org/docs/Tutorials/Toy/Ch-5/>Chapter 5: Partial Lowering to Lower-Level Dialects for Optimization</a></li><li><a href=https://mlir.llvm.org/docs/Tutorials/Toy/Ch-6/>Chapter 6: Lowering to LLVM and CodeGeneration</a></li><li><a href=https://mlir.llvm.org/docs/Tutorials/Toy/Ch-7/>Chapter 7: Adding a Composite Type to Toy</a></li></ul></li><li class=has-sub-menu><a href=https://mlir.llvm.org/docs/Tutorials/transform/>Transform Dialect Tutorial+</a><ul class=sub-menu><li><a href=https://mlir.llvm.org/docs/Tutorials/transform/Ch0/>Chapter 0: A Primer on “Structured” Linalg Operations</a></li><li><a href=https://mlir.llvm.org/docs/Tutorials/transform/Ch1/>Chapter 1: Combining Existing Transformations</a></li><li><a href=https://mlir.llvm.org/docs/Tutorials/transform/Ch2/>Chapter 2: Adding a Simple New Transformation Operation</a></li><li><a href=https://mlir.llvm.org/docs/Tutorials/transform/Ch3/>Chapter 3: More than Simple Transform Operations</a></li><li><a href=https://mlir.llvm.org/docs/Tutorials/transform/Ch4/>Chapter 4: Matching Payload with Transform Operations</a></li><li><a href=https://mlir.llvm.org/docs/Tutorials/transform/ChH/>Chapter H: Reproducing Halide Schedule</a></li></ul></li><li><a href=https://mlir.llvm.org/docs/Tutorials/UnderstandingTheIRStructure/>Understanding the IR Structure</a></li><li><a href=https://mlir.llvm.org/docs/Tutorials/MlirOpt/>Using `mlir-opt`</a></li><li><a href=https://mlir.llvm.org/docs/Tutorials/DataFlowAnalysis/>Writing DataFlow Analyses in MLIR</a></li></ul></li></ul></li></ul></nav><div class=sidebar-footer></div></div></div><a href=# id=backtothetop-fixed class=backtothetop data-backtothetop-duration=600 data-backtothetop-easing=easeOutQuart data-backtothetop-fixed-fadein=1000 data-backtothetop-fixed-fadeout=1000 data-backtothetop-fixed-bottom=10 data-backtothetop-fixed-right=20> </a></div></body></html>