Skip to content
New issue

Have a question about this project? Sign up for a free GitHub account to open an issue and contact its maintainers and the community.

Sign up for GitHub

By clicking “Sign up for GitHub”, you agree to our terms of service and privacy statement. We’ll occasionally send you account related emails.

Already on GitHub? Sign in to your account

Jump to bottom

[mlir][sparse] add sparsification options to pretty print and debug s… #80205

Merged
merged 2 commits into from
Feb 1, 2024
Merged

[mlir][sparse] add sparsification options to pretty print and debug s… #80205

merged 2 commits into from
Feb 1, 2024

Conversation

PeimingLiu
Copy link
Member

…parse loops.

@llvmbot llvmbot added mlir:sparse Sparse compiler in MLIR mlir labels Jan 31, 2024
@llvmbot
Copy link
Member

llvmbot commented Jan 31, 2024

@llvm/pr-subscribers-mlir

Author: Peiming Liu (PeimingLiu)

Changes

…parse loops.

Full diff: https://github.com/llvm/llvm-project/pull/80205.diff

2 Files Affected:

  • (modified) mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.h (+17-2)
  • (modified) mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.td (+7)
diff --git a/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.h b/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.h
index e93e2aefb344f..8b2875a751d4a 100644
--- a/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.h
+++ b/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.h
@@ -47,6 +47,12 @@ enum class ReinterpretMapScope {
   kExceptGeneric, // reinterprets operation other than linalg.generic
 };
 
+/// Defines a scope for reinterpret map pass.
+enum class DebugSparseIteration {
+  kNone,          // generate fully inlined (and functional) sparse iteration
+  kInterfaceOnly, // generate only place-holder for sparse iteration
+};
+
 #define GEN_PASS_DECL
 #include "mlir/Dialect/SparseTensor/Transforms/Passes.h.inc"
 
@@ -74,11 +80,20 @@ std::unique_ptr<Pass> createPreSparsificationRewritePass();
 
 /// Options for the Sparsification pass.
 struct SparsificationOptions {
+  SparsificationOptions(SparseParallelizationStrategy p, DebugSparseIteration d,
+                        bool enableRT)
+      : parallelizationStrategy(p), debugSparseIteration(d),
+        enableRuntimeLibrary(enableRT) {}
+
   SparsificationOptions(SparseParallelizationStrategy p, bool enableRT)
-      : parallelizationStrategy(p), enableRuntimeLibrary(enableRT) {}
+      : SparsificationOptions(p, DebugSparseIteration::kNone, enableRT) {}
+
   SparsificationOptions()
-      : SparsificationOptions(SparseParallelizationStrategy::kNone, true) {}
+      : SparsificationOptions(SparseParallelizationStrategy::kNone,
+                              DebugSparseIteration::kNone, true) {}
+
   SparseParallelizationStrategy parallelizationStrategy;
+  DebugSparseIteration debugSparseIteration;
   bool enableRuntimeLibrary;
 };
 
diff --git a/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.td b/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.td
index f38779ed9ed2b..126b91510d391 100644
--- a/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.td
+++ b/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.td
@@ -130,6 +130,13 @@ def SparsificationPass : Pass<"sparsification", "ModuleOp"> {
              clEnumValN(mlir::SparseParallelizationStrategy::kAnyStorageAnyLoop,
                         "any-storage-any-loop",
                         "Enable sparse parallelization for any storage and loop."))}]>,
+    Option<"debugSparseIteration", "debug-sparse-iteration", "mlir::DebugSparseIteration",
+           "mlir::DebugSparseIteration::kNone",
+           "Pretty print sparse loops to debug sparse iteration", [{llvm::cl::values(
+             clEnumValN(mlir::DebugSparseIteration::kNone, "none",
+                        "Turn off pretty printing and generates functional code."),
+             clEnumValN(mlir::DebugSparseIteration::kInterfaceOnly, "interface-only",
+                        "Generate non-functional interfaces for sparse iteration."))}]>,
     Option<"enableRuntimeLibrary", "enable-runtime-library", "bool",
            "true", "Enable runtime library for manipulating sparse tensors">,
   ];

@llvmbot
Copy link
Member

llvmbot commented Jan 31, 2024

@llvm/pr-subscribers-mlir-sparse

Author: Peiming Liu (PeimingLiu)

Changes

…parse loops.

Full diff: https://github.com/llvm/llvm-project/pull/80205.diff

2 Files Affected:

  • (modified) mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.h (+17-2)
  • (modified) mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.td (+7)
diff --git a/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.h b/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.h
index e93e2aefb344f..8b2875a751d4a 100644
--- a/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.h
+++ b/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.h
@@ -47,6 +47,12 @@ enum class ReinterpretMapScope {
   kExceptGeneric, // reinterprets operation other than linalg.generic
 };
 
+/// Defines a scope for reinterpret map pass.
+enum class DebugSparseIteration {
+  kNone,          // generate fully inlined (and functional) sparse iteration
+  kInterfaceOnly, // generate only place-holder for sparse iteration
+};
+
 #define GEN_PASS_DECL
 #include "mlir/Dialect/SparseTensor/Transforms/Passes.h.inc"
 
@@ -74,11 +80,20 @@ std::unique_ptr<Pass> createPreSparsificationRewritePass();
 
 /// Options for the Sparsification pass.
 struct SparsificationOptions {
+  SparsificationOptions(SparseParallelizationStrategy p, DebugSparseIteration d,
+                        bool enableRT)
+      : parallelizationStrategy(p), debugSparseIteration(d),
+        enableRuntimeLibrary(enableRT) {}
+
   SparsificationOptions(SparseParallelizationStrategy p, bool enableRT)
-      : parallelizationStrategy(p), enableRuntimeLibrary(enableRT) {}
+      : SparsificationOptions(p, DebugSparseIteration::kNone, enableRT) {}
+
   SparsificationOptions()
-      : SparsificationOptions(SparseParallelizationStrategy::kNone, true) {}
+      : SparsificationOptions(SparseParallelizationStrategy::kNone,
+                              DebugSparseIteration::kNone, true) {}
+
   SparseParallelizationStrategy parallelizationStrategy;
+  DebugSparseIteration debugSparseIteration;
   bool enableRuntimeLibrary;
 };
 
diff --git a/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.td b/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.td
index f38779ed9ed2b..126b91510d391 100644
--- a/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.td
+++ b/mlir/include/mlir/Dialect/SparseTensor/Transforms/Passes.td
@@ -130,6 +130,13 @@ def SparsificationPass : Pass<"sparsification", "ModuleOp"> {
              clEnumValN(mlir::SparseParallelizationStrategy::kAnyStorageAnyLoop,
                         "any-storage-any-loop",
                         "Enable sparse parallelization for any storage and loop."))}]>,
+    Option<"debugSparseIteration", "debug-sparse-iteration", "mlir::DebugSparseIteration",
+           "mlir::DebugSparseIteration::kNone",
+           "Pretty print sparse loops to debug sparse iteration", [{llvm::cl::values(
+             clEnumValN(mlir::DebugSparseIteration::kNone, "none",
+                        "Turn off pretty printing and generates functional code."),
+             clEnumValN(mlir::DebugSparseIteration::kInterfaceOnly, "interface-only",
+                        "Generate non-functional interfaces for sparse iteration."))}]>,
     Option<"enableRuntimeLibrary", "enable-runtime-library", "bool",
            "true", "Enable runtime library for manipulating sparse tensors">,
   ];

@PeimingLiu PeimingLiu merged commit 20a9fa3 into llvm:main Feb 1, 2024
3 of 4 checks passed
@PeimingLiu PeimingLiu deleted the debug-loop-structure branch February 1, 2024 19:51
ichaer added a commit to ichaer/llvm-project-onesided_lower_bound that referenced this pull request Feb 2, 2024
@ichaer
Merge remote-tracking branch 'llvm/main' into onesided_lower_bound
f17fa58 
* llvm/main: (500 commits)
  [docs] Add beginner-focused office hours (llvm#80308)
  [mlir][sparse] external entry method wrapper for sparse tensors (llvm#80326)
  [StackSlotColoring] Ignore non-spill objects in RemoveDeadStores. (llvm#80242)
  [libc][stdbit] fix return types (llvm#80337)
  Revert "[RISCV] Refine cost on Min/Max reduction" (llvm#80340)
  [TTI]Add support for strided loads/stores.
  [analyzer][HTMLRewriter] Cache partial rewrite results. (llvm#80220)
  [flang][openacc][openmp] Use #0 from hlfir.declare value when generating bound ops (llvm#80317)
  [AArch64][PAC] Expand blend(reg, imm) operation in aarch64-pauth pass (llvm#74729)
  [SHT_LLVM_BB_ADDR_MAP][llvm-readobj] Implements llvm-readobj handling for PGOAnalysisMap. (llvm#79520)
  [libc] add bazel support for most of unistd (llvm#80078)
  [clang-tidy] Remove enforcement of rule C.48 from cppcoreguidelines-prefer-member-init (llvm#80330)
  [OpenMP] Fix typo (NFC) (llvm#80332)
  [BOLT] Enable re-writing of Linux kernel binary (llvm#80228)
  [BOLT] Adjust section sizes based on file offsets (llvm#80226)
  [libc] fix stdbit include test when not all entrypoints are available (llvm#80323)
  [RISCV][GISel] RegBank select and instruction select for vector G_ADD, G_SUB (llvm#74114)
  [RISCV] Add srmcfg CSR from Ssqosid extension. (llvm#79914)
  [mlir][sparse] add sparsification options to pretty print and debug s… (llvm#80205)
  [RISCV][MC] MC layer support for the experimental zalasr extension (llvm#79911)
  ...
agozillon pushed a commit to agozillon/llvm-project that referenced this pull request Feb 5, 2024
@PeimingLiu @agozillon
[mlir][sparse] add sparsification options to pretty print and debug s… (
60c0c67 
llvm#80205)

…parse loops.
Sign up for free to join this conversation on GitHub. Already have an account? Sign in to comment
Reviewers

@aartbik aartbik aartbik approved these changes

@yinying-lisa-li yinying-lisa-li yinying-lisa-li approved these changes

@matthias-springer matthias-springer Awaiting requested review from matthias-springer matthias-springer is a code owner

Assignees
No one assigned
Labels
mlir:sparse Sparse compiler in MLIR mlir
Projects
None yet
Milestone
No milestone
Development

Successfully merging this pull request may close these issues.
4 participants
@PeimingLiu @llvmbot @aartbik @yinying-lisa-li