renaming

nod-ai · Nov 29, 2024 · a53b7db · a53b7db
1 parent 8ff316c
commit a53b7db
Show file tree

Hide file tree

Showing 11 changed files with 41 additions and 40 deletions.
diff --git a/compiler/plugins/target/AMD-AIE/iree-amd-aie/Test/samples/matmul_pack_peel_objectfifo.mlir b/compiler/plugins/target/AMD-AIE/iree-amd-aie/Test/samples/matmul_pack_peel_objectfifo.mlir
@@ -1,6 +1,6 @@
 // This pipeline is obtained by going into Passes.cpp, and dumping the pass pipeline (at the end of addAMDAIEObjectFifoLoweringPasses) using `passManager.dump()`. This test is included, as it can be useful to have a reference in IR of all the passes that are run.
 
-// RUN: iree-opt --pass-pipeline="builtin.module(fold-memref-alias-ops,iree-amdaie-distribute-l1-allocations,iree-amdaie-convert-to-dma,iree-amdaie-normalize-loop-bounds,iree-amdaie-insert-cores,iree-amdaie-localize-logicalobjectfifo,cse,iree-amdaie-distribute-cores-and-objectfifos,cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-split-logical-objectfifos-for-connection-reuse,cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-assign-tiles,cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-dma-to-circular-dma,func.func(iree-amdaie-create-aie-workgroup),cse,iree-amdaie-dma-cse,iree-amdaie-hoist-logical-objectfifo,iree-amdaie-canonicalize-doubly-strided-op{fold-single-dims=false},iree-amdaie-flatten-logicalobjectfifo,iree-amdaie-assign-logical-objectfifo-depth{l1-buffer-depth=2 l2-buffer-depth=2 l3-buffer-depth=1},iree-amdaie-access-to-acquire-release,iree-amdaie-none-access-to-temporary-buffer,iree-amdaie-assign-connection-types,cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-dma-composition{only-zero-stride-on-outer-dim=true},cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-dma-cse,iree-amdaie-assign-npu-dma-bd-ids,cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-controlcode-loop-unroll,cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-dma-cse,iree-amdaie-canonicalize-doubly-strided-op{fold-single-dims=false},canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-convert-core-forall-to-for,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-assign-channels,cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-objfifo-bufferization,iree-amdaie-connection-to-flow,iree-amdaie-assign-packet-ids,iree-amdaie-controlcode-to-half-dma-cpy-nd,iree-amdaie-controlcode-to-npu,iree-amdaie-controlcode-to-transaction,iree-amdaie-acquire-release-to-use-lock,iree-amdaie-canonicalize-npu-dma-cpy-nd{nb-dimensions=4},canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-sink-into-core,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-lower-to-aie,iree-amdaie-remove-memoryspace)" --split-input-file %s | FileCheck %s
+// RUN: iree-opt --pass-pipeline="builtin.module(fold-memref-alias-ops,iree-amdaie-distribute-l1-allocations,iree-amdaie-convert-to-dma,iree-amdaie-normalize-loop-bounds,iree-amdaie-insert-cores,iree-amdaie-localize-logicalobjectfifo,cse,iree-amdaie-distribute-cores-and-objectfifos,cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-split-logical-objectfifos-for-connection-reuse,cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-assign-tiles,cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-dma-to-circular-dma,func.func(iree-amdaie-create-aie-workgroup),cse,iree-amdaie-dma-cse,iree-amdaie-hoist-logical-objectfifo,iree-amdaie-canonicalize-doubly-strided-op{fold-single-dims=false},iree-amdaie-flatten-logicalobjectfifo,iree-amdaie-assign-logical-objectfifo-depth{l1-buffer-depth=2 l2-buffer-depth=2 l3-buffer-depth=1},iree-amdaie-access-to-acquire-release,iree-amdaie-none-access-to-temporary-buffer,iree-amdaie-assign-connection-types,cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-dma-composition{only-zero-stride-on-outer-dim=true},cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-dma-cse,iree-amdaie-assign-npu-dma-bd-ids,cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-controlcode-loop-unroll,cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-dma-cse,iree-amdaie-canonicalize-doubly-strided-op{fold-single-dims=false},canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-convert-core-forall-to-for,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-assign-channels,cse,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-objfifo-bufferization,iree-amdaie-connection-to-flow,iree-amdaie-assign-packet-ids,iree-amdaie-npu-dma-to-half-dma-cpy-nd,iree-amdaie-controlcode-lowering,iree-amdaie-controlcode-to-transaction,iree-amdaie-acquire-release-to-use-lock,iree-amdaie-canonicalize-npu-dma-cpy-nd{nb-dimensions=4},canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-sink-into-core,canonicalize{  max-iterations=10 max-num-rewrites=-1 region-simplify=normal test-convergence=false top-down=true},iree-amdaie-lower-to-aie,iree-amdaie-remove-memoryspace)" --split-input-file %s | FileCheck %s
 
 
 

diff --git a/...aie/Transforms/AMDAIEControlCodeToNpu.cpp → .../Transforms/AMDAIEControlCodeLowering.cpp b/...aie/Transforms/AMDAIEControlCodeToNpu.cpp → .../Transforms/AMDAIEControlCodeLowering.cpp
@@ -14,7 +14,7 @@
 #include "mlir/Transforms/DialectConversion.h"
 #include "mlir/Transforms/GreedyPatternRewriteDriver.h"
 
-#define DEBUG_TYPE "iree-amdaie-controlcode-to-npu"
+#define DEBUG_TYPE "iree-amdaie-controlcode-lowering"
 
 namespace mlir::iree_compiler::AMDAIE {
 
@@ -216,16 +216,17 @@ struct HalfDmaCpyNdToNpuConverter final
 };
 
 namespace {
-class AMDAIEControlCodeToNpuPass
-    : public impl::AMDAIEControlCodeToNpuBase<AMDAIEControlCodeToNpuPass> {
+class AMDAIEControlCodeLoweringPass
+    : public impl::AMDAIEControlCodeLoweringBase<
+          AMDAIEControlCodeLoweringPass> {
  public:
   void getDependentDialects(DialectRegistry &registry) const override {
     registry.insert<AMDAIEDialect>();
   }
   void runOnOperation() override;
 };
 
-void AMDAIEControlCodeToNpuPass::runOnOperation() {
+void AMDAIEControlCodeLoweringPass::runOnOperation() {
   Operation *parentOp = getOperation();
   MLIRContext *context = &getContext();
 
@@ -254,8 +255,8 @@ void AMDAIEControlCodeToNpuPass::runOnOperation() {
 
 }  // namespace
 
-std::unique_ptr<Pass> createAMDAIEControlCodeToNpuPass() {
-  return std::make_unique<AMDAIEControlCodeToNpuPass>();
+std::unique_ptr<Pass> createAMDAIEControlCodeLoweringPass() {
+  return std::make_unique<AMDAIEControlCodeLoweringPass>();
 }
 
 }  // namespace mlir::iree_compiler::AMDAIE
diff --git a/...forms/AMDAIEControlCodeToHalfDmaCpyNd.cpp → ...Transforms/AMDAIENpuDmaToHalfDmaCpyNd.cpp b/...forms/AMDAIEControlCodeToHalfDmaCpyNd.cpp → ...Transforms/AMDAIENpuDmaToHalfDmaCpyNd.cpp
@@ -14,11 +14,11 @@
 #include "mlir/Transforms/DialectConversion.h"
 #include "mlir/Transforms/GreedyPatternRewriteDriver.h"
 
-#define DEBUG_TYPE "iree-amdaie-controlcode-to-half-dma-cpy-nd"
+#define DEBUG_TYPE "iree-amdaie-npu-dma-to-half-dma-cpy-nd"
 
 namespace mlir::iree_compiler::AMDAIE {
 
-struct DmaCpyNdToHalfDmaCpyNdConverter final
+struct NpuDmaToHalfDmaCpyNdConverter final
     : OpConversionPattern<AMDAIE::NpuDmaCpyNdOp> {
   using OpConversionPattern::OpConversionPattern;
 
@@ -93,24 +93,24 @@ struct DmaCpyNdToHalfDmaCpyNdConverter final
 };
 
 namespace {
-class AMDAIEControlCodeToHalfDmaCpyNdPass
-    : public impl::AMDAIEControlCodeToHalfDmaCpyNdBase<
-          AMDAIEControlCodeToHalfDmaCpyNdPass> {
+class AMDAIENpuDmaToHalfDmaCpyNdPass
+    : public impl::AMDAIENpuDmaToHalfDmaCpyNdBase<
+          AMDAIENpuDmaToHalfDmaCpyNdPass> {
  public:
   void getDependentDialects(DialectRegistry &registry) const override {
     registry.insert<AMDAIEDialect>();
   }
   void runOnOperation() override;
 };
 
-void AMDAIEControlCodeToHalfDmaCpyNdPass::runOnOperation() {
+void AMDAIENpuDmaToHalfDmaCpyNdPass::runOnOperation() {
   Operation *parentOp = getOperation();
   MLIRContext *context = &getContext();
   RewritePatternSet patterns(context);
   ConversionTarget conversionTarget(*context);
   conversionTarget.addLegalDialect<AMDAIEDialect>();
   conversionTarget.addIllegalOp<AMDAIE::NpuDmaCpyNdOp>();
-  patterns.insert<DmaCpyNdToHalfDmaCpyNdConverter>(context);
+  patterns.insert<NpuDmaToHalfDmaCpyNdConverter>(context);
   if (failed(applyPartialConversion(parentOp, conversionTarget,
                                     std::move(patterns)))) {
     return signalPassFailure();
@@ -119,8 +119,8 @@ void AMDAIEControlCodeToHalfDmaCpyNdPass::runOnOperation() {
 
 }  // namespace
 
-std::unique_ptr<Pass> createAMDAIEControlCodeToHalfDmaCpyNdPass() {
-  return std::make_unique<AMDAIEControlCodeToHalfDmaCpyNdPass>();
+std::unique_ptr<Pass> createAMDAIENpuDmaToHalfDmaCpyNdPass() {
+  return std::make_unique<AMDAIENpuDmaToHalfDmaCpyNdPass>();
 }
 
 }  // namespace mlir::iree_compiler::AMDAIE
diff --git a/compiler/plugins/target/AMD-AIE/iree-amd-aie/Transforms/CMakeLists.txt b/compiler/plugins/target/AMD-AIE/iree-amd-aie/Transforms/CMakeLists.txt
@@ -60,8 +60,7 @@ iree_cc_library(
     "AMDAIEConnectionToFlow.cpp"
     "AMDAIEConvertToDma.cpp"
     "AMDAIEControlCodeForallToFor.cpp"
-    "AMDAIEControlCodeToHalfDmaCpyNd.cpp"
-    "AMDAIEControlCodeToNpu.cpp"
+    "AMDAIEControlCodeLowering.cpp"
     "AMDAIEControlCodeLoopUnroll.cpp"
     "AMDAIEControlCodeToTransaction.cpp"
     "AMDAIEConvertCoreForallToFor.cpp"
@@ -96,6 +95,7 @@ iree_cc_library(
     "AMDAIEMapForallToCores.cpp"
     "AMDAIENoneAccessToTemporaryBuffer.cpp"
     "AMDAIENormalizeLoopBounds.cpp"
+    "AMDAIENpuDmaToHalfDmaCpyNd.cpp"
     "AMDAIEObjFifoBufferization.cpp"
     "AMDAIEPackAndTranspose.cpp"
     "AMDAIEPad.cpp"

diff --git a/compiler/plugins/target/AMD-AIE/iree-amd-aie/Transforms/PassDetail.h b/compiler/plugins/target/AMD-AIE/iree-amd-aie/Transforms/PassDetail.h
@@ -37,8 +37,7 @@ namespace mlir::iree_compiler::AMDAIE {
 #define GEN_PASS_DEF_AMDAIECONNECTIONTOFLOW
 #define GEN_PASS_DEF_AMDAIECONTROLCODEFORALLTOFOR
 #define GEN_PASS_DEF_AMDAIECONTROLCODELOOPUNROLL
-#define GEN_PASS_DEF_AMDAIECONTROLCODETOHALFDMACPYND
-#define GEN_PASS_DEF_AMDAIECONTROLCODETONPU
+#define GEN_PASS_DEF_AMDAIECONTROLCODELOWERING
 #define GEN_PASS_DEF_AMDAIECONTROLCODETOTRANSACTION
 #define GEN_PASS_DEF_AMDAIECONVERTCOREFORALLTOFOR
 #define GEN_PASS_DEF_AMDAIECREATEAIEWORKGROUP
@@ -74,6 +73,7 @@ namespace mlir::iree_compiler::AMDAIE {
 #define GEN_PASS_DEF_AMDAIEMAPFORALLTOCORES
 #define GEN_PASS_DEF_AMDAIENONEACCESSTOTEMPORARYBUFFER
 #define GEN_PASS_DEF_AMDAIENORMALIZELOOPBOUNDS
+#define GEN_PASS_DEF_AMDAIENPUDMATOHALFDMACPYND
 #define GEN_PASS_DEF_AMDAIEOBJFIFOBUFFERIZATION
 #define GEN_PASS_DEF_AMDAIEPACKANDTRANSPOSE
 #define GEN_PASS_DEF_AMDAIECONVERTTODMA

diff --git a/compiler/plugins/target/AMD-AIE/iree-amd-aie/Transforms/Passes.cpp b/compiler/plugins/target/AMD-AIE/iree-amd-aie/Transforms/Passes.cpp
@@ -664,8 +664,8 @@ void addAMDAIEObjectFifoLoweringPasses(
   passManager.addPass(createAMDAIEConnectionToFlowPass());
   passManager.addPass(createAMDAIEAssignPacketIdsPass());
 
-  passManager.addPass(createAMDAIEControlCodeToHalfDmaCpyNdPass());
-  passManager.addPass(createAMDAIEControlCodeToNpuPass());
+  passManager.addPass(createAMDAIENpuDmaToHalfDmaCpyNdPass());
+  passManager.addPass(createAMDAIEControlCodeLoweringPass());
   passManager.addPass(createAMDAIEControlCodeToTransactionPass());
 
   addAMDAIEToAIEPasses(passManager, insertLoopAroundCoreBlock);

diff --git a/compiler/plugins/target/AMD-AIE/iree-amd-aie/Transforms/Passes.h b/compiler/plugins/target/AMD-AIE/iree-amd-aie/Transforms/Passes.h
@@ -121,12 +121,9 @@ std::unique_ptr<Pass> createAMDAIEControlCodeForallToForPass();
 /// Pass to unroll the loops within the control code regions.
 std::unique_ptr<Pass> createAMDAIEControlCodeLoopUnrollPass();
 
-/// Pass to convert control code DMA operations into HalfDmaCpyNd
-std::unique_ptr<Pass> createAMDAIEControlCodeToHalfDmaCpyNdPass();
-
 /// Pass to convert control code HalfDmaCpyNd into NPU WriteBd, AddressPatch,
-/// PushToQueue operations
-std::unique_ptr<Pass> createAMDAIEControlCodeToNpuPass();
+/// PushToQueue operations.
+std::unique_ptr<Pass> createAMDAIEControlCodeLoweringPass();
 
 /// Pass to convert control code into a transaction binary.
 std::unique_ptr<Pass> createAMDAIEControlCodeToTransactionPass(
@@ -258,6 +255,9 @@ std::unique_ptr<Pass> createAMDAIENoneAccessToTemporaryBufferPass();
 /// Normalize the loop bounds of `scf.for` and `scf.forall`.
 std::unique_ptr<Pass> createAMDAIENormalizeLoopBoundsPass();
 
+/// Pass to convert control code DMA operations into HalfDmaCpyNd.
+std::unique_ptr<Pass> createAMDAIENpuDmaToHalfDmaCpyNdPass();
+
 /// Create a pass to bufferize logical objectFifos.
 std::unique_ptr<Pass> createAMDAIEObjFifoBufferizationPass();
 

diff --git a/compiler/plugins/target/AMD-AIE/iree-amd-aie/Transforms/Passes.td b/compiler/plugins/target/AMD-AIE/iree-amd-aie/Transforms/Passes.td
@@ -165,16 +165,10 @@ def AMDAIEControlCodeLoopUnroll :
   let constructor = "mlir::iree_compiler::AMDAIE::createAMDAIEControlCodeLoopUnrollPass()";
 }
 
-def AMDAIEControlCodeToHalfDmaCpyNd :
-    Pass<"iree-amdaie-controlcode-to-half-dma-cpy-nd", ""> {
-  let summary = "Lower control code ops to HalfDmaCpyNd operations";
-  let constructor = "mlir::iree_compiler::AMDAIE::createAMDAIEControlCodeToHalfDmaCpyNdPass()";
-}
-
-def AMDAIEControlCodeToNpu :
-    Pass<"iree-amdaie-controlcode-to-npu", ""> {
+def AMDAIEControlCodeLowering :
+    Pass<"iree-amdaie-controlcode-lowering", ""> {
   let summary = "Lower control code ops to the most basic NPU write/sync/patch instructions";
-  let constructor = "mlir::iree_compiler::AMDAIE::createAMDAIEControlCodeToNpuPass()";
+  let constructor = "mlir::iree_compiler::AMDAIE::createAMDAIEControlCodeLoweringPass()";
 }
 
 
@@ -543,6 +537,12 @@ def AMDAIENormalizeLoopBounds :
   let constructor = "mlir::iree_compiler::AMDAIE::createAMDAIENormalizeLoopBoundsPass()";
 }
 
+def AMDAIENpuDmaToHalfDmaCpyNd :
+    Pass<"iree-amdaie-npu-dma-to-half-dma-cpy-nd", ""> {
+  let summary = "Lower DmaCpNd ops to HalfDmaCpyNd ops";
+  let constructor = "mlir::iree_compiler::AMDAIE::createAMDAIENpuDmaToHalfDmaCpyNdPass()";
+}
+
 def AMDAIEObjFifoBufferization :
     Pass<"iree-amdaie-objfifo-bufferization", ""> {
   let summary = "Bufferize logical objectFifos.";

diff --git a/compiler/plugins/target/AMD-AIE/iree-amd-aie/Transforms/test/CMakeLists.txt b/compiler/plugins/target/AMD-AIE/iree-amd-aie/Transforms/test/CMakeLists.txt
@@ -26,8 +26,7 @@ iree_lit_test_suite(
     "connection_to_flow.mlir"
     "controlcode_forall_to_for.mlir"
     "controlcode_loop_unrolling.mlir"
-    "controlcode_to_half_dma_cpy_nd.mlir"
-    "controlcode_to_npu.mlir"
+    "controlcode_lowering.mlir"
     "controlcode_to_transaction.mlir"
     "convert_core_forall_to_for.mlir"
     "create_aie_workgroup.mlir"
@@ -65,6 +64,7 @@ iree_lit_test_suite(
     "map_forall_to_cores.mlir"
     "none_access_to_temporary_buffer.mlir"
     "normalize_loop_bounds.mlir"
+    "npu_dma_to_half_dma_cpy_nd.mlir"
     "obj_fifo_bufferization.mlir"
     "pack_and_transpose_level1.mlir"
     "pack_and_transpose_level2.mlir"

diff --git a/...e/Transforms/test/controlcode_to_npu.mlir → ...Transforms/test/controlcode_lowering.mlir b/...e/Transforms/test/controlcode_to_npu.mlir → ...Transforms/test/controlcode_lowering.mlir
@@ -1,4 +1,4 @@
-// RUN: iree-opt --pass-pipeline="builtin.module(iree-amdaie-controlcode-to-npu)" --split-input-file --verify-diagnostics %s | FileCheck %s
+// RUN: iree-opt --pass-pipeline="builtin.module(iree-amdaie-controlcode-lowering)" --split-input-file --verify-diagnostics %s | FileCheck %s
 
 // expected-error @+1 {{op has no AMDAIEDevice in the target attribute configuration}}
 module {

diff --git a/.../test/controlcode_to_half_dma_cpy_nd.mlir → ...orms/test/npu_dma_to_half_dma_cpy_nd.mlir b/.../test/controlcode_to_half_dma_cpy_nd.mlir → ...orms/test/npu_dma_to_half_dma_cpy_nd.mlir
@@ -1,4 +1,4 @@
-// RUN: iree-opt --pass-pipeline="builtin.module(iree-amdaie-controlcode-to-half-dma-cpy-nd)" --split-input-file --verify-diagnostics %s | FileCheck %s
+// RUN: iree-opt --pass-pipeline="builtin.module(iree-amdaie-npu-dma-to-half-dma-cpy-nd)" --split-input-file --verify-diagnostics %s | FileCheck %s
 
 // CHECK-LABEL: @no_amdaie_device
 module {