Sync Sub-channel quantized type from llvm-project (pre-merge) to npu-plugin

mlir/include/mlir-c/Dialect/Quant.h

@@ -172,6 +172,47 @@ mlirUniformQuantizedPerAxisTypeGetQuantizedDimension(MlirType type);
 MLIR_CAPI_EXPORTED bool
 mlirUniformQuantizedPerAxisTypeIsFixedPoint(MlirType type);
 
+//===---------------------------------------------------------------------===//
+// UniformQuantizedSubChannelType
+//===---------------------------------------------------------------------===//
+
+/// Returns `true` if the given type is a UniformQuantizedSubChannel.
+MLIR_CAPI_EXPORTED bool
+mlirTypeIsAUniformQuantizedSubChannelType(MlirType type);
+
+/// Creates a UniformQuantizedSubChannelType with the given parameters.
+///
+/// The type is owned by the context. `scalesAttr` and `zeroPointsAttr` must be
+/// DenseElementsAttrs.  `quantizedDimensions` and `blockSizes`
+/// point to `blockSizeInfoLength` number of elements, describing respectively
+/// the quantization axis and corresponding block size.
+MLIR_CAPI_EXPORTED MlirType mlirUniformQuantizedSubChannelTypeGet(
+    unsigned flags, MlirType storageType, MlirType expressedType,
+    MlirAttribute scalesAttr, MlirAttribute zeroPointsAttr,
+    intptr_t blockSizeInfoLength, int32_t *quantizedDimensions,
+    int64_t *blockSizes, int64_t storageTypeMin, int64_t storageTypeMax);
+
+/// Returns the number of block sizes provided in type.
+MLIR_CAPI_EXPORTED intptr_t
+mlirUniformQuantizedSubChannelTypeGetNumBlockSizes(MlirType type);
+
+/// Returns the quantized dimension at the given position.
+MLIR_CAPI_EXPORTED int32_t
+mlirUniformQuantizedSubChannelTypeGetQuantizedDimension(MlirType type,
+                                                        intptr_t pos);
+
+/// Returns the block size at the given position.
+MLIR_CAPI_EXPORTED int64_t
+mlirUniformQuantizedSubChannelTypeGetBlockSize(MlirType type, intptr_t pos);
+
+/// Returns the scales of the quantized type.
+MLIR_CAPI_EXPORTED MlirAttribute
+mlirUniformQuantizedSubChannelTypeGetScales(MlirType type);
+
+/// Returns the zero-points of the quantized type.
+MLIR_CAPI_EXPORTED MlirAttribute
+mlirUniformQuantizedSubChannelTypeGetZeroPoints(MlirType type);
+
 //===---------------------------------------------------------------------===//
 // CalibratedQuantizedType
 //===---------------------------------------------------------------------===//

mlir/include/mlir/Dialect/Quant/QuantDialectBytecode.td

@@ -13,6 +13,7 @@
 #ifndef QUANT_BYTECODE
 #define QUANT_BYTECODE
 
+include "mlir/IR/BuiltinDialectBytecode.td"
 include "mlir/IR/BytecodeBase.td"
 
 def DoubleAPFloat:
@@ -81,6 +82,21 @@ def UniformQuantizedPerAxisType: DialectType<(type
   }];
 }
 
+def UniformQuantizedSubChannelType
+    : DialectType<(type VarInt:$flags, Type:$storageType, Type:$expressedType,
+          SignedVarInt:$storageTypeMin, SignedVarInt:$storageTypeMax,
+          Array<SignedVarIntList>:$quantizedDimensions,
+          Array<SignedVarIntList>:$blockSizes, DenseElementsAttr:$scales,
+          DenseElementsAttr:$zeroPoints)> {
+  // Note: builder order differs from bytecode.
+  let cBuilder = [{
+      get<$_resultType>(context, flags, storageType, expressedType, scales,
+        zeroPoints, llvm::to_vector(llvm::map_range(quantizedDimensions,
+        [](int64_t dim) { return static_cast<int32_t>(dim);})), blockSizes,
+        storageTypeMin, storageTypeMax)
+  }];
+}
+
 def QuantileQuantizedType: DialectType<(type
   VarInt:$flags,
   Type:$storageType,
@@ -118,16 +134,16 @@ def QuantileQuantizedPerAxisType: DialectType<(type
 /// compatibility with older bytecode.
 
 def QuantDialectTypes : DialectTypes<"Quant"> {
-  let elems = [
-    ReservedOrDead,
-    AnyQuantizedType,
-    AnyQuantizedTypeWithExpressedType,
-    CalibratedQuantizedType,
-    UniformQuantizedType,
-    UniformQuantizedPerAxisType,
-    QuantileQuantizedType,
-    QuantileQuantizedPerAxisType
+  let elems = [ReservedOrDead, 
+               AnyQuantizedType,
+               AnyQuantizedTypeWithExpressedType, 
+               CalibratedQuantizedType,
+               UniformQuantizedType, 
+               UniformQuantizedPerAxisType, 
+               UniformQuantizedSubChannelType,
+               QuantileQuantizedType,
+               QuantileQuantizedPerAxisType
   ];
 }
 
-#endif // QUANT_BYTECODE
+#endif // QUANT_BYTECODE