`zhigh.Add` (::onnx_mlir::zhigh::ZHighAddOp)

ZHigh Add operation

ZHigh operation to perform an Add. This operation does not support broadcasting.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand Description

X unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Y unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.AvgPool2D` (::onnx_mlir::zhigh::ZHighAvgPool2DOp)

ZHigh 2D average pooling operation

ZHigh operation to perform 2D average pooling.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`kernel_shape`	::mlir::ArrayAttr	64-bit integer array attribute
`strides`	::mlir::ArrayAttr	64-bit integer array attribute
`padding_type`	::mlir::StringAttr	string attribute

Operands:

Operand	Description
`input`	unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC

Results:

Result	Description
`output`	unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC

`zhigh.BatchNorm` (::onnx_mlir::zhigh::ZHighBatchNormOp)

ZHigh batchnorm operation

ZHigh operation to perform batchnorm.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand	Description
`input`	unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC
`a`	unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D
`b`	unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D

Results:

Result	Description
`output`	unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC

`zhigh.Conv2D` (::onnx_mlir::zhigh::ZHighConv2DOp)

ZHigh 2D convolution operation

ZHigh operation to perform 2D convolution.

input: [num_batches, height_in, width_in, channels_in]
input_kernel: [kernel_height, kernel_width, channels_in, channels_out]
input_bias: [channels_out]
kernel_shape: 1D array of kernel height and width
strides: 1D array of stride height and width
padding_type: SAME_PADDING or VALID_PADDING
act_func: ACT_NONE or ACT_RELU
output: [num_batches, height_out, width_out, channels_out]

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`kernel_shape`	::mlir::ArrayAttr	64-bit integer array attribute
`strides`	::mlir::ArrayAttr	64-bit integer array attribute
`padding_type`	::mlir::StringAttr	string attribute
`act_func`	::mlir::StringAttr	string attribute

Operands:

Operand	Description
`input`	unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC
`input_kernel`	unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK
`input_bias`	unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or none type

Results:

Result	Description
`output`	unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC

`zhigh.DLF16ToF32` (::onnx_mlir::zhigh::ZHighDLF16ToF32Op)

ZHigh DLF16ToF32 operation

ZHigh operation to convert a tensor of dlfloat16 to a tensor of f32.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand	Description
`In`	tensor of 16-bit float values

Results:

Result	Description
`Out`	tensor of 32-bit float values

`zhigh.Div` (::onnx_mlir::zhigh::ZHighDivOp)

ZHigh Div operation

ZHigh operation to perform a Div. This operation does not support broadcasting.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand Description

X unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Y unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.Exp` (::onnx_mlir::zhigh::ZHighExpOp)

ZHigh Exp operation

ZHigh operation to perform a Exp.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand Description

X unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.F32ToDLF16` (::onnx_mlir::zhigh::ZHighF32ToDLF16Op)

ZHigh F32ToDLF16 operation

ZHigh operation to convert a tensor of f32 to a tensor of dlfloat16.

Optional saturation indicates whether the CPU tensor is saturated before stickification or not. If it is saturated, the dlfloat16 range would be used. Saturation if off if saturation == 0 or it is not given. Otherwise, it is on.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`saturation`	::mlir::IntegerAttr	64-bit signed integer attribute

Operands:

Operand	Description
`In`	tensor of 32-bit float values

Results:

Result	Description
`Out`	tensor of 16-bit float values

`zhigh.FixGRUY` (::onnx_mlir::zhigh::ZHighFixGRUYOp)

Fix Y result of GRU for sequence_lens

Fix Y result of GRU by padding value after sequence_lens.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand	Description
`Y`	tensor of 16-bit float values or tensor of 32-bit float values or tensor of 64-bit float values or none type
`sequence_lens`	tensor of 32-bit signless integer values or none type
`initial_h`	tensor of 16-bit float values or tensor of 32-bit float values or tensor of 64-bit float values or none type

Results:

Result	Description
`output`	tensor of 16-bit float values or tensor of 32-bit float values or tensor of 64-bit float values or none type

`zhigh.FixGRUYh` (::onnx_mlir::zhigh::ZHighFixGRUYhOp)

Fix Yh result of GRU for sequence_lens

Fix Yh result of GRU by picking the value in Y according to sequence_lens.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand	Description
`Y`	tensor of 16-bit float values or tensor of 32-bit float values or tensor of 64-bit float values or none type
`sequence_lens`	tensor of 32-bit signless integer values or none type

Results:

Result	Description
`output`	tensor of 16-bit float values or tensor of 32-bit float values or tensor of 64-bit float values or none type

`zhigh.GRU` (::onnx_mlir::zhigh::ZHighGRUOp)

ZHigh GRU operation

zHigh operation to perform a GRU.
Shape for input is [S, B, I]. Shape for h0 is [D, B, H].
Shape for input_weights is [D, I, 3*H].
Shape for hidden_weights is [D, H, 3*H].
Shape for input_bias and hidden_bias is [D, 3*H].
Shape for hn_output is [S, D, B, H] if return all timesteps and [1, D, B, H] if return the final step only.
S is timesteps, D is the number of directions (1 for unidirectional and
2 for bidirectional), B is batch size, I is input size, and
H is hidden size.
direction accepts "forward", "reverse", or "bidirectional
return_all_steps: -1 returns all timesteps, 0: returns only the last timestep."

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`hidden_size`	::mlir::IntegerAttr	64-bit signed integer attribute
`direction`	::mlir::StringAttr	string attribute
`return_all_steps`	::mlir::IntegerAttr	64-bit signed integer attribute

Operands:

Operand	Description
`input`	unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS
`h0`	unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or none type
`input_weights`	unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH
`input_bias`	unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or none type
`hidden_weights`	unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH
`hidden_bias`	unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or none type

Results:

Result	Description
`hn_output`	unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS

`zhigh.Gelu` (::onnx_mlir::zhigh::ZHighGeluOp)

ZHigh Gelu operation

"ZHigh operation to perform a Gelu."

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`approximate`	::mlir::StringAttr	string attribute

Operands:

Operand Description

X unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.InvSqrt` (::onnx_mlir::zhigh::ZHighInvSqrtOp)

ZHigh InvSqrt operation

ZHigh operation to perform a InvSqrt.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand Description

X unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.LSTM` (::onnx_mlir::zhigh::ZHighLSTMOp)

ZHigh LSTM operation

zHigh operation to perform a LSTM.

Shape for input is [S, B, I]. Shape for h0 and c0 is [D, B, H].
Shape for input_weights is [D, I, 4*H].
Shape for hidden_weights is [D, H, 4*H].
Shape for input_bias and hidden_bias is [D, 4*H].
Shape for hn_output is [S, D, B, H] if return all timesteps and [1, D, B, H] if return the final step only.
Shape for cf_output is [1, D, B, H].
S is timesteps, D is the number of directions (1 for unidirectional and
2 for bidirectional), B is batch size, I is input size, and
H is hidden size.
direction accepts "forward", "reverse", or "bidirectional
return_all_steps: -1 returns all timesteps, 0: returns only the last timestep.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`hidden_size`	::mlir::IntegerAttr	64-bit signed integer attribute
`direction`	::mlir::StringAttr	string attribute
`return_all_steps`	::mlir::IntegerAttr	64-bit signed integer attribute

Operands:

Operand	Description
`input`	unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS
`h0`	unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or none type
`c0`	unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or none type
`input_weights`	unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO
`input_bias`	unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or none type
`hidden_weights`	unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO
`hidden_bias`	unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or none type

Results:

Result	Description
`hn_output`	unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS
`cf_output`	unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS

`zhigh.LeakyRelu` (::onnx_mlir::zhigh::ZHighLeakyReluOp)

ZHigh LeakyRelu operation

"ZHigh operation to perform a LeakyRelu."

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`alpha`	::mlir::FloatAttr	32-bit float attribute

Operands:

Operand Description

X unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.Log` (::onnx_mlir::zhigh::ZHighLogOp)

ZHigh Log operation

ZHigh operation to perform a Log.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand Description

X unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.MatMul` (::onnx_mlir::zhigh::ZHighMatMulOp)

ZHigh MatMul operation

ZHigh operation to perform a MatMul.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`transposeA`	::mlir::IntegerAttr	64-bit signed integer attribute
`transposeB`	::mlir::IntegerAttr	64-bit signed integer attribute

Operands:

Operand	Description
`X`	unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS
`Y`	unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS
`B`	unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or none type

Results:

Result	Description
`Out`	unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS

`zhigh.Max` (::onnx_mlir::zhigh::ZHighMaxOp)

ZHigh Max operation

ZHigh operation to perform a Max. This operation does not support broadcasting.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand Description

X unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Y unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.MaxPool2D` (::onnx_mlir::zhigh::ZHighMaxPool2DOp)

ZHigh 2D max pooling operation

ZHigh operation to perform 2D max pooling.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`kernel_shape`	::mlir::ArrayAttr	64-bit integer array attribute
`strides`	::mlir::ArrayAttr	64-bit integer array attribute
`padding_type`	::mlir::StringAttr	string attribute

Operands:

Operand	Description
`input`	unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC

Results:

Result	Description
`output`	unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC

`zhigh.MeanReduce2d` (::onnx_mlir::zhigh::ZHighMeanReduce2DOp)

ZHigh 2D mean reduce operation

ZHigh operation to perform 2D mean reduce. Given an input 4D tensor, returns a downsampled tensor reducing the middle 2nd and 3rd dimensions to a size of 1 based on the mean of the original values. Input and Output tensors should be in the 3D layout.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand	Description
`input`	unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC

Results:

Result	Description
`output`	unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC

`zhigh.Min` (::onnx_mlir::zhigh::ZHighMinOp)

ZHigh Min operation

ZHigh operation to perform a Min. This operation does not support broadcasting.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand Description

X unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Y unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.Mul` (::onnx_mlir::zhigh::ZHighMulOp)

ZHigh Mul operation

ZHigh operation to perform a Mul. This operation does not support broadcasting.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand Description

X unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Y unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.QuantizedMatMul` (::onnx_mlir::zhigh::ZHighQuantizedMatMulOp)

ZHigh QuantizedMatMul operation

ZHigh operation to perform a quantized MatMul.

OutRecScaleIn and OutOffsetIn are recscale and offset for the output. If OutRecScaleIn is given, it will be passed to OutRecScale. If it is None, OutRescScale is set to 1.0. If OutOffsetIn is given, it will be passed to OutOffset. If it is None, OutOffset is set to 0.0.

PreComputedBias: -1 bias is re-computed, 0: bias is not pre-computed.

DequantizeOutput indicates if the output is dequantized to real dfloat16 or not. If not, the output is int8 but stored in dlfloat (int8-as-dlfloat).

DequantizeOutput: -1 output is dequantized, 0: output is not dequantized.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`PreComputedBias`	::mlir::IntegerAttr	64-bit signed integer attribute
`DisableClipping`	::mlir::IntegerAttr	64-bit signed integer attribute
`DequantizeOutput`	::mlir::IntegerAttr	64-bit signed integer attribute

Operands:

Operand	Description
`X`	unranked tensor of 8-bit signless integer or 16-bit float values or 2D tensor of 8-bit signless integer or 16-bit float values with layout _2D or unranked tensor of 8-bit signless integer or 16-bit float values or 3D tensor of 8-bit signless integer or 16-bit float values with layout _3DS
`XRecScale`	0D tensor of 32-bit float values
`XOffset`	0D tensor of 32-bit float values
`Y`	unranked tensor of 8-bit signless integer or 16-bit float values or 2D tensor of 8-bit signless integer or 16-bit float values with layout _2D or unranked tensor of 8-bit signless integer or 16-bit float values or 3D tensor of 8-bit signless integer or 16-bit float values with layout _3DS
`YRecScale`	0D tensor of 32-bit float values
`YOffset`	0D tensor of 32-bit float values
`B`	unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 8-bit signless integer or 16-bit float values or 1D tensor of 8-bit signless integer or 16-bit float values with layout _1D or unranked tensor of 8-bit signless integer or 16-bit float values or 2D tensor of 8-bit signless integer or 16-bit float values with layout _2DS or none type
`BRecScale`	0D tensor of 32-bit float values or none type
`BOffset`	0D tensor of 32-bit float values or none type
`OutRecScaleIn`	0D tensor of 32-bit float values or none type
`OutOffsetIn`	0D tensor of 32-bit float values or none type

Results:

Result	Description
`Out`	unranked tensor of 8-bit signless integer or 16-bit float values or 2D tensor of 8-bit signless integer or 16-bit float values with layout _2D or unranked tensor of 8-bit signless integer or 16-bit float values or 3D tensor of 8-bit signless integer or 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS
`OutRecScale`	0D tensor of 32-bit float values
`OutOffset`	0D tensor of 32-bit float values

`zhigh.QuantizedStick` (::onnx_mlir::zhigh::ZHighQuantizedStickOp)

ZHigh QuantizedStick operation

ZHigh operation to perform a quantized Stick. Type is one of values: dlfloat16, int8, and weights. sym_mode indicates whether to use symmetric quantization or not to compute the output rescale and offset. sym_mode is only effective when the input rescale and offset are None. By default, asymmetric quantization is used.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`layout`	::mlir::StringAttr	string attribute
`quantized_type`	::mlir::StringAttr	string attribute
`sym_mode`	::mlir::IntegerAttr	64-bit signless integer attribute

Operands:

Operand	Description
`In`	tensor of 32-bit float values or tensor of 8-bit signless integer values or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS
`InRecScale`	0D tensor of 32-bit float values or none type
`InOffset`	0D tensor of 32-bit float values or none type

Results:

Result	Description
`Out`	unranked tensor of 8-bit signless integer or 16-bit float values or 1D tensor of 8-bit signless integer or 16-bit float values with layout _1D or unranked tensor of 8-bit signless integer or 16-bit float values or 2D tensor of 8-bit signless integer or 16-bit float values with layout _2D or unranked tensor of 8-bit signless integer or 16-bit float values or 3D tensor of 8-bit signless integer or 16-bit float values with layout _3D or unranked tensor of 8-bit signless integer or 16-bit float values or 2D tensor of 8-bit signless integer or 16-bit float values with layout _2DS or unranked tensor of 8-bit signless integer or 16-bit float values or 3D tensor of 8-bit signless integer or 16-bit float values with layout _3DS or none type
`RecScale`	0D tensor of 32-bit float values
`Offset`	0D tensor of 32-bit float values

`zhigh.ReduceMax` (::onnx_mlir::zhigh::ZHighReduceMaxOp)

ZHigh ReduceMax operation

ZHigh operation to perform a ReduceMax. op_type: REDUCE_OP_MAXIMUM or REDUCE_OP_MINIMUM.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`op_type`	::mlir::StringAttr	string attribute

Operands:

Operand Description

data unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

output unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.ReduceMin` (::onnx_mlir::zhigh::ZHighReduceMinOp)

ZHigh ReduceMin operation

ZHigh operation to perform a ReduceMin. op_type: REDUCE_OP_MAXIMUM or REDUCE_OP_MINIMUM.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`op_type`	::mlir::StringAttr	string attribute

Operands:

Operand Description

data unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

output unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.Relu` (::onnx_mlir::zhigh::ZHighReluOp)

ZHigh Relu operation

"ZHigh operation to perform a Relu."

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand Description

X unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.Sigmoid` (::onnx_mlir::zhigh::ZHighSigmoidOp)

ZHigh Sigmoid operation

ZHigh operation to perform a Sigmoid.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand Description

X unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.Softmax` (::onnx_mlir::zhigh::ZHighSoftmaxOp)

ZHigh Softmax operation

ZHigh operation to perform a Softmax. act_func: ACT_NONE or ACT_LOG.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`act_func`	::mlir::StringAttr	string attribute

Operands:

Operand	Description
`X`	unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS

Results:

Result	Description
`Out`	unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS

`zhigh.Sqrt` (::onnx_mlir::zhigh::ZHighSqrtOp)

ZHigh Sqrt operation

ZHigh operation to perform a Sqrt.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand Description

X unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.StickForGRU` (::onnx_mlir::zhigh::ZHighStickForGRUOp)

ZHigh stick operation for GRU

ZHigh operation to perform a stick for GRU. Variadic: list of pointers for input data to be transformed:

GRU concatenated: 3 data pointers, one for each input gate in (Z)update, Reset, Hidden, (ZRH) gate order

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand	Description
`z_gate`	tensor of 32-bit float values
`r_gate`	tensor of 32-bit float values
`h_gate`	tensor of 32-bit float values

Results:

Result	Description
`out`	unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH

`zhigh.StickForLSTM` (::onnx_mlir::zhigh::ZHighStickForLSTMOp)

ZHigh stick operation for LSTM

ZHigh operation to perform a stick for LSTM. Variadic: list of pointers for input data to be transformed:

LSTM concatenated: 4 data pointers, one for each input gate in Forget, Input, Cell, Output (FICO) order,

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand	Description
`f_gate`	tensor of 32-bit float values
`i_gate`	tensor of 32-bit float values
`c_gate`	tensor of 32-bit float values
`o_gate`	tensor of 32-bit float values

Results:

Result	Description
`out`	unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO

`zhigh.Stick` (::onnx_mlir::zhigh::ZHighStickOp)

ZHigh Stick operation

ZHigh operation to perform a Stick."

If layout=NHWC, input must be in NCHW and output will be in NHWC.

Optional saturation indicates whether the CPU tensor is saturated before stickification or not. If it is saturated, the dlfloat16 range would be used. Saturation if off if saturation == 0 or it is not given. Otherwise, it is on.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`layout`	::mlir::StringAttr	string attribute
`saturation`	::mlir::IntegerAttr	64-bit signed integer attribute

Operands:

Operand	Description
`In`	tensor of 32-bit float values or none type

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or none type

`zhigh.StickifiedConstantOfShape` (::onnx_mlir::zhigh::ZHighStickifiedConstantOfShapeOp)

ZHigh Stickified Constant operation for a dynamic shape

This operator produces a constant tensor to store stickified data. The stickified data is defined by a f32 scalar value, a dynamic shape and a layout. Stickified data is 4K-aligned.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`value`	::mlir::FloatAttr	32-bit float attribute
`layout`	::mlir::StringAttr	string attribute

Operands:

Operand	Description
`shape`	tensor of 64-bit signless integer values

Results:

Result Description

output unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.StickifiedConstant` (::onnx_mlir::zhigh::ZHighStickifiedConstantOp)

ZHigh Stickified Constant operation

This operator produces a constant tensor to store stickified data. Stickified data is opaque and must be 4K-aligned. One who produces the stickified data must make sure its size in bytes consistent with the output tensor's size.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface)

Effects: MemoryEffects::Effect{}

Attributes:

Attribute	MLIR Type	Description
`value`	::mlir::Attribute	any attribute
`alignment`	::mlir::IntegerAttr	64-bit signless integer attribute

Results:

Result Description

output unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH or unranked tensor of 8-bit signless integer or 16-bit float values or 1D tensor of 8-bit signless integer or 16-bit float values with layout _1D or unranked tensor of 8-bit signless integer or 16-bit float values or 2D tensor of 8-bit signless integer or 16-bit float values with layout _2D or unranked tensor of 8-bit signless integer or 16-bit float values or 3D tensor of 8-bit signless integer or 16-bit float values with layout _3D or unranked tensor of 8-bit signless integer or 16-bit float values or 2D tensor of 8-bit signless integer or 16-bit float values with layout _2DS or unranked tensor of 8-bit signless integer or 16-bit float values or 3D tensor of 8-bit signless integer or 16-bit float values with layout _3DS

`zhigh.Sub` (::onnx_mlir::zhigh::ZHighSubOp)

ZHigh Sub operation

ZHigh operation to perform a Sub. This operation does not support broadcasting.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand Description

X unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Y unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.Tanh` (::onnx_mlir::zhigh::ZHighTanhOp)

ZHigh Tanh operation

ZHigh operation to perform a Tanh.

Traits: AlwaysSpeculatableImplTrait, SameOperandsAndResultLayout

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand Description

X unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

Results:

Result Description

Out unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout FICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout ZRH or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BFICO or unranked tensor of 16-bit float values or 2D/3D tensor of 16-bit float values with layout BZRH

`zhigh.Unstick` (::onnx_mlir::zhigh::ZHighUnstickOp)

ZHigh Unstick operation

ZHigh operation to perform a Unstick.

Traits: AlwaysSpeculatableImplTrait

Interfaces: ConditionallySpeculatable, NoMemoryEffect (MemoryEffectOpInterface), ShapeHelperOpInterface, ShapeInferenceOpInterface

Effects: MemoryEffects::Effect{}

Operands:

Operand Description

In unranked tensor of 16-bit float values or 1D tensor of 16-bit float values with layout _1D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2D or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3D or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4D or unranked tensor of 16-bit float values or 2D tensor of 16-bit float values with layout _2DS or unranked tensor of 16-bit float values or 3D tensor of 16-bit float values with layout _3DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout _4DS or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NHWC or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout NCHW or unranked tensor of 16-bit float values or 4D tensor of 16-bit float values with layout HWCK

Results:

Result	Description
`Out`	tensor of 32-bit float values

Files

zhigh.md

Latest commit

History

zhigh.md

File metadata and controls

zhigh.Add (::onnx_mlir::zhigh::ZHighAddOp)

Operands:

Results:

zhigh.AvgPool2D (::onnx_mlir::zhigh::ZHighAvgPool2DOp)

Attributes:

Operands:

Results:

zhigh.BatchNorm (::onnx_mlir::zhigh::ZHighBatchNormOp)

Operands:

Results:

zhigh.Conv2D (::onnx_mlir::zhigh::ZHighConv2DOp)

Attributes:

Operands:

Results:

zhigh.DLF16ToF32 (::onnx_mlir::zhigh::ZHighDLF16ToF32Op)

Operands:

Results:

zhigh.Div (::onnx_mlir::zhigh::ZHighDivOp)

Operands:

Results:

zhigh.Exp (::onnx_mlir::zhigh::ZHighExpOp)

Operands:

Results:

zhigh.F32ToDLF16 (::onnx_mlir::zhigh::ZHighF32ToDLF16Op)

Attributes:

Operands:

Results:

zhigh.FixGRUY (::onnx_mlir::zhigh::ZHighFixGRUYOp)

Operands:

Results:

zhigh.FixGRUYh (::onnx_mlir::zhigh::ZHighFixGRUYhOp)

Operands:

Results:

zhigh.GRU (::onnx_mlir::zhigh::ZHighGRUOp)

Attributes:

Operands:

Results:

zhigh.Gelu (::onnx_mlir::zhigh::ZHighGeluOp)

Attributes:

Operands:

Results:

zhigh.InvSqrt (::onnx_mlir::zhigh::ZHighInvSqrtOp)

Operands:

Results:

zhigh.LSTM (::onnx_mlir::zhigh::ZHighLSTMOp)

Attributes:

Operands:

Results:

zhigh.LeakyRelu (::onnx_mlir::zhigh::ZHighLeakyReluOp)

Attributes:

Operands:

Results:

zhigh.Log (::onnx_mlir::zhigh::ZHighLogOp)

Operands:

Results:

zhigh.MatMul (::onnx_mlir::zhigh::ZHighMatMulOp)

Attributes:

Operands:

Results:

zhigh.Max (::onnx_mlir::zhigh::ZHighMaxOp)

Operands:

Results:

zhigh.MaxPool2D (::onnx_mlir::zhigh::ZHighMaxPool2DOp)

Attributes:

Operands:

Results:

zhigh.MeanReduce2d (::onnx_mlir::zhigh::ZHighMeanReduce2DOp)

Operands:

Results:

zhigh.Min (::onnx_mlir::zhigh::ZHighMinOp)

Operands:

Results:

zhigh.Mul (::onnx_mlir::zhigh::ZHighMulOp)

Operands:

`zhigh.Add` (::onnx_mlir::zhigh::ZHighAddOp)

`zhigh.AvgPool2D` (::onnx_mlir::zhigh::ZHighAvgPool2DOp)

`zhigh.BatchNorm` (::onnx_mlir::zhigh::ZHighBatchNormOp)

`zhigh.Conv2D` (::onnx_mlir::zhigh::ZHighConv2DOp)

`zhigh.DLF16ToF32` (::onnx_mlir::zhigh::ZHighDLF16ToF32Op)

`zhigh.Div` (::onnx_mlir::zhigh::ZHighDivOp)

`zhigh.Exp` (::onnx_mlir::zhigh::ZHighExpOp)

`zhigh.F32ToDLF16` (::onnx_mlir::zhigh::ZHighF32ToDLF16Op)

`zhigh.FixGRUY` (::onnx_mlir::zhigh::ZHighFixGRUYOp)

`zhigh.FixGRUYh` (::onnx_mlir::zhigh::ZHighFixGRUYhOp)

`zhigh.GRU` (::onnx_mlir::zhigh::ZHighGRUOp)

`zhigh.Gelu` (::onnx_mlir::zhigh::ZHighGeluOp)

`zhigh.InvSqrt` (::onnx_mlir::zhigh::ZHighInvSqrtOp)

`zhigh.LSTM` (::onnx_mlir::zhigh::ZHighLSTMOp)

`zhigh.LeakyRelu` (::onnx_mlir::zhigh::ZHighLeakyReluOp)

`zhigh.Log` (::onnx_mlir::zhigh::ZHighLogOp)

`zhigh.MatMul` (::onnx_mlir::zhigh::ZHighMatMulOp)

`zhigh.Max` (::onnx_mlir::zhigh::ZHighMaxOp)

`zhigh.MaxPool2D` (::onnx_mlir::zhigh::ZHighMaxPool2DOp)

`zhigh.MeanReduce2d` (::onnx_mlir::zhigh::ZHighMeanReduce2DOp)

`zhigh.Min` (::onnx_mlir::zhigh::ZHighMinOp)

`zhigh.Mul` (::onnx_mlir::zhigh::ZHighMulOp)

`zhigh.QuantizedMatMul` (::onnx_mlir::zhigh::ZHighQuantizedMatMulOp)

`zhigh.QuantizedStick` (::onnx_mlir::zhigh::ZHighQuantizedStickOp)

`zhigh.ReduceMax` (::onnx_mlir::zhigh::ZHighReduceMaxOp)

`zhigh.ReduceMin` (::onnx_mlir::zhigh::ZHighReduceMinOp)

`zhigh.Relu` (::onnx_mlir::zhigh::ZHighReluOp)

`zhigh.Sigmoid` (::onnx_mlir::zhigh::ZHighSigmoidOp)

`zhigh.Softmax` (::onnx_mlir::zhigh::ZHighSoftmaxOp)

`zhigh.Sqrt` (::onnx_mlir::zhigh::ZHighSqrtOp)

`zhigh.StickForGRU` (::onnx_mlir::zhigh::ZHighStickForGRUOp)

`zhigh.StickForLSTM` (::onnx_mlir::zhigh::ZHighStickForLSTMOp)

`zhigh.Stick` (::onnx_mlir::zhigh::ZHighStickOp)

`zhigh.StickifiedConstantOfShape` (::onnx_mlir::zhigh::ZHighStickifiedConstantOfShapeOp)

`zhigh.StickifiedConstant` (::onnx_mlir::zhigh::ZHighStickifiedConstantOp)

`zhigh.Sub` (::onnx_mlir::zhigh::ZHighSubOp)

`zhigh.Tanh` (::onnx_mlir::zhigh::ZHighTanhOp)

`zhigh.Unstick` (::onnx_mlir::zhigh::ZHighUnstickOp)