[QNN][Legalize] Specialize for Platforms without any fast Int8 arithm…

…etic units.

python/tvm/relay/qnn/op/legalizations.py

@@ -22,10 +22,43 @@
 from tvm import relay
 from .. import op as reg
 
+#################################################
+# Register the functions for different operators.
+#################################################
+
 # Registering QNN Conv2D legalization function.
 @reg.register_qnn_legalize("qnn.conv2d")
 def legalize_qnn_conv2d(attrs, inputs, types):
-    """Legalizes QNN conv2d op.
+    return qnn_conv2d_legalize(attrs, inputs, types)
+
+# Registering QNN dense legalization function.
+@reg.register_qnn_legalize("qnn.dense")
+def legalize_qnn_dense(attrs, inputs, types):
+    return qnn_dense_legalize(attrs, inputs, types)
+
+# Default to None. If overridden by target, this will not be run.
+# Generic QNN Conv2D legalization function.
+@tvm.target.generic_func
+def qnn_conv2d_legalize(attrs, inputs, types):
+    """Default legalization is None."""
+    return None
+
+# Generic QNN Conv2D legalization function.
+@tvm.target.generic_func
+def qnn_dense_legalize(attrs, inputs, types):
+    """Default legalization is None."""
+    return None
+
+###################
+# Helper functions.
+###################
+
+# Helper function for lowering in the abscence of fast Int8 arithmetic units.
+def helper_no_fast_int8_hw_legalization(attrs, inputs, types, relay_op):
+    """ Converts QNN operators into a sequence of Relay operators that are friendly to HW that do
+    not have fast Int8 arithmetic. For example, for ARM, LLVM utilizes the assembly instructions
+    much more efficiently if the convolution or dense operator input datatypes are int16 instead of
+    int8. More details are present at https://github.com/apache/incubator-tvm/pull/4277.
 
     Parameters
     ----------
@@ -41,19 +74,27 @@ def legalize_qnn_conv2d(attrs, inputs, types):
     result : tvm.relay.Expr
         The legalized expr
     """
-    return qnn_conv2d_legalize(attrs, inputs, types)
 
-# Generic QNN Conv2D legalization function.
-@tvm.target.generic_func
-def qnn_conv2d_legalize(attrs, inputs, types):
-    """Default legalization is None."""
-    return None
+    # Collect the input exprs.
+    data, kernel = inputs
 
-# Intel x86 QNN Conv2D legalization function.
-@qnn_conv2d_legalize.register('cpu')
-def _qnn_conv2d_legalize(attrs, inputs, types):
-    """Legalizes QNN conv2d op. VNNI supports u8 x i8 fast conv/MM. If the dtypes are already good,
-    we dont transform. Else, we shift the tensor values and zero points to change the dtype.
+    input_zp = attrs['input_zero_point']
+    kernel_zp = attrs['kernel_zero_point']
+
+    shift_data = relay.subtract(relay.cast(data, dtype='int16'),
+                                relay.const(input_zp, 'int16'))
+    shift_kernel = relay.subtract(relay.cast(kernel, dtype='int16'),
+                                  relay.const(kernel_zp, 'int16'))
+    new_attrs = {k : attrs[k] for k in attrs.keys()}
+    del new_attrs['kernel_zero_point']
+    del new_attrs['input_zero_point']
+    return relay_op(shift_data, shift_kernel, **new_attrs)
+
+# Helper function to change dtypes to uint8 x int8. Intel VNNI instructions prefer this setting.
+def helper_change_dtypes_to_uint8_int8(attrs, inputs, types, relay_op):
+    """Legalizes QNN conv2d/dense op for Intel HW. VNNI supports u8 x i8 fast conv/MM. If the dtypes
+    are already good, we dont transform. Else, we shift the tensor values and zero points to change
+    the dtype.
 
     Converting from int8 to uint8 can be done in following manner.
 
@@ -95,26 +136,13 @@ def _shift(data, out_dtype):
         data_modified = relay.cast(data_modified, out_dtype)
         return data_modified
 
-    def _is_int8_hw_support(target):
-        """
-        Checks to ensure that we can use Intel DLBoost instructions - Check if the target is skylake
-        and above.
-        """
-        supported_arches = {'-mcpu=skylake-avx512', '-mcpu=cascadelake'}
-        return supported_arches.intersection(set(target.options))
-
     # Collect the dtypes.
     data_dtype = types[0].dtype
     kernel_dtype = types[1].dtype
 
     # Collect the input exprs.
     data, kernel = inputs
 
-    # The VNNI transformations are applicable only Skylake and above.g
-    target = tvm.target.current_target(allow_none=False)
-    if not _is_int8_hw_support(target):
-        return None
-
     # VNNI supports u8 x i8 fast conv/MM. Don't do anything if it is already satisfied.
     if data_dtype == 'uint8' and kernel_dtype == 'int8':
         return None
@@ -137,4 +165,120 @@ def _is_int8_hw_support(target):
     new_attrs = {k : attrs[k] for k in attrs.keys()}
     new_attrs['input_zero_point'] = input_zp
     new_attrs['kernel_zero_point'] = kernel_zp
-    return relay.qnn.op.conv2d(data, kernel, **new_attrs)
+    return relay_op(data, kernel, **new_attrs)
+
+# Helper function to change dtypes to be same. ARM dotprod instructions prefer this setting.
+def helper_change_dtypes_to_be_same(attrs, inputs, types, relay_op):
+    """ Sometimes MxNet + MLDNN can lead to uint8 x int8 datatypes for the conv inputs. However,
+    many devices like ARM prefer the datatypes to be same for the HW units. This helper transforms
+    conv2d/dense such that both the dtypes are same.
+
+    Parameters
+    ----------
+    attrs : tvm.attrs.Attrs
+        Attributes of current convolution
+    inputs : list of tvm.relay.Expr
+        The args of the Relay expr to be legalized
+    types : list of types
+        List of input and output types
+
+    Returns
+    -------
+    result : tvm.relay.Expr
+        The legalized expr
+    """
+
+    def _shift(data, out_dtype):
+        """Shifts (add/subtracts) the qnn tensor with +/-128)"""
+        if out_dtype == 'uint8':
+            shift = 128
+        elif out_dtype == 'int8':
+            shift = -128
+        else:
+            raise ValueError("Unsupport out dtype.")
+        data_modified = relay.cast(data, 'int32')
+        data_modified = relay.add(data_modified, relay.const(shift, 'int32'))
+        data_modified = relay.cast(data_modified, out_dtype)
+        return data_modified
+
+    # Collect the dtypes.
+    data_dtype = types[0].dtype
+    kernel_dtype = types[1].dtype
+
+    # Collect the input exprs.
+    data, kernel = inputs
+
+    # VNNI supports u8 x i8 fast conv/MM. Don't do anything if it is already satisfied.
+    if data_dtype == kernel_dtype:
+        return None
+
+    assert 'int8' in data_dtype and 'int8' in kernel_dtype, \
+            "Qnn Conv2D only accepts uint8 or int8 inputs"
+
+    # Shift input if necessary.
+    input_zp = attrs['input_zero_point']
+    data = _shift(data, kernel_dtype)
+    if data_dtype == 'int8':
+        input_zp = input_zp + 128
+    elif data_dtype == 'uint8':
+        input_zp = input_zp - 128
+    else:
+        raise RuntimeError("Qnn Conv2D only accepts uint8 or int8 inputs")
+
+    # Call qnn.conv2d with modified inputs and zero points.
+    new_attrs = {k : attrs[k] for k in attrs.keys()}
+    new_attrs['input_zero_point'] = input_zp
+    return relay_op(data, kernel, **new_attrs)
+
+def is_fast_int8_hw_present():
+    """
+    Checks to ensure that we can use Intel DLBoost instructions - Check if the target is skylake
+    and above.
+    """
+    target = tvm.target.current_target(allow_none=False)
+    intel_supported_arches = {'-mcpu=skylake-avx512', '-mcpu=cascadelake'}
+    is_present_intel = intel_supported_arches.intersection(set(target.options))
+
+    arm_supported_attr = '+v8.2a,+dotprod'
+    is_present_arm = False
+    for opt in target.options:
+        if arm_supported_attr in opt:
+            is_present_arm = True
+
+    return is_present_intel or is_present_arm
+
+########################
+# ARM CPU legalizations.
+########################
+
+@qnn_conv2d_legalize.register('arm_cpu')
+def _qnn_conv2d_legalize_arm_cpu(attrs, inputs, types):
+    # ARM likes the dtypes to be same.
+    if is_fast_int8_hw_present():
+        return helper_change_dtypes_to_be_same(attrs, inputs, types, relay.qnn.op.conv2d)
+    return helper_no_fast_int8_hw_legalization(attrs, inputs, types, relay.nn.conv2d)
+
+@qnn_dense_legalize.register('arm_cpu')
+def _qnn_dense_legalize_arm_cpu(attrs, inputs, types):
+    # ARM likes the dtypes to be same.
+    if is_fast_int8_hw_present():
+        return helper_change_dtypes_to_be_same(attrs, inputs, types, relay.qnn.op.dense)
+    return helper_no_fast_int8_hw_legalization(attrs, inputs, types, relay.nn.dense)
+
+##########################
+# Intel CPU legalizations.
+##########################
+
+@qnn_conv2d_legalize.register('cpu')
+def _qnn_conv2d_legalize_intel_cpu(attrs, inputs, types):
+    # The VNNI transformations prefer uint8 x int8 datatypes.
+    if is_fast_int8_hw_present():
+        return helper_change_dtypes_to_uint8_int8(attrs, inputs, types, relay.qnn.op.conv2d)
+    return helper_no_fast_int8_hw_legalization(attrs, inputs, types, relay.nn.conv2d)
+
+@qnn_dense_legalize.register('cpu')
+def _qnn_dense_legalize_intel_cpy(attrs, inputs, types):
+    # The VNNI transformations prefer uint8 x int8 datatypes.
+    if is_fast_int8_hw_present():
+        return helper_change_dtypes_to_uint8_int8(attrs, inputs, types, relay.qnn.op.dense)
+    return helper_no_fast_int8_hw_legalization(attrs, inputs, types, relay.nn.dense)