[TOPI] Support grouped conv1d

Generalize the conv2d compute statement to a generic convNd that
supports any layout and groups. Replace some existing conv2d and conv1d
compute statements with this generic compute. Also add a topi
group_conv1d compute that uses the generic convNd compute. Existing
schedules for conv1d work with group_conv1d, so they are reused.

Author: Tristan Konolige

python/tvm/autotvm/task/topi_integration.py

@@ -233,7 +233,10 @@ def wrapper(outs, *args, **kwargs):
             """wrapper function for topi schedule"""
             workload = get_workload(outs, task_name)
             if workload is None:
-                raise RuntimeError("Cannot find workload in attribute of this schedule")
+                raise RuntimeError(
+                    f"Cannot find TOPI workload {task_name}. "
+                    "Is it registered with `register_topi_compute`?"
+                )
             tgt = Target.current()
             cfg = DispatchContext.current.query(tgt, workload)
             return topi_schedule(cfg, outs, *args, **kwargs)
@@ -253,7 +256,7 @@ def traverse(tensors):
         for t in tensors:
             op = t.op
             wkl = traverse(op.input_tensors)
-            if wkl:
+            if wkl is not None:
                 return wkl
 
             if "workload" in op.attrs:

python/tvm/relay/frontend/onnx.py

@@ -526,23 +526,6 @@ def _impl_v1(cls, inputs, attr, params):
                 raise tvm.error.OpAttributeInvalid(msg.format(attr["auto_pad"]))
             attr.pop("auto_pad")
 
-        # Check if the requested convolution is a group conv1d, if so convert it to conv2d.
-        # TODO(jwfromm) Remove once proper group_conv1d is supported.
-        group_conv1d = False
-        if dimension_picker("conv")(attr) == "conv1d" and attr.get("group") != 1:
-            group_conv1d = True
-            # Expand input from NCW to NCHW
-            data = _op.expand_dims(data, axis=2)
-            # Expand kernel from OIW to OIHW
-            kernel = _op.expand_dims(kernel, axis=2)
-            # Add new value to kernel_shape, strices, dilation, pads, if needed
-            attr["kernel_shape"] = [1] + list(attr["kernel_shape"])
-            if "strides" in attr:
-                attr["strides"] = [1] + list(attr["strides"])
-            if "dilations" in attr:
-                attr["dilations"] = [1] + list(attr["dilations"])
-            if "pads" in attr:
-                attr["pads"] = [0, attr["pads"][0], 0, attr["pads"][1]]
         attr["channels"] = kernel_shapes[0][0]
         out = AttrCvt(
             op_name=dimension_picker("conv"),
@@ -555,10 +538,6 @@ def _impl_v1(cls, inputs, attr, params):
             custom_check=dimension_constraint(),
         )([data, kernel], attr, params)
 
-        # If this was a group_conv1d, squish output back to NCW.
-        if group_conv1d:
-            out = _op.squeeze(out, axis=[2])
-
         use_bias = len(inputs) == 3
         if use_bias:
             out = _op.nn.bias_add(out, inputs[2])

python/tvm/relay/op/strategy/cuda.py

@@ -689,20 +689,36 @@ def conv1d_strategy_cuda(attrs, inputs, out_type, target):
     if dilation[0] < 1:
         raise ValueError("dilation should be a positive value")
     strategy = _op.OpStrategy()
-    if layout == "NCW":
-        strategy.add_implementation(
-            wrap_compute_conv1d(topi.cuda.conv1d_ncw),
-            wrap_topi_schedule(topi.cuda.schedule_conv1d_ncw),
-            name="conv1d_ncw.cuda",
-        )
-    elif layout == "NWC":
-        strategy.add_implementation(
-            wrap_compute_conv1d(topi.cuda.conv1d_nwc),
-            wrap_topi_schedule(topi.cuda.schedule_conv1d_nwc),
-            name="conv1d_nwc.cuda",
-        )
+    if attrs.groups == 1:
+        if layout == "NCW":
+            strategy.add_implementation(
+                wrap_compute_conv1d(topi.cuda.conv1d_ncw),
+                wrap_topi_schedule(topi.cuda.schedule_conv1d_ncw),
+                name="conv1d_ncw.cuda",
+            )
+        elif layout == "NWC":
+            strategy.add_implementation(
+                wrap_compute_conv1d(topi.cuda.conv1d_nwc),
+                wrap_topi_schedule(topi.cuda.schedule_conv1d_nwc),
+                name="conv1d_nwc.cuda",
+            )
+        else:
+            raise ValueError("Unsupported conv1d layout {}".format(layout))
     else:
-        raise ValueError("Unsupported conv1d layout {}".format(layout))
+        if layout == "NCW":
+            strategy.add_implementation(
+                wrap_compute_group_conv1d(topi.cuda.group_conv1d_ncw),
+                wrap_topi_schedule(topi.cuda.schedule_group_conv1d_ncw),
+                name="group_conv1d_ncw.cuda",
+            )
+        elif layout == "NWC":
+            strategy.add_implementation(
+                wrap_compute_group_conv1d(topi.cuda.group_conv1d_nwc),
+                wrap_topi_schedule(topi.cuda.schedule_group_conv1d_nwc),
+                name="group_conv1d_nwc.cuda",
+            )
+        else:
+            raise ValueError("Unsupported conv1d layout {}".format(layout))
     return strategy
 
 

python/tvm/relay/op/strategy/generic.py

@@ -637,6 +637,49 @@ def conv1d_strategy(attrs, inputs, out_type, target):
     return strategy
 
 
+def wrap_compute_group_conv1d(topi_compute):
+    """wrap conv1d topi compute"""
+
+    def _compute_group_conv1d(attrs, inputs, out_type):
+        """Compute definition of conv1d"""
+        strides = get_const_tuple(attrs.strides)
+        padding = get_const_tuple(attrs.padding)
+        dilation = get_const_tuple(attrs.dilation)
+        out_dtype = attrs.out_dtype
+        out_dtype = inputs[0].dtype if out_dtype in ("same", "") else out_dtype
+        return [
+            topi_compute(inputs[0], inputs[1], strides, padding, dilation, attrs.groups, out_dtype)
+        ]
+
+    return _compute_group_conv1d
+
+
+@override_native_generic_func("group_conv1d_strategy")
+def group_conv1d_strategy(attrs, inputs, out_type, target):
+    """group_conv1d generic strategy"""
+    logger.warning("group_conv1d is not optimized for this platform.")
+    layout = attrs.data_layout
+    dilation = get_const_tuple(attrs.dilation)
+    if dilation[0] < 1:
+        raise ValueError("dilation should be a positive value")
+    strategy = _op.OpStrategy()
+    if layout == "NCW":
+        strategy.add_implementation(
+            wrap_compute_conv1d(topi.nn.group_conv1d_ncw),
+            wrap_topi_schedule(topi.generic.schedule_group_conv1d_ncw),
+            name="group_conv1d_ncw.generic",
+        )
+    elif layout == "NWC":
+        strategy.add_implementation(
+            wrap_compute_conv1d(topi.nn.group_conv1d_nwc),
+            wrap_topi_schedule(topi.generic.schedule_group_conv1d_nwc),
+            name="group_conv1d_nwc.generic",
+        )
+    else:
+        raise ValueError("Unsupported conv1d layout {}".format(layout))
+    return strategy
+
+
 # conv1d_transpose
 def wrap_compute_conv1d_transpose(topi_compute):
     """wrap conv1d_transpose topi compute"""

python/tvm/relay/op/strategy/x86.py

@@ -360,24 +360,41 @@ def conv3d_strategy_cpu(attrs, inputs, out_type, target):
 def conv1d_strategy_cpu(attrs, inputs, out_type, target):
     """conv1d x86 strategy"""
     layout = attrs.data_layout
+    groups = attrs.groups
     dilation = get_const_tuple(attrs.dilation)
     if dilation[0] < 1:
         raise ValueError("dilation should be a positive value")
     strategy = _op.OpStrategy()
-    if layout == "NCW":
-        strategy.add_implementation(
-            wrap_compute_conv1d(topi.nn.conv1d_ncw),
-            wrap_topi_schedule(topi.x86.schedule_conv1d_ncw),
-            name="conv1d_ncw.x86",
-        )
-    elif layout == "NWC":
-        strategy.add_implementation(
-            wrap_compute_conv1d(topi.nn.conv1d_nwc),
-            wrap_topi_schedule(topi.x86.schedule_conv1d_nwc),
-            name="conv1d_nwc.x86",
-        )
+    if groups == 1:
+        if layout == "NCW":
+            strategy.add_implementation(
+                wrap_compute_conv1d(topi.nn.conv1d_ncw),
+                wrap_topi_schedule(topi.x86.schedule_conv1d_ncw),
+                name="conv1d_ncw.x86",
+            )
+        elif layout == "NWC":
+            strategy.add_implementation(
+                wrap_compute_conv1d(topi.nn.conv1d_nwc),
+                wrap_topi_schedule(topi.x86.schedule_conv1d_nwc),
+                name="conv1d_nwc.x86",
+            )
+        else:
+            raise ValueError("Unsupported conv1d layout {}".format(layout))
     else:
-        raise ValueError("Unsupported conv1d layout {}".format(layout))
+        if layout == "NCW":
+            strategy.add_implementation(
+                wrap_compute_group_conv1d(topi.nn.group_conv1d_ncw),
+                wrap_topi_schedule(topi.x86.schedule_group_conv1d_ncw),
+                name="group_conv1d_ncw.x86",
+            )
+        elif layout == "NWC":
+            strategy.add_implementation(
+                wrap_compute_group_conv1d(topi.nn.group_conv1d_nwc),
+                wrap_topi_schedule(topi.x86.schedule_group_conv1d_nwc),
+                name="group_conv1d_nwc.x86",
+            )
+        else:
+            raise ValueError("Unsupported conv1d layout {}".format(layout))
     return strategy
 
 

python/tvm/topi/cuda/conv1d.py

@@ -29,8 +29,7 @@ def conv1d_ncw(cfg, data, kernel, strides, padding, dilation, out_dtype="float32
     return nn.conv1d_ncw(data, kernel, strides, padding, dilation, out_dtype)
 
 
-@autotvm.register_topi_schedule("conv1d_ncw.cuda")
-def schedule_conv1d_ncw(cfg, outs):
+def _schedule_conv1d_ncw(cfg, outs):
     """TOPI schedule callback of conv1d ncw for cuda gpu
 
     Parameters
@@ -51,7 +50,7 @@ def schedule_conv1d_ncw(cfg, outs):
     s = te.create_schedule([x.op for x in outs])
 
     def _callback(op):
-        if op.tag == "conv1d_ncw":
+        if op.tag == "conv1d_ncw" or op.tag == "group_conv1d_ncw":
             pad_data = op.input_tensors[0]
             kernel = op.input_tensors[1]
             conv = op.output(0)
@@ -140,13 +139,27 @@ def _callback(op):
     return s
 
 
+@autotvm.register_topi_schedule("conv1d_ncw.cuda")
+def schedule_conv1d_ncw(cfg, outs):
+    return _schedule_conv1d_ncw(cfg, outs)
+
+
+@autotvm.register_topi_compute("group_conv1d_ncw.cuda")
+def group_conv1d_ncw(cfg, data, kernel, strides, padding, dilation, groups, out_dtype="float32"):
+    return nn.group_conv1d_ncw(data, kernel, strides, padding, dilation, groups, out_dtype)
+
+
+@autotvm.register_topi_schedule("group_conv1d_ncw.cuda")
+def schedule_group_conv1d_ncw(cfg, outs):
+    return _schedule_conv1d_ncw(cfg, outs)
+
+
 @autotvm.register_topi_compute("conv1d_nwc.cuda")
 def conv1d_nwc(cfg, data, kernel, strides, padding, dilation, out_dtype="float32"):
     return nn.conv1d_nwc(data, kernel, strides, padding, dilation, out_dtype)
 
 
-@autotvm.register_topi_schedule("conv1d_nwc.cuda")
-def schedule_conv1d_nwc(cfg, outs):
+def _schedule_conv1d_nwc(cfg, outs):
     """TOPI schedule callback of conv1d nwc for cuda gpu
 
     Parameters
@@ -167,7 +180,7 @@ def schedule_conv1d_nwc(cfg, outs):
     s = te.create_schedule([x.op for x in outs])
 
     def _callback(op):
-        if op.tag == "conv1d_nwc":
+        if op.tag == "conv1d_nwc" or op.tag == "group_conv1d_nwc":
             pad_data = op.input_tensors[0]
             kernel = op.input_tensors[1]
             conv = op.output(0)
@@ -254,3 +267,18 @@ def _callback(op):
     traverse_inline(s, outs[0].op, _callback)
 
     return s
+
+
+@autotvm.register_topi_schedule("conv1d_nwc.cuda")
+def schedule_conv1d_nwc(cfg, outs):
+    return _schedule_conv1d_nwc(cfg, outs)
+
+
+@autotvm.register_topi_compute("group_conv1d_nwc.cuda")
+def group_conv1d_nwc(cfg, data, kernel, strides, padding, dilation, groups, out_dtype="float32"):
+    return nn.group_conv1d_nwc(data, kernel, strides, padding, dilation, groups, out_dtype)
+
+
+@autotvm.register_topi_schedule("group_conv1d_nwc.cuda")
+def schedule_group_conv1d_nwc(cfg, outs):
+    return _schedule_conv1d_nwc(cfg, outs)

python/tvm/topi/generic/nn.py

@@ -54,6 +54,40 @@ def schedule_conv1d_nwc(outs):
     return _default_schedule(outs, False)
 
 
+def schedule_group_conv1d_ncw(outs):
+    """Schedule for group_conv1d_ncw
+
+    Parameters
+    ----------
+    outs: Array of Tensor
+          The computation graph description of group_conv1d_ncw
+          in the format of an array of tensors.
+
+    Returns
+    -------
+    sch: Schedule
+        The computation schedule for the op.
+    """
+    return _default_schedule(outs, False)
+
+
+def schedule_group_conv1d_nwc(outs):
+    """Schedule for group_conv1d_nwc
+
+    Parameters
+    ----------
+    outs: Array of Tensor
+          The computation graph description of group_conv1d_nwc
+          in the format of an array of tensors.
+
+    Returns
+    -------
+    sch: Schedule
+        The computation schedule for the op.
+    """
+    return _default_schedule(outs, False)
+
+
 def schedule_conv2d_hwcn(outs):
     """Schedule for conv2d_hwcn