[NNVM, TOPI] Bug fixes (apache#24)

* bug fix

* passing the parameters when building the nnvm graph before extracting the tasks in autotvm

* bug fix for operator fusion

* fixing integration and tuning scripts

Author: tmoreau89

nnvm/python/nnvm/top/nn.py

@@ -98,25 +98,25 @@ def compute_conv2d(attrs, inputs, _):
     if groups == 1 and layout == 'NCHW4c' and inputs[0].dtype == 'int8':
         # pylint: disable=assignment-from-no-return
         out = topi.nn.conv2d(inputs[0], inputs[1], strides, padding,
-                             dilation, layout, out_dtype=out_dtype)
+                             dilation, layout, out_dtype)
         # pylint: enable=assignment-from-no-return
     elif groups == 1:
         out = topi.nn.conv2d(
-            inputs[0], inputs[1], strides, padding, dilation, layout, out_dtype=out_dtype)
+            inputs[0], inputs[1], strides, padding, dilation, layout, out_dtype)
     elif layout == "NCHW" and \
          groups == get_const_int(inputs[0].shape[1]) and \
          groups == channels:
         out = topi.nn.depthwise_conv2d_nchw(
-            inputs[0], inputs[1], strides, padding, dilation, out_dtype=out_dtype)
+            inputs[0], inputs[1], strides, padding, dilation, out_dtype)
     elif layout in ["NCHW", "NCHW4c"]:
         out = topi.nn.group_conv2d_nchw(inputs[0], inputs[1], strides, padding, dilation, groups,
-                                        out_dtype=out_dtype)
+                                        out_dtype)
     elif layout == "NHWC" and \
          kernel_layout == "HWOI" and \
          groups == get_const_int(inputs[0].shape[3]) and \
          groups == channels:
         out = topi.nn.depthwise_conv2d_nhwc(
-            inputs[0], inputs[1], strides, padding, dilation, out_dtype=out_dtype)
+            inputs[0], inputs[1], strides, padding, dilation, out_dtype)
     else:
         raise ValueError("not support arbitrary group number for now")
 

python/tvm/autotvm/task/nnvm_integration.py

@@ -16,7 +16,7 @@
 logger = logging.getLogger('autotvm')
 
 
-def extract_from_graph(graph, shape, dtype, target, symbols, target_host=None):
+def extract_from_graph(graph, shape, dtype, target, symbols, params, target_host=None):
     """ Extract tuning tasks from a nnvm graph.
 
     This function collects tuning tasks by building the graph and trace all the calls to topi.
@@ -33,6 +33,8 @@ def extract_from_graph(graph, shape, dtype, target, symbols, target_host=None):
         The compilation target
     symbols : Array of nnvm.symbol
         Array of nnvm symbols want to be tuned
+    params : dict of str to NDArray
+        The parameter dictionary.
     target_host: tvm.target.Target
         The host compilation target
 
@@ -66,7 +68,8 @@ def extract_from_graph(graph, shape, dtype, target, symbols, target_host=None):
         # run compiler to collect all TOPI calls during compilation
         nnvm.compiler.engine.clear_cache()
         with ApplyHistoryBest([]):
-            nnvm.compiler.build(graph, target=target, shape=shape, dtype=dtype)
+            nnvm.compiler.build(graph, target=target, shape=shape, dtype=dtype,
+                                target_host=target_host, params=params)
         nnvm.compiler.engine.clear_cache()
 
         logger.disabled = old_state
@@ -80,7 +83,7 @@ def extract_from_graph(graph, shape, dtype, target, symbols, target_host=None):
                          template_key='direct')
             tasks.append(tsk)
         except topi.InvalidShapeError:
-            print("shape error")
+            print("[Warning] invalid shape")
 
     return tasks
 

vta/python/vta/top/vta_conv2d.py

@@ -190,7 +190,7 @@ def schedule_packed_conv2d(cfg, outs,
     ewise_inputs = []
     ewise_ops = []
     conv2d_res = []
-    assert output.op.input_tensors[0].dtype == "int32"
+    assert "int" in output.op.input_tensors[0].dtype
 
     def _traverse(op):
         if topi.tag.is_broadcast(op.tag):

vta/scripts/tune_conv.py

@@ -20,7 +20,7 @@ def my_clip(x, a_min, a_max):
     x = tvm.compute(x.shape, lambda *i: tvm.max(x(*i), const_min), name="clipB")
     return x
 
-def conv2d(N, CI, H, W, CO, KH, KW, strides, padding, in_dtype, out_dtype):
+def conv2d(N, CI, H, W, CO, KH, KW, strides, padding, dilation, in_dtype, out_dtype):
     data_shape = (N//env.BATCH, CI//env.BLOCK_IN, H, W, env.BATCH, env.BLOCK_IN)
     kernel_shape = (CO//env.BLOCK_OUT, CI//env.BLOCK_IN, KH, KW, env.BLOCK_OUT, env.BLOCK_IN)
     bias_shape = (N//env.BATCH, CO//env.BLOCK_OUT, 1, 1, env.BATCH, env.BLOCK_OUT)
@@ -33,7 +33,7 @@ def conv2d(N, CI, H, W, CO, KH, KW, strides, padding, in_dtype, out_dtype):
     kernel = tvm.placeholder(kernel_shape, name="kernel", dtype=env.wgt_dtype)
 
     with tvm.target.vta():
-        res = topi.nn.conv2d(data, kernel, padding=padding, strides=strides,
+        res = topi.nn.conv2d(data, kernel, padding=padding, strides=strides, dilation=dilation,
                              layout='NCHW%dn%dc' % (env.BATCH, env.BLOCK_IN), out_dtype='int32')
         res = topi.add(res, bias)
         res = topi.right_shift(res, 8)
@@ -46,13 +46,13 @@ def conv2d(N, CI, H, W, CO, KH, KW, strides, padding, in_dtype, out_dtype):
         s = tvm.create_schedule([res.op])
 
 
-    return s, [data, kernel, bias, res] 
+    return s, [data, kernel, bias, res]
 
 if __name__ == '__main__':
-    N, CI, H, W, CO, KH, KW, strides, padding, in_dtype, out_dtype = \
-        1, 64, 56, 56, 64, 3, 3, (1, 1), (1, 1), 'int8', 'int32'
+    N, CI, H, W, CO, KH, KW, strides, padding, dilation, in_dtype, out_dtype = \
+        1, 64, 56, 56, 64, 3, 3, (1, 1), (1, 1), (1, 1), 'int8', 'int32'
 
-    task = autotvm.task.create(conv2d, args=(N, CI, H, W, CO, KH, KW, strides, padding, in_dtype, out_dtype),
+    task = autotvm.task.create(conv2d, args=(N, CI, H, W, CO, KH, KW, strides, padding, dilation, in_dtype, out_dtype),
             target=tvm.target.vta(env.MODEL), target_host=env.target_host, template_key='direct')
     print(task.config_space)
 
@@ -62,7 +62,7 @@ def conv2d(N, CI, H, W, CO, KH, KW, strides, padding, in_dtype, out_dtype):
 
     measure_option = autotvm.measure_option(
             builder=autotvm.LocalBuilder(build_func=vta.vta_autotvm_build_func),
-            runner=autotvm.RPCRunner(env.TARGET, 'fleet', 9190, number=4, repeat=3, timeout=30,
+            runner=autotvm.RPCRunner(env.TARGET, '10.77.1.109', 9190, number=4, repeat=3, timeout=30,
                                      check_correctness=True))
 
     tuner = autotvm.tuner.RandomTuner(task)

vta/scripts/tune_resnet.py

@@ -89,8 +89,9 @@ def extract_tasks(sym, params, target, target_host):
     sym = vta.graph.pack(sym, shape_dict, env.BATCH, env.BLOCK_OUT)
 
     with vta.build_config():
-        tasks = autotvm.task.extract_from_graph(sym, target=target, target_host=target_host,
-                                                shape=shape_dict, dtype=dtype_dict, symbols=(nnvm.sym.conv2d,))
+        tasks = autotvm.task.extract_from_graph(sym, shape=shape_dict, dtype=dtype_dict, target=target,
+                                                params=params, symbols=(nnvm.sym.conv2d,), target_host=target_host,
+                                                )
     return tasks
 
 
@@ -169,7 +170,7 @@ def tune_tasks(tasks,
 
         'measure_option':  autotvm.measure_option(
                 builder=autotvm.LocalBuilder(build_func=vta.vta_autotvm_build_func),
-                runner=autotvm.RPCRunner(env.TARGET, 'fleet', 9190,
+                runner=autotvm.RPCRunner(env.TARGET, '10.77.1.109', 9190,
                     number=4, repeat=3, timeout=60,
                     check_correctness=True))
     }
@@ -202,7 +203,7 @@ def tune_tasks(tasks,
 
         # upload module to device
         print("Upload...")
-        remote = autotvm.measure.request_remote(env.TARGET, 'fleet', 9190, timeout=10000)
+        remote = autotvm.measure.request_remote(env.TARGET, '10.77.1.109', 9190, timeout=10000)
         remote.upload(tmp.relpath(filename))
         rlib = remote.load_module(filename)
 

vta/tests/python/integration/test_benchmark_topi_conv2d.py

@@ -47,7 +47,7 @@ def run_cpu_conv2d(env, remote, wl, target):
 
     with target:
         res_conv = topi.nn.conv2d(
-            data, kernel, (wl.hstride, wl.wstride), (wl.hpad, wl.wpad), "NCHW", "int32")
+            data, kernel, (wl.hstride, wl.wstride), (wl.hpad, wl.wpad), (1, 1), "NCHW", "int32")
         res = topi.right_shift(res_conv, 8)
         res = my_clip(res, 0, 127)
         res = topi.cast(res, "int8")
@@ -202,7 +202,7 @@ def run_vta_conv2d(env, remote, wl, target, check_correctness=True, print_ir=Fal
 
     with target:
         res_conv = topi.nn.conv2d(
-            data, kernel, (wl.hstride, wl.wstride), (wl.hpad, wl.wpad),
+            data, kernel, (wl.hstride, wl.wstride), (wl.hpad, wl.wpad), (1, 1),
             "NCHW%dn%dc" % (env.BATCH, env.BLOCK_IN), 'int32')
         res = topi.right_shift(res_conv, 8)
         res = topi.add(res, bias)