test_winograd_nnpack.py 6.22 KB
Newer Older
1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16
# Licensed to the Apache Software Foundation (ASF) under one
# or more contributor license agreements.  See the NOTICE file
# distributed with this work for additional information
# regarding copyright ownership.  The ASF licenses this file
# to you under the Apache License, Version 2.0 (the
# "License"); you may not use this file except in compliance
# with the License.  You may obtain a copy of the License at
#
#   http://www.apache.org/licenses/LICENSE-2.0
#
# Unless required by applicable law or agreed to in writing,
# software distributed under the License is distributed on an
# "AS IS" BASIS, WITHOUT WARRANTIES OR CONDITIONS OF ANY
# KIND, either express or implied.  See the License for the
# specific language governing permissions and limitations
# under the License.
hlu1 committed
17 18 19 20 21 22 23 24 25
import numpy as np
import tvm
from tvm import autotvm
from tvm.autotvm.task.space import FallbackConfigEntity
from tvm.contrib import nnpack
from tvm.contrib.pickle_memoize import memoize
import topi
import topi.testing
from topi.util import get_const_tuple
26
from pytest import skip
hlu1 committed
27 28 29 30 31 32 33 34 35 36 37 38 39 40 41 42 43 44 45 46 47 48 49 50 51 52 53 54 55 56 57 58 59 60 61 62


def verify_conv2d_nchw(batch, in_channel, in_size, num_filter, kernel, stride, padding, dilation=1, add_bias=False, add_relu=False,
        devices=['cuda', 'llvm -device=arm_cpu', 'opencl -device=mali']):
    print("Workload: (%d, %d, %d, %d, %d, %d, %d, %d)" % (batch, in_channel, in_size, num_filter, kernel, stride, padding, dilation))

    in_height = in_width = in_size

    A = tvm.placeholder((batch, in_channel, in_height, in_width), name='A')
    W = tvm.placeholder((num_filter, in_channel, kernel, kernel), name='W')
    bias = tvm.placeholder((num_filter, 1, 1), name='bias')

    a_shape = get_const_tuple(A.shape)
    w_shape = get_const_tuple(W.shape)
    bias_shape = get_const_tuple(bias.shape)
    dtype = A.dtype

    @memoize("topi.tests.test_topi_conv2d_nchw.verify_conv2d_nchw")
    def get_ref_data():
        a_np = np.random.uniform(size=a_shape).astype(dtype)
        w_np = np.random.uniform(size=w_shape).astype(dtype)
        b_np = np.random.uniform(size=bias_shape).astype(dtype)
        dw_np = topi.testing.dilate_python(w_np, (1, 1, dilation, dilation))
        c_np = topi.testing.conv2d_nchw_python(a_np, dw_np, stride, padding)
        if add_bias:
            b_np = np.random.uniform(size=bias_shape).astype(dtype)
            c_np += b_np
        if add_relu:
            c_np = np.maximum(c_np, 0)
        return a_np, w_np, b_np, c_np

    a_np, w_np, b_np, c_np = get_ref_data()

    def check_device(device):
        ctx = tvm.context(device, 0)
        if not ctx.exist:
63
            skip("s is not enabled" % device)
hlu1 committed
64 65 66 67 68 69 70 71 72 73 74 75 76 77 78 79 80 81 82 83 84 85 86 87 88 89 90 91 92 93 94 95 96 97 98 99 100 101
        print("Running on target: %s" % device)
        with tvm.target.create(device):
            C = topi.nn.conv2d(A, W, stride, padding, dilation, layout='NCHW', out_dtype=dtype)
            if add_bias:
                C = topi.add(C, bias)
            if add_relu:
                C = topi.nn.relu(C)
            s = topi.generic.schedule_conv2d_nchw([C])

        a = tvm.nd.array(a_np, ctx)
        w = tvm.nd.array(w_np, ctx)
        b = tvm.nd.array(b_np, ctx)
        c = tvm.nd.array(np.zeros(get_const_tuple(C.shape), dtype=C.dtype), ctx)
        if add_bias:
            func = tvm.build(s, [A, W, bias, C], device, name="relu_%d_%d_%d_%d_%d_%d_%d_%d" % (batch, in_channel, in_size, num_filter, kernel, stride, padding, dilation))
            func(a, w, b, c)
        else:
            func = tvm.build(s, [A, W, C], device, name="relu_%d_%d_%d_%d_%d_%d_%d_%d" % (batch, in_channel, in_size, num_filter, kernel, stride, padding, dilation))
            func(a, w, c)
        tvm.testing.assert_allclose(c.asnumpy(), c_np, rtol=1e-4)


    for device in devices:
        check_device(device)


class WinogradFallback(autotvm.FallbackContext):
    def _query_inside(self, target, workload):
        key = (target, workload)
        if key in self.memory:
            return self.memory[key]
        cfg = FallbackConfigEntity()
        cfg.template_key = 'winograd_nnpack_fp32'
        self.memory[key] = cfg
        return cfg

def test_conv2d_nchw():
    if not tvm.get_global_func("tvm.contrib.nnpack.convolution_inference_without_weight_transform", True):
102
        skip("extern function is not available")
hlu1 committed
103 104

    if not nnpack.is_available():
105
        skip("nnpack is not available")
hlu1 committed
106 107 108 109 110 111 112 113 114 115 116 117 118 119 120 121 122 123 124 125 126 127 128 129 130 131 132 133 134 135 136 137 138 139 140 141

    devices = ['llvm -device=arm_cpu']
    autotvm.DispatchContext.current.silent = True
    with WinogradFallback():
        # resnet 18 workloads
        verify_conv2d_nchw(1, 64, 56, 64, 3, 1, 1, devices=devices)
        verify_conv2d_nchw(1, 128, 28, 128, 3, 1, 1, devices=devices)
        verify_conv2d_nchw(1, 256, 14, 256, 3, 1, 1, devices=devices)
        verify_conv2d_nchw(1, 512, 7, 512, 3, 1, 1, devices=devices)

        # unet workloads
        verify_conv2d_nchw(1, 3, 192, 12, 3, 1, 1, add_bias=True, devices=devices)
        verify_conv2d_nchw(1, 4, 192, 12, 3, 1, 1, add_bias=True, devices=devices)
        verify_conv2d_nchw(1, 12, 96, 24, 3, 1, 1, add_bias=True, devices=devices)
        verify_conv2d_nchw(1, 24, 48, 48, 3, 1, 1, add_bias=True, devices=devices)
        verify_conv2d_nchw(1, 48, 24, 96, 3, 1, 1, add_bias=True, devices=devices)
        verify_conv2d_nchw(1, 96, 12, 180, 3, 1, 1, add_bias=True, devices=devices)
        verify_conv2d_nchw(1, 180, 6, 220, 3, 1, 1, add_bias=True, devices=devices)
        verify_conv2d_nchw(1, 220, 6, 180, 3, 1, 1, add_bias=True, devices=devices)
        verify_conv2d_nchw(1, 180, 12, 96, 3, 1, 1, add_bias=True, devices=devices)
        verify_conv2d_nchw(1, 96, 24, 48, 3, 1, 1, add_bias=True, devices=devices)
        verify_conv2d_nchw(1, 48, 48, 24, 3, 1, 1, add_bias=True, devices=devices)
        verify_conv2d_nchw(1, 24, 96, 12, 3, 1, 1, add_bias=True, devices=devices)
        verify_conv2d_nchw(1, 12, 192, 1, 3, 1, 1, add_bias=True, devices=devices)

        # relu, bias
        verify_conv2d_nchw(1, 64, 56, 64, 3, 1, 1, add_bias=True, devices=devices)
        verify_conv2d_nchw(1, 64, 56, 64, 3, 1, 1, add_relu=True, devices=devices)
        verify_conv2d_nchw(1, 64, 56, 64, 3, 1, 1, add_relu=True, add_bias=True, devices=devices)

        # werid workloads
        verify_conv2d_nchw(1, 3, 3, 3, 3, 1, 1, devices=devices)
        verify_conv2d_nchw(1, 13, 71, 59, 3, 1, 1, devices=devices)


if __name__ == "__main__":
142 143
    import pytest
    pytest.main()