wehub-resource-sync
102 lines
3.0 KiB
Python
102 lines
3.0 KiB
Python
# Copyright (c) 2025 PaddlePaddle Authors. All Rights Reserved.
|
|
#
|
|
# Licensed 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.
|
|
|
|
import os
|
|
import unittest
|
|
|
|
import numpy as np
|
|
|
|
import paddle
|
|
import paddle.incubate.cc as pcc
|
|
import paddle.incubate.cc.typing as pct
|
|
|
|
os.environ["AP_WORKSPACE_DIR"] = "/tmp/paddle/ap"
|
|
|
|
|
|
def GetPirProgram(fused_func, tensor_args):
|
|
dtypes = tuple(tensor.dtype for tensor in tensor_args)
|
|
func = fused_func.func_overload_ctx.dtypes2func.get(dtypes, None)
|
|
return str(func.infer_program.forward_program)
|
|
|
|
|
|
def IsSupportDevice():
|
|
if paddle.is_compiled_with_cuda():
|
|
prop = paddle.device.cuda.get_device_properties()
|
|
cc = prop.major * 10 + prop.minor
|
|
return cc == 80
|
|
|
|
if paddle.is_compiled_with_rocm():
|
|
return True
|
|
|
|
return False
|
|
|
|
|
|
class TestMatmulEpilogue(unittest.TestCase):
|
|
def setUp(self):
|
|
dtype = 'float16'
|
|
x_shape = [32, 16, 16]
|
|
self.x = paddle.randn(x_shape, dtype=dtype)
|
|
self.x.stop_gradient = False
|
|
|
|
y_shape = [16, 16]
|
|
self.y = paddle.randn(y_shape, dtype=dtype)
|
|
self.y.stop_gradient = False
|
|
|
|
b_shape = [32, 16, 16]
|
|
self.b = paddle.randn(b_shape, dtype=dtype)
|
|
self.b.stop_gradient = False
|
|
|
|
def getSubGraph(self):
|
|
B = pct.DimVar(32)
|
|
M = pct.DimVar(16)
|
|
K = pct.DimVar(16)
|
|
N = pct.DimVar(16)
|
|
DType = pct.DTypeVar("T", "float16")
|
|
|
|
def foo(
|
|
x: pct.Tensor([B, M, K], DType),
|
|
w: pct.Tensor([K, N], DType),
|
|
b: pct.Tensor([B, M, N], DType),
|
|
):
|
|
y = paddle.matmul(x, w)
|
|
tmp = paddle.nn.functional.relu(y)
|
|
tmp2 = tmp + b
|
|
return tmp2
|
|
|
|
return foo
|
|
|
|
def test_subgraph(self):
|
|
foo = self.getSubGraph()
|
|
backend_device = 'dcu' if paddle.is_compiled_with_rocm() else 'cuda'
|
|
fused_foo = pcc.compile(
|
|
foo,
|
|
ap_path=f"{os.path.dirname(paddle.__file__)}/apy/matmul_pass",
|
|
backend_device=backend_device,
|
|
)
|
|
generated_pir_program = GetPirProgram(
|
|
fused_foo, [self.x, self.y, self.b]
|
|
)
|
|
self.assertTrue(
|
|
'pd_op.ap_variadic' in generated_pir_program, "fusion failed"
|
|
)
|
|
if IsSupportDevice():
|
|
ap_outs = fused_foo(self.x, self.y, self.b)
|
|
dy_outs = foo(self.x, self.y, self.b)
|
|
for dy_out, ap_out in zip(dy_outs, ap_outs):
|
|
np.testing.assert_allclose(dy_out, ap_out, atol=1e-1)
|
|
|
|
|
|
if __name__ == "__main__":
|
|
unittest.main()
|