[AMP OP&Test] arange op support fp16/bf16 #51106
Conversation
|
你的PR提交成功，感谢你对开源项目的贡献! |
|
❌ The PR is not created using PR's template. You can refer to this Demo. |
|
|
|
|
| self.case = (0, 5, 1) | ||
|
|
||
| def test_check_output(self): | ||
| self.check_output(atol=1e-3) |
|
|
||
| namespace phi { | ||
|
|
||
| template <typename T> | ||
| __global__ void Range(T start, T step, int64_t size, T* out) { | ||
| CUDA_KERNEL_LOOP(index, size) { out[index] = start + step * index; } | ||
| using MPType = typename phi::dtype::MPTypeTrait<T>::Type; | ||
| CUDA_KERNEL_LOOP(index, size) { out[index] = static_cast<T>(static_cast<MPType>(start) + static_cast<MPType>(step) * index); } |
There was a problem hiding this comment.
start和step是在计算中是固定的值，可以在计算前做static_cast成一个临时变量
| @@ -39,7 +45,8 @@ void ArangeKernel(const Context& dev_ctx, | |||
| T step_value = GetValue<T, Context>(dev_ctx, step); | |||
There was a problem hiding this comment.
建议直接在这里拿到start、end、step的时候直接转成MPType，后续都使用MPType的值去计算
| @@ -1,4 +1,4 @@ | |||
| # Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. | |||
| # Copyright (c) 2020 PaddlePaddle Authors. All Rights Reserved. | |||
There was a problem hiding this comment.
这里其实不用修改，但是如果改了就把2020改成2023吧
| or not core.is_bfloat16_supported(core.CUDAPlace(0)), | ||
| "core is not compiled with CUDA and not support the bfloat16", | ||
| ) | ||
| def test_check_output(self): |
There was a problem hiding this comment.
修改输入输出这部分不建议写在test_check_output中，建议直接继承OpTest重写setUp，其他参考TestArangeOp写就行
| self.outputs = { | ||
| 'Out': convert_float_to_uint16(np.arange(self.case[0], self.case[1], self.case[2])) | ||
| } | ||
| self.check_output(atol=1e-2) |
|
@xiegegege Performance of model ResNet50_bs32_dygraph has been increased from 248.402 to 480.483,(480.483-248.402)/248.402 equals 0.9343,which is greater than threshold 0.06 |
python/paddle/tensor/creation.py
Outdated
| @@ -1234,7 +1234,7 @@ def arange(start=0, end=None, step=1, dtype=None, name=None): | |||
| check_dtype( | |||
| dtype, | |||
| 'dtype', | |||
| ['float32', 'float64', 'int32', 'int64'], | |||
| ['float32', 'float64', 'int32', 'int64', 'float16', 'bfloat16'], | |||
| class TestBFloat16ArangeOp(OpTest): | ||
| def setUp(self): | ||
| self.op_type = "range" | ||
| self.__class__.op_type = self.op_type |
PR types
Others
PR changes
OPs
Describe
增加了单测