模型调用Pad接口填充报错For ‘Pad’, output buffer memset failed.

[longvoyage]

1 系统环境

• 硬件环境(Ascend/GPU/CPU): Ascend/GPU/CPU
• MindSpore版本: mindformer =2.1.0
• 执行模式（PyNative/ Graph）: 不限
• Python版本: Python=3.10
• 操作系统平台: Ubuntu18.04

2 报错信息

2.1 脚本信息

import mindspore
from mindspore import Tensor, nn, ops
import numpy as np
class TestNet(nn.Cell):
    def __init__(self):
        super(Net, self).__init__()
        self.pad = nn.Pad(paddings=((0, 0), (0, 999999)))
    def construct(self, x):
        return self.pad(x)
x =  Tensor(np.random.randn(9999, 9999))
pad = TestNet()
output = pad(x)

2.2 报错信息

该报错在开启Padding的Conv2d，DepthwiseConv2d，MaxPooling，AveragePooling等多个算子执行时均会出现。

For ‘Pad’, output buffer memset failed.

3 根因分析

参数设置过大,显存不足,无法执行.

4 解决方案

降低参数值

import mindspore
from mindspore import Tensor, nn, ops
import numpy as np
class TestNet(nn.Cell):
    def __init__(self):
        super(Net, self).__init__()
        self.pad = nn.Pad(paddings=((0, 0), (0, 99)))
    def construct(self, x):
        return self.pad(x)
x =  Tensor(np.random.randn(99, 99))
pad = TestNet()
output = pad(x)