pre_prepare: 1.1178
F0101 00:58:12.392118 726 conv2d_functor.tcc:44] Check failed: dev->IsInAddressRange(reinterpret_cast<size_t>(ptr)) Pointer 0x36d7b0 doesn't fit in FPGA device RAINMAN_BOARD_V3
*** Check failure stack trace: ***
@ 0xb67a637a google::LogMessage::Fail()
@ 0xb67a7542 google::LogMessage::SendToLog()
@ 0xb67a5fc8 google::LogMessage::Flush()
@ 0xb67a84c2 google::LogMessageFatal::~LogMessageFatal()
@ 0xb6a65e98 raintime::nn::functor::Conv2DOperatorFpga<>()
@ 0xb6a6c9a4 raintime::nn::ops::Conv2DOpImpl<>::Run()
@ 0xb6a4d9de raintime::sg::SGRunner::Run()
@ 0x18646 main
@ 0xb5d3d7f0 __libc_start_main
Aborted

@Corerain 是tutorial2的，人脸识别，拿来训练了一下，发现不能生成sg。但是这个例子提供了训练好的checkpoint文件，那个是可以的。麻烦能核实一下吗？

@Corerain 你好，我想问一下问什么你们的例子训练的模型生成sg也出现相同的问题(Check whether your GraphDef-interpreting binary is up to date with your GraphDef-generating binary)。例子也需要改才能用吗？

已经解决了，参数设置的问题

terminate called after throwing an instance of 'cv::Exception'
what(): OpenCV(3.4.4) /home/morris/workspace/system_dev/buildroot-2018.02.4/output/build/opencv3-3.4.4/modules/imgproc/src/resize.cpp:3784: error: (-215:Assertion failed) !ssize.empty() in function 'resize'

@Corerain 请问训练是用gpu，模型在cpu的docker中去编译，这个没有关系吗？还是说必须要用gpu的docker呢？

@Corerain 请问您是在plumber_env_cpu_16bit下复现的吗？

已经解决了，是路径的问题

root@c683f04e158c:/workspace/16bit/single_img_tt3/build# make
Scanning dependencies of target demo_runner_face5b
[ 25%] Building CXX object CMakeFiles/demo_runner_face5b.dir/RainBuilder.cc.o
In file included from /workspace/16bit/single_img_tt3/RainBuilder.cc:1:0:
/workspace/16bit/single_img_tt3/RainBuilder.hpp:4:10: fatal error: raintime/core.h: No such file or directory
#include "raintime/core.h"

@Corerain said in 使用plumber生成SG失败，一直报错:

community@corerain.com 已经邮箱发给你了，谢谢。