Created by: nlj0011

您好，在跑distribute_deepwalk的训练阶段时提示offset is too big，然后训练中断。请问这个要怎么修改呀，多谢了~ paddle的版本是1.6，以下是server具体的报错信息。 terminate called after throwing an instance of 'Xbyak::Error' what(): offset is too big W1212 15:27:18.687328 19979 init.cc:212] *** Aborted at 1576135638 (unix time) try "date -d @1576135638" if you are using GNU date *** W1212 15:27:18.690397 19979 init.cc:212] PC: @ 0x0 (unknown) W1212 15:27:18.690459 19979 init.cc:212] *** SIGABRT (@0x1f400003fd1) received by PID 16337 (TID 0x7f38bbfff700) from PID 16337; stack trace: *** W1212 15:27:18.693048 19979 init.cc:212] @ 0x7f3ac3ba5160 (unknown) W1212 15:27:18.695641 19979 init.cc:212] @ 0x7f3ac31133f7 __GI_raise W1212 15:27:18.698221 19979 init.cc:212] @ 0x7f3ac31147d8 __GI_abort W1212 15:27:18.699606 19979 init.cc:212] @ 0x7f3a65bdbc65 __gnu_cxx::__verbose_terminate_handler() W1212 15:27:18.700861 19979 init.cc:212] @ 0x7f3a65bd9e06 __cxxabiv1::__terminate() W1212 15:27:18.702225 19979 init.cc:212] @ 0x7f3a65bd9e33 std::terminate() W1212 15:27:18.703485 19979 init.cc:212] @ 0x7f3a65c2c935 execute_native_thread_routine W1212 15:27:18.705993 19979 init.cc:212] @ 0x7f3ac3b9d1c3 start_thread W1212 15:27:18.708621 19979 init.cc:212] @ 0x7f3ac31c512d __clone W1212 15:27:18.711194 19979 init.cc:212] @ 0x0 (unknown)

另外worker有一处报错为PaddleCheckError: Expected posix_memalign(&p, alignment, size) == 0, but received posix_memalign(&p, alignment, size):12 != 0:0.