显卡配置:GeForce RTX 3090,CUDA Version: 11.2
Makefile配置:GPU=1 CUDNN=1 OPENCV=1
make 报错
nvcc fatal : Unsupported gpu architecture 'compute_30' Makefile:168: recipe for target 'obj/convolutional_kernels.o' failed
删除第18行:-gencode arch=compute_30,code=sm_30
make clean
make 报错
compilation terminated due to -Wfatal-errors.
Makefile:161: recipe for target 'obj/convolutional_layer.o' failed
make: *** [obj/convolutional_layer.o] Error 1
修改 Makefile配置:CUDNN=0
本文探讨了在使用GeForce RTX 3090显卡时,遇到CUDA Version 11.2的Makefile配置错误,特别是针对compute_30架构的不兼容问题。通过清理并调整CUDNN配置,解决了编译时的错误,涉及关键词如CUDA、GPU架构、CUDNN、OpenCV和Makefile。

2274

被折叠的 条评论
为什么被折叠?



