caffe

配置环境

caffe

cpu-only openblas

protobuf

编译caffe需要3.6~3.10版本，否则会报错

dnf install只能安装3.19版本

需要从源码编译，这里选择了3.9版本  protobuf的github仓  从源码编译安装

caffe-gpu mode

caffe的gpu模式需要用到cuda

make all会报错

模型

caffe train –solver=

caffe test -model= -weights=

caffe time -model

Berkeley-trained models

models/bvlc_googlenet

Network in Network model

没提供deploy.prototxt 需要由train_val.prototxt转换而来

Models from the BMVC-2014 paper "Return of the Devil in the Details: Delving Deep into Convolutional Nets"

 VGG_CNN_S

 VGG_CNN_M

Age and Gender Classification using Convolutional Neural Networks

age

gender

phoronix-test-suite

openbenchmarking用的工具pts

从源码编译安装

caffe

应该是先装caffe 然后再benchamrk

卡在装那一步 缺依赖

如果有小伙伴会用pts来benchmark,欢迎在评论区和我交流~