Merge pull request #2229 from LDOUBLEV/fix_2.0

enable memory optim

Merge pull request #2229 from LDOUBLEV/fix_2.0
enable memory optim
a7937102 · zhoujun · GitHub · ef3f72f0 · 4431a09c · a7937102
显示空白变更内容
内联并排

Showing with 2 addition and 1 deletion

tools/infer/utility.py tools/infer/utility.py +2 -1

未找到文件。
--- a/tools/infer/utility.py
+++ b/tools/infer/utility.py
@@ -129,7 +129,8 @@ def create_predictor(args, mode, logger):
            #config.set_mkldnn_op({'conv2d', 'depthwise_conv2d', 'pool2d', 'batch_norm'})
            args.rec_batch_num = 1

-    # config.enable_memory_optim()
+    # enable memory optim
+    config.enable_memory_optim()
    config.disable_glog_info()

    config.delete_pass("conv_transpose_eltwiseadd_bn_fuse_pass")