From cad4d1618fee74471d335314cb77070fee951a42 Mon Sep 17 00:00:00 2001
From: AlexeyAB <alexeyab84@gmail.com>
Date: Sun, 25 Feb 2018 13:29:44 +0000
Subject: [PATCH] Added support for Tensor Cores CC >= 7.0 (V100). For FP16/32 (mixed precision) define CUDNN_HALF should be used.
---
src/network.c | 3 +++
1 files changed, 3 insertions(+), 0 deletions(-)
diff --git a/src/network.c b/src/network.c
index c906b58..d23468d 100644
--- a/src/network.c
+++ b/src/network.c
@@ -316,6 +316,8 @@
net->layers[i].batch = b;
#ifdef CUDNN
if(net->layers[i].type == CONVOLUTIONAL){
+ cudnn_convolutional_setup(net->layers + i, cudnn_fastest);
+ /*
layer *l = net->layers + i;
cudnn_convolutional_setup(l, cudnn_fastest);
// check for excessive memory consumption
@@ -327,6 +329,7 @@
cudnn_convolutional_setup(l, cudnn_smallest);
l->workspace_size = get_workspace_size(*l);
}
+ */
}
#endif
}
--
Gitblit v1.10.0