From 6332ea99abd7ca7782cb22c8fafd938918536ca7 Mon Sep 17 00:00:00 2001
From: AlexeyAB <alexeyab84@gmail.com>
Date: Thu, 22 Feb 2018 21:13:08 +0000
Subject: [PATCH] one more fix
---
src/convolutional_kernels.cu | 4 +++-
1 files changed, 3 insertions(+), 1 deletions(-)
diff --git a/src/convolutional_kernels.cu b/src/convolutional_kernels.cu
index 3918e5e..ee9b534 100644
--- a/src/convolutional_kernels.cu
+++ b/src/convolutional_kernels.cu
@@ -77,7 +77,8 @@
__global__ void cuda_f32_to_f16(float* input_f32, size_t size, half *output_f16)
{
int idx = blockIdx.x * blockDim.x + threadIdx.x;
- if (idx < size) output_f16[idx] = __float2half_rn(input_f32[idx]);
+ if (idx < size) output_f16[idx] = __float2half(input_f32[idx]);
+ //if (idx < size) *((unsigned int *)output_f16 + idx) = __float2half(input_f32[idx]);
}
void cuda_convert_f32_to_f16(float* input_f32, size_t size, half *output_f16) {
@@ -88,6 +89,7 @@
{
int idx = blockIdx.x * blockDim.x + threadIdx.x;
if (idx < size) output_f32[idx] = __half2float(input_f16[idx]);
+ //if (idx < size) output_f32[idx] = __half2float(*((unsigned int *)input_f16 + idx));
}
void cuda_convert_f16_to_f32(half* input_f16, size_t size, float *output_f32) {
--
Gitblit v1.10.0