From e7688a05a194e3c8baf3c11fbf09b7f5e8640a77 Mon Sep 17 00:00:00 2001
From: Joseph Redmon <pjreddie@gmail.com>
Date: Wed, 06 May 2015 21:08:16 +0000
Subject: [PATCH] no idea
---
src/blas_kernels.cu | 8 ++++----
1 files changed, 4 insertions(+), 4 deletions(-)
diff --git a/src/blas_kernels.cu b/src/blas_kernels.cu
index d6f7143..636a9b5 100644
--- a/src/blas_kernels.cu
+++ b/src/blas_kernels.cu
@@ -15,10 +15,10 @@
if(i < N) X[i*INCX] *= ALPHA;
}
-__global__ void mask_kernel(int n, float *x, float *mask, int mod)
+__global__ void mask_kernel(int n, float *x, float *mask)
{
int i = (blockIdx.x + blockIdx.y*gridDim.x) * blockDim.x + threadIdx.x;
- if(i < n) x[i] = (i%mod && !mask[(i/mod)*mod]) ? 0 : x[i];
+ if(i < n && mask[i] == 0) x[i] = 0;
}
__global__ void copy_kernel(int N, float *X, int OFFX, int INCX, float *Y, int OFFY, int INCY)
@@ -49,9 +49,9 @@
check_error(cudaPeekAtLastError());
}
-extern "C" void mask_ongpu(int N, float * X, float * mask, float mod)
+extern "C" void mask_ongpu(int N, float * X, float * mask)
{
- mask_kernel<<<cuda_gridsize(N), BLOCK>>>(N, X, mask, mod);
+ mask_kernel<<<cuda_gridsize(N), BLOCK>>>(N, X, mask);
check_error(cudaPeekAtLastError());
}
--
Gitblit v1.10.0