提交 7d48ea8a authored 作者: Ian Goodfellow's avatar Ian Goodfellow

pep8

上级 07ec3ea1
......@@ -821,15 +821,18 @@ class GpuCAReduce(GpuOp):
def _k_init(self, *args):
return """
const int threadCount = blockDim.x * blockDim.y * blockDim.z;
const int threadNum = threadIdx.z * blockDim.x * blockDim.y + threadIdx.y * blockDim.x + threadIdx.x;
const int threadNum = threadIdx.z * blockDim.x * blockDim.y
+ threadIdx.y * blockDim.x + threadIdx.x;
extern __shared__ float buf[];
float myresult = 0.0f;
if (warpSize != 32)
{
// TODO: set error code
// 2012-09-17 IG: as of today, Fred says this is unlikely to change for any GPU released
// anytime soon. Hard to fix as _k_init is called by c_support_code_apply and
// 2012-09-17 IG: as of today, Fred says this is unlikely
// to change for any GPU released
// anytime soon. Hard to fix as _k_init is called by
// c_support_code_apply and
// c_support_code_apply is not passed a 'fail' code string.
Z[0] = -666;
return;
......
Markdown 格式
0%
您添加了 0 到此讨论。请谨慎行事。
请先完成此评论的编辑!
注册 或者 后发表评论