提交 54a3a323 authored 作者: Olivier Delalleau's avatar Olivier Delalleau

Minor typo fixes in comments

上级 e0f4ef5b
...@@ -2969,14 +2969,14 @@ def local_elemwise_fusion_op(OP, max_input_fct=lambda node: 1024): ...@@ -2969,14 +2969,14 @@ def local_elemwise_fusion_op(OP, max_input_fct=lambda node: 1024):
:param max_input_fct: a function that returns the maximum number of inputs :param max_input_fct: a function that returns the maximum number of inputs
that this elemwise can take (useful for GpuElemwise). that this elemwise can take (useful for GpuElemwise).
GPU kernel currently have a limit of 256 bytes for GPU kernel currently has a limit of 256 bytes for
the size of all parameter passed to it. As currently the size of all parameters passed to it. As currently
we pass many information only by parameter, we must we pass many information only by parameter, we must
limit how many op we fuse together to don't bust that limit how many ops we fuse together to avoid busting
256 limit. that 256 limit.
On the CPU we limit to 1024 input variable to the On the CPU we limit to 1024 input variable to the
resulting fused op. This is big enought that resulting fused op. This is big enough that
if we hit it, I'm not sure it will affect performance. if we hit it, I'm not sure it will affect performance.
""" """
def local_fuse(node): def local_fuse(node):
......
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