- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
I mounted a directory of host to mic0. Then I wrote a toy MIC program in native mode to read file from this nfs directory. It turned out that the performance of reading from host disk is extemely terrible, only got less than 100MB/s for the first reading (assuming that no memory buffer is used). Is it normal? Any reason about this?
Thanks!
Link Copied
2 Replies
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
You need to parallelize your file IO. The Linux kernel syscalls usually assume that the caller does the parallelization. This might help improve the IO performance that you are seeing.
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
You might also find the following blog helpful: http://software.intel.com/en-us/blogs/2014/01/07/improving-file-io-performance-on-intel-xeon-phi

Reply
Topic Options
- Subscribe to RSS Feed
- Mark Topic as New
- Mark Topic as Read
- Float this Topic for Current User
- Bookmark
- Subscribe
- Printer Friendly Page