- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hello,
I am currently building a real-time multiprocessor scheduling algorithm that will preempt
tasks and move them between processors frequently. One of our biggest concerns is that
there will be a serious cache hit whenever a task moves between processors. Our target application
is a DSP program, and as is my understand such programs typically have a small cache footprint. I was wondering, if we were to move our DSP program over to the IPP, how large would the cache footprint be?
Thanks for the help,
Aaron
I am currently building a real-time multiprocessor scheduling algorithm that will preempt
tasks and move them between processors frequently. One of our biggest concerns is that
there will be a serious cache hit whenever a task moves between processors. Our target application
is a DSP program, and as is my understand such programs typically have a small cache footprint. I was wondering, if we were to move our DSP program over to the IPP, how large would the cache footprint be?
Thanks for the help,
Aaron
Link Copied
1 Reply
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
Hi Aaron,
I think it is not related to IPP directly. You can write program which work with memory in efficient way with or without using IPP. Generally speaking, you should reduce each task memory requirement to not allow one taskloadallcache memory.
Regards,
Vladimir
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