- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
We are seeing are builds change on a day to day basis when none of the source code (HDL and Nios SW) changes. We perform nightly builds and regression testing and from one day to the next with no changes other than the next day, we sometimes pass and meet timing, we pass and our timing analysis has changed from the previous day's numbers, or we fail miserably on all fronts, but an interim build with no changes will succeed.
Is there a magic switch that we need to flip in the tools to make things consistent?Link Copied
1 Reply
- Mark as New
- Bookmark
- Subscribe
- Mute
- Subscribe to RSS Feed
- Permalink
- Report Inappropriate Content
For the same source code, you should trust Quartus to produce exactly the same fitting.
However, variation of build may occur due to either fitter random seed value change or timestamp change(which in effect is an invisible change of source code). This is the case with a device working close to its maximum performance in terms of speed/resource. For example, Altera states that seed value may change timing performance by 10% whatever that means. The same applies to timestamp or any other trivial change. It is clealy a limitation on the fitter technology - may God help them.
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