What's new for Excel add-in in PPS SP1

Peter Eb. has excelent post in Technet forums regarding SP1 improvements for Excel add-in:

"Here's a quick highlight of the changes our team made in SP1 to improve performance. The main scenarios addressed are "matrix rendering/what-if" and "assignment opening".

1. There is an option to clear the changelist after submission per form. This is for scenarios where large volumes of data are submitted at once, and there is no risk of validation errors. (avoids having the contributor clear the cache manually) This is also useful for long running assignments that may accumulate a large number of cells changed over multiple submissions
2. There is an option to turn off what-if per matrix. This is for scenarios where multiple matrices are being used that reference the writeback model, but might not have aggregated values that are affected by the what-if. (note this is also handy for shares input cases, where the value cannot be what-if'd by Analysis Services)
3. Formula handling. The formula scan is now much more efficient and the matching of formulas to the mdx location is also vastly improved.
4. Auto-recalc. For data entry scenarios when auto-recalc is enabled (remember spreading forces this mode on) we determined that most times the matrix is statically sized. (aka no grow/shrink). So in this case the addin just renders the data grid, formatting and the row/column headers are not touched. It's not quite Excel calculation fast, but definitely for smaller matrices the experience is much improved for contributors.


Mainly assignment opening benefits from the rendering improvements. But specifically in some scenarios the clear changelist after submission is a useful new option. There are some other changes as well (like the where clause change I've talked about frequently), but I think these are the visible changes that have not yet been identified. (I'll go back over the changes later and provide more details if I missed anything)

Some informal (but real world) feedback from internal deployments and some preview customers tell us that common refresh/what-if operations in SP1 are 50% of the time in RTM or better. In one extreme case which could take advantage of the option to turn of what-if the time was even 5%, but I don't think that will be typical. Opening assignments sounds like SP1 is 50% to 33% of the RTM time. So far I haven't seen any negative performance feedback, knock wood, cross fingers, and any other anti-jinx superstitions.

I can't guarantee that those performance numbers will match your customers experience, but am optimistic that formula users and spreading users will see improvements. (Localized builds will follow and will see the same perf. improvements)"

You can find complete discussion on:
http://forums.microsoft.com/TechNet/ShowPost.aspx?PostID=3445168&SiteID=17