Data Values Extracted From Two Or More Clustered Hyperion Financial Management (HFM) Servers Is Different Due To Floating Point Arithmetic
(Doc ID 2337888.1)
Last updated on JULY 08, 2022
Applies to:
Hyperion Financial Management - Version 11.1.2.0.00 and laterInformation in this document applies to any platform.
HFM calculations are carried out with 15 significant decimal digits; this includes the digits to the left and right of the the decimal point. The round-off errors are considered normal for floating point calculations. In fact, the relative error is 1.29 x 10^-15.
There are multiple sources for round-off errors in floating point calculations, including conversions between decimal and binary (used internally to store numbers), internal calculations such as aggregation, consolidation, etc, and storage and retrieval in the database.
The order of calculations can also affect the visibility of round-off errors. That is 15 significant decimal digits is the maximum accuracy can expect with the product.
HS.Round() functions can be used to narrow down numbers especially in dynamic account rules.
When one or more servers are being used in an HFM deployment, each server may have slight differences in the order of operations that causes minute differences in results.
There is no workaround for this, nor is there any type of fix to accommodate this fact of computer science.
Symptoms
After consolidation is complete a data extract retrieves two different values for same POV.
The usually manifests itself in a very small error in the last decimal place.
For example
312981.610000002 from server X
312981.610000001 from server Y
Cause
To view full details, sign in with your My Oracle Support account. |
|
Don't have a My Oracle Support account? Click to get started! |
In this Document
| Symptoms |
| Cause |
| Solution |