Performance considerations when calculating cost and valuation prices
If you calculate and actualize cost and valuation prices, and performance is low, you can decide to use extra servers ('parallel bshells') to increase the speed of the calculation and actualization process.
Calculation and actualization
Calculation in parallel bshell mode is done bottom-up, and level by level, starting with the lowest level (purchased items). Within a level, the items are calculated in single-level mode.
If you actualize cost and valuation prices in parallel bshell mode, no report is printed with actualized costs and valuations. Instead, after running the Calculate Standard Cost (ticpr2210m000) session, you can use the Print Inventory Revaluation Transactions (whina1422m000) session to print the revaluation transactions.
Batch size
In the Calculate Standard Cost (ticpr2210m000) session on the Parallel Bshells tab, in the Batch size field, you can define the number of items that are submitted in one go to the parallel servers (bshells). If you define a large batch, you reduce the overhead per item, which can speed up the calculation and actualization process. However, a large batch can also slow down the process because of the waiting times at the end of a level. When in doubt, use the default of this field.
Logfile
In you are calculating or actualizing in
parallel bshell mode, warnings that are normally send to a printer or display
device, are now logged in logfile log.cpr2210
. Use the Display Error Log (ttstperrlog) session to review the
logfile.
Additional information about the calculation and
actualization process, for example, the number of items and runtimes, can be
logged in log file log.cprdll2001
. To get this additional logfile, add the following setting to your BWC
file:
-set LOGGER=1 -set LOGGER_LEVEL=INFO