We have programs which run many times slower under ARev3 than ARev1.

This is both on stand-alone and network PCs.

ARev3 is setup as: ..\arev.exe , /s /x /m4096

For example, one program uses arrays extensively:

 select customer

 loop

     readnext @id else stop

     read @record from customer.file, @id ..

     ...

     stats += 1

     repeat

This runs some 50 times slower under ARev3 than ARev1.

Displaying menus under ARev3 is some 4 times slower.

ARev3 programs run twice as slow on a note book off-line as they do on a network.

As the slowness is so wide-spread and so chronic, there must be a simple common reason.

The programs, menus, popups have been copied directly from ARev 1 to ARev3. New data files have been created under ARev3 with dictionary names similar to ARev1 with '.' replaced by '_'

(eg age.today -] age_today). Similarly with file names.

OI programs sharing the ARev3 files run at a very good speed, possibly faster than ARev1.

As it will be some time before all the systems will be converted to OI, we are taking a lot of flak for the dreadful slowness of ARev3.

I have searched this and other data bases for a document with suggestions I can believe in without success.

Can any one help?

Other than the inception of expanded memory, and different indexing, AREV 1 & 3 are internally (especially the exe's) quite similar in their string handling and RBASIC interpretation layers.

Try disabling the expanded memory to give a more reliable comparison.

Which network are you on? Are the network drivers set correctly? Are you using the NPP? Under Windows NT, byte range locking (DOS 3.1 driver) can be especially slow. Under Novell the differences between AREV 1 & 3 are virtually undetectable.

Perhaps a sizelock is set on a critical file, causing it to be heavily in overflow in the AREV 3 edition of your test bed. Overflow accounts for many speed problems. Other problems which are invisible but which impact performance include disk fragmentation.

How are you comparing the two products? Same data sets being attached? In this case you'd have to zip up the same paths of data files and explode each prior to use. Remember that the first time AREV 3 attaches files it converts the dictionaries to a new format. This can take some additional time.

Check that your "yield time to windows" flag (in AREV 3.12) is unset, and that your *.pif files (if any) have the same Windows foreground and background timeslice (and other) settings.

Steve

View this thread on the forum...