Your test is flawed in that it doesn't reflect the speed difference in the
database engine between multi and single user. Write a loop that just does
the access without the Screen display for an accurate measure of the Engine.
Screen writes can have a great deal to do with video card, buss, VRam,
etc.....
----- Original Message -----
From: "Frank Radice" <[EMAIL PROTECTED]>
To: "RBASE mailing list (E-mail)" <[EMAIL PROTECTED]>
Cc: <[EMAIL PROTECTED]>
Sent: Sunday, June 24, 2001 10:55 PM
Subject: SET MULTI vs speed
> I wrote a simple CMD file that selected 3 values from a table (integer,
> text10 & text25) and displayed the resulting value. The integer values
run
> from 1 to 1000. I put the select & write in a WHILE loop and incremented
an
> int var so that it runs thru the loop 1000 times (selecting the row with
> the corresponding int value) & displays the results. I had it calculate
the
> time it took & display that value.
> The integer in the table is indexed.
>
> I got the following results.
>
> Database on MULTI setting Version Time, sec
> Speed factor
> ------------------ -------------------- ------------------------------
--
> ------------- ------------------
> Local drive C ON 6.5++ DOS 4.80
> Local drive C OFF 6.5++ DOS 1.21
> 4x faster
>
> speed ratio = 4.6
> Network ON 6.5++ DOS 27.85
> Network OFF 6.5++ DOS 1.23
> 18.6x faster
>
> --------------------------------------------------------------------------
--
> --------------------------------------------
>
> Local drive C ON 6.5++ WIN 3.57
> Local drive C OFF 6.5++ WIN 1.32
> 2.7x faster
>
> speed ratio = 4.6
> Network ON 6.5++ WIN 17.14
> Network OFF 6.5++ WIN 1.36
> 12.6x faster
>
> =============================================================
> The time is an average for several runs. I used computers (with approx
the
> same processor speed) running both Win98 & Win2000 and got similar
results.
> The network server is running Win2000.
>
> The 6.5++ WIN seems to run a little faster with MULTI ON and a little
slower
> with MULTI OFF, relative to 6.5++ DOS but the ratio of the speed with
MULTI
> ON for Local vs Network are both about 4.6 for both DOS & WIN versions.
> The speed with MULTI OFF does not vary significantly for Local vs
Network.
>
> Why does SET MULTI ON cause such a huge speed degration? Why would MULTI
> ON cause it to run slower on a network then on a local drive, whereas,
MULTI
> OFF runs the same on a network and/or local drive?
>
> Has anyone else observe this kind of behavior?
>
> Frank Radice
>
> [EMAIL PROTECTED]
>
