NA-User
[Top] [All Lists]

[na-user] Computer Speed Question

To: <na-user@contesting.com>
Subject: [na-user] Computer Speed Question
From: k8cc <k8cc@mediaone.net> (k8cc)
Date: Fri, 08 Dec 2000 18:16:00 -0500
At 03:10 PM 12/8/00 -0600, Jim George wrote:

>As a new serious NA user, I did a serious effort in the CQWW CW contest
>with NA 10.51.  Overall, I like NA and feel comfortable with it.
>
>However a problem developed as the QSO database increased towards my
>eventual 2600 QSO total, and that was an increasing delay in the time
>needed to process a QSO and log it.  I noted that QSOs with call signs
>starting with "J" took longer than QSOs with less common prefixes.  Among
>the worst were the JA1s
><snip>
>
>Other data:  My PC has enough room for 16000 QSOs with the 8 MByte RAM.  I
>use the Super Check Partial feature with the Master.dta file available from
>the NA web site (which is quite good).  I haven't tried to run the database
>with SCP disabled.  That could make a difference, but I like that during a
>contest and would rather use that feature.  My PC is quite old, a 25 MHz
>386SX, so the problem might be as basic as needing to buy a faster PC.  I
>am prepared to do that if that's really needed.  But I wanted to pose the
>issue to this esteemed group of more experienced users, plus Dave.

When running Super Check Partial, processor speed and disk drive 
performance is critical.  Each time you type a alphanumeric character 
(after the third character prior to 10.51 and after the second character in 
10.51) NA has to pull a whole string of callsigns off the disk drive, one 
character at a time to fill the partial window.

As you've surmised, your CPU is probably marginal to run Super Check 
Partial.  Prior to a last year, we were running a motley assortment of CPUs 
in the K8CC multi-multi shack; everything from 486DLC-40s to 
486DX4-100s.  We put 6000+ WPX QSOs into a multi-multi log on more than one 
occasion with no complaints from the ops.  My recommendation is that 
anything from a 486DX2-66 or above should work OK.

One thing that helped these 486s get by is the use of a disk cache.  Some 
of these computers only had 2 MB of RAM, but we would set up at least 1MB 
as a SmartDrive disk cache.  If MASTER.DTA can set undisturbed in the disk 
cache, then callsigns are pulled from RAM which is a lot faster.  Since 
most MASTER.DTA files are roughly 600K, 1 MB leaves room for the typical 
.QDF log file to sit there too.  (If the SmartDrive buffer is too small, 
the MASTER.DTA would have to be swapped in and out which would be a mess.)

Motherboards and processors are so cheap anymore that upgrading is pretty 
painless.  Last year we put new AMD K6-400 CPUs and motherboards in every 
shack computer.  Unfortunately most new motherboards won't accept old 30 
pin SIMM memory, so you'll probably need new RAM too.  Motherboard, CPU and 
a 32 MB DIMM memory stick cost about $250 per machine.  We kept everything 
else - video, HDD, expansion cards, etc.  One thing, its getting harder and 
harder to find modern motherboards with ISA slots.

>A second issue is that I am used (from my prior CT experience) to entering
>the call with the "insert" key and continuing to type in the last two or
>three characters of the call.  NA, however, transmits the call as it was in
>the "call field" at the time the "insert" or "enter" key was pressed, and
>does not pick up the characters typed after the "enter" fubnction was
>activated.  Sometimes it will allow one more character, but usually none.
>CT allows the user to continue to type characters and sends every character
>entered in the keyboard buffer at the time the call is sent.  Is this a
>fundamental difference between NA and CT?  It's a pretty minor issue, but
>as a slowish typist, over the years I have gotten used to entering a call
>and then typing the last part so I can start transmitting sooner, rather
>than waiting until the entire call is entered in the call field.  Comments
>on this-other than I need to be a fast touch typist!

NA does indeed do this, and has for years.  Suppose I'm going to send 
KA5XYZ - if I type KA5 and press insert, NA will start sending KA5.  As 
long as I keep ahead of it, NA will keep sending the digits, so if I hit X 
before the "5", NA will send "X", if I hit the Y before "X" is done, NA 
will send "Y", etc.  If I'm slow and NA catches up with me, it no longer 
continues to send any additional letters I might enter.

Perhaps CT takes it farther than we do?  I'm open to ideas on this.


Dave/K8CC


--
Submissions:              na-user@contesting.com
Administrative requests:  na-user-REQUEST@contesting.com
WWW:                      http://datom.contesting.com/
Questions:                owner-na-user@contesting.com


<Prev in Thread] Current Thread [Next in Thread>