|
|
| Author |
Message |
GTBecker
Joined: 18 Jan 2006
Posts: 472
Location: Cape Coral
|
|
Posted: 29 June 2010, 3:24 AM Post subject: |
 |
|
Further, Variant handling differs if the s5 calculation is offered differently:
| Code: | s5 = (s1 * s2) + (s3 - s4)
|
... for Doubles yields
s5= 0.000953339808631881000, the same as without the parens around (s3 - s4)
and for Variants yields
s5= 0.000953339808631836000, a third - very close - value. Is this the denormalization issue Don shows in Singles, above? |
|
| Back to top |
|
 |
FFMan
Joined: 09 Jan 2010
Posts: 264
|
|
Posted: 29 June 2010, 13:58 PM Post subject: |
|
|
Interesting work in VB. Personally I never used variants, I always kinda figured these were for lazy people but i'm sure its not as easy as that.
However, some more analysis reveals the problem may not be as bad as i thought. The calcs that showed the differences were used to judge when 2 GPS reading straddled my imaginary line, and once used for that were discarded. Subsequent calcs then worked out the intersect point and time offset etc
Over 3 test laps my time offset difference when compared zbasic to excel were:-
-0.0136401088851807 -0.0054256324306400 -0.0066200165252785
What i don't know, until the next race on sunday is how accurate this is compared to transponder based timing but it seems that it should be possible to improved the resolution substantially from what 10hz suggests you should get.
I'll post back when i have race results. |
|
| Back to top |
|
|
|
ZBasic Home
Forum Rules
Memberlist
Usergroups
RSS Feed
Site Search
Links
Download
Digests and Subscriptions
Profile
Log in to check your private messages
Log in
Register