I ran some tests, and it looks like we'd only get a score in the sixties. I'm not worried though, because I don't know of a single bot in the rumble that would fire bullets less than power 1.0 at close range, except in the endgame. And even if there was one, it would likely have a HOT gun, which would be dodged by orbit movement anyway.
Hmm, apparently not quite as good as 119. The scores may not have quite stabilised yet, but it has dropped down to second with 78.06 APS, about -1.2 IIRC :(
On the plus side I briefly get to retake the lead :)
Not to worry though, I have the superlatively evil (nasty, unpleasant, dishonourable, cheating or the insult of your choice) version 1.2 of PralDeGuerre under development. Now if only it performs up to the statistical predictions...
By the way, how much space is that move change saving? I think Sheldor said 2 bytes.
An alternative that might perform better (can't say for sure, you would have to benchmark it) would be to use the old move, but change the "... - (1.1 - 1e-8) ) <= 1)" to "... - 1) <= 2)".
This is slightly less accurate as the energy detect is .001..3.999 rather than .001..3.001, but it could work better.
You would also need to update the rand chance to be "4 / 0.6*Math.sqrt(..." rather than "3.0/(0.6*Math.sqrt(.." to account for the wider window.
The drop in score is due to two problems with the table. I just fixed both of them.
Yep, this method of enemy fire detection is two bytes smaller than the previous. The other
char based check you just posted would be (going by memory here) only one byte smaller than the original, which would be one byte bigger than my new method.
It may be compiler dependent, I am just using standard jdk 1.6, but every time I have sized that it has been 2 bytes different.
1 is dconst_1, size 1 byte.
Any other double constant (except zero and possibly stuff that can be cast from an int constant) is
ldc2_w, indexbyte1, indexbyte2, size 3 bytes.
As to which is functionally superior I am not sure, I haven't done extensive benchmarking on that variant of Yatagan movement as those extra 2 bytes were not enough for me to add anything else, so I didn't use/test it much.
You do not have permission to edit this page, for the following reasons:
- The action you have requested is limited to users in the group: Users.
- You must confirm your email address before editing pages. Please set and validate your email address through your user preferences.
You can view and copy the source of this page.Skilgannon (talk)
Return to Thread:Talk:Yatagan/Source/1.2.0/reply (14).
Ah well, I had a few hours at the top, but it was not to last.
It has only had 1000 battles, but this version looks to have lost half a point from 1.1.9.
So I was testing some other stuff and decided while I was at it to bung some yatagan variants through my benchmarks.
Note, take the figures with a grain of salt as there can be up to 1% jitter and these are all quite close. Also my figures often show .5 to 1% higher than the rumble, but what I got was:
- jk.sheldor.nano.Yatagan 1.1.7 79.6%
- jk.sheldor.nano.Yatagan 1.2.1 78.8%
- nz.jdc.nano.Yatagan117a 1.0.0 79.4%
- nz.jdc.nano.Yatagan117b 1.0.0 79.0%
1.1.7 and 1.2.1 are your released versions, both scores align reasonably with the rumble and show 1.2.1 being down a little.
117a is a copy of 1.1.7 with the movement code change suggested above.
117b is a copy of 117a, but with the addition of setAdjustGunForRobotTurn(true);
Not sure what is causing the (relative) weakness in 1.2.1, it could be the change in energy detection, or it could be putting oscillate first in the table, or another table bug. Another possibility is the gun adjust. I have varying results with that and often find it hurts rather than helps the score. The result of the otherwise identical 117b being worse than 117a leads me to believe the gun adjust is a factor.
It looks like putting the oscillate first was the issue, 1.2.2 is tied with 1.1.7. I wonder if the
setAdjustGunForRobotTurn(true); would be better replaced by
setAdjustRadarForGunTurn(true); to cut down on the radar lock issues?
That seems to have made quite a large difference. I suppose I should not be surprised about a pattern matcher not liking gaps in its pattern. @#$!&* pattern guns, they are rather strong. I'm not sure even my cheaty dev version will have much margin over 1.2.3. I had better get back to the tuning. Or start praying for inspiration from saint Michael.