Difference between revisions of "Talk:Horizon/Gun Research"
Jump to navigation
Jump to search
(Archiving talk) |
(Robobot 0.1 : correcting user page links) |
||
| Line 1: | Line 1: | ||
==== ARCHIVED TALK: THROUGH 20071112 ==== | ==== ARCHIVED TALK: THROUGH 20071112 ==== | ||
| − | Note how I have already ''proven'' my gun is a piece of junk. -- [[AaronR]] | + | Note how I have already ''proven'' my gun is a piece of junk. -- [[User:AaronR|AaronR]] |
<code>Challenge complete, took 5 hrs 41 min 50 sec</code> | <code>Challenge complete, took 5 hrs 41 min 50 sec</code> | ||
| − | ... running with two threads on a dual-core processor. Man, I really need to upgrade Horizon to use [[KdTree/BucketPRKdTree|Simonton's tree]]. -- [[AaronR]] | + | ... running with two threads on a dual-core processor. Man, I really need to upgrade Horizon to use [[KdTree/BucketPRKdTree|Simonton's tree]]. -- [[User:AaronR|AaronR]] |
| − | * These battles take 15-20 seconds each on my machine using DCResearch. It may be well worth your time to plug in the tree. Is this really the time to run 15 seasons of '''fast learning'''?? -- [[Simonton]] | + | * These battles take 15-20 seconds each on my machine using DCResearch. It may be well worth your time to plug in the tree. Is this really the time to run 15 seasons of '''fast learning'''?? -- [[User:Simonton|Simonton]] |
| − | ** Yes... and the battles take 2:30 - 3:15 each with Horizon. =( Of course, part of the problem is that my bot loses a higher percentage of the rounds, meaning there is more idle time than with a better gun like DCResearch. -- [[AaronR]] | + | ** Yes... and the battles take 2:30 - 3:15 each with Horizon. =( Of course, part of the problem is that my bot loses a higher percentage of the rounds, meaning there is more idle time than with a better gun like DCResearch. -- [[User:AaronR|AaronR]] |
| − | I'll have 50 seasons of the new 0000 ready later. Using the tree really is much, much faster. It seems like there is a bug somewhere, though, since just about every one of the new version's scores is lower than the older release's. -- [[AaronR]] | + | I'll have 50 seasons of the new 0000 ready later. Using the tree really is much, much faster. It seems like there is a bug somewhere, though, since just about every one of the new version's scores is lower than the older release's. -- [[User:AaronR|AaronR]] |
<code>Challenge complete, took 5 hrs 22 min 47 sec</code> | <code>Challenge complete, took 5 hrs 22 min 47 sec</code> | ||
| − | ... running with one thread on a single-core processor, over more than three times the number of seasons as before. I'm happy now. -- [[AaronR]] | + | ... running with one thread on a single-core processor, over more than three times the number of seasons as before. I'm happy now. -- [[User:AaronR|AaronR]] |
| − | You should send me your gun code. Maybe I can find what's holding you back. -- [[Simonton]] | + | You should send me your gun code. Maybe I can find what's holding you back. -- [[User:Simonton|Simonton]] |
| − | * Thanks. I'll upload with source the TC version of the bot to my "tests" slot on the repository; I was just about to remove 1.0betaF3a from the rumble anyway. Chances are, the main bug is something really, really obvious, but that I haven't noticed because it's in the parts of the code I haven't looked at since the first release. =P ''Edit: the uploaded version is basically 0000, not 0001, so it has the smaller cluster size.'' -- [[AaronR]] | + | * Thanks. I'll upload with source the TC version of the bot to my "tests" slot on the repository; I was just about to remove 1.0betaF3a from the rumble anyway. Chances are, the main bug is something really, really obvious, but that I haven't noticed because it's in the parts of the code I haven't looked at since the first release. =P ''Edit: the uploaded version is basically 0000, not 0001, so it has the smaller cluster size.'' -- [[User:AaronR|AaronR]] |
| − | * Can you email it to me? As usual, the repository is down. -- [[Simonton]] | + | * Can you email it to me? As usual, the repository is down. -- [[User:Simonton|Simonton]] |
| − | * Sent. You know, this is getting really irritating... -- [[AaronR]] | + | * Sent. You know, this is getting really irritating... -- [[User:AaronR|AaronR]] |
---- | ---- | ||
Latest revision as of 08:42, 22 May 2009
ARCHIVED TALK: THROUGH 20071112
Note how I have already proven my gun is a piece of junk. -- AaronR
Challenge complete, took 5 hrs 41 min 50 sec
... running with two threads on a dual-core processor. Man, I really need to upgrade Horizon to use Simonton's tree. -- AaronR
- These battles take 15-20 seconds each on my machine using DCResearch. It may be well worth your time to plug in the tree. Is this really the time to run 15 seasons of fast learning?? -- Simonton
- Yes... and the battles take 2:30 - 3:15 each with Horizon. =( Of course, part of the problem is that my bot loses a higher percentage of the rounds, meaning there is more idle time than with a better gun like DCResearch. -- AaronR
I'll have 50 seasons of the new 0000 ready later. Using the tree really is much, much faster. It seems like there is a bug somewhere, though, since just about every one of the new version's scores is lower than the older release's. -- AaronR
Challenge complete, took 5 hrs 22 min 47 sec
... running with one thread on a single-core processor, over more than three times the number of seasons as before. I'm happy now. -- AaronR
You should send me your gun code. Maybe I can find what's holding you back. -- Simonton
- Thanks. I'll upload with source the TC version of the bot to my "tests" slot on the repository; I was just about to remove 1.0betaF3a from the rumble anyway. Chances are, the main bug is something really, really obvious, but that I haven't noticed because it's in the parts of the code I haven't looked at since the first release. =P Edit: the uploaded version is basically 0000, not 0001, so it has the smaller cluster size. -- AaronR
- Can you email it to me? As usual, the repository is down. -- Simonton
- Sent. You know, this is getting really irritating... -- AaronR
