Ns Specific Stats
Tweedle
Join Date: 2002-11-07 Member: 7581Members
Join Date: 2002-11-07 Member: 7581Members
<div class="IPBDescription">Worth the trouble?</div> Right now I'm using Psychostats and it works fine.. but..
I don't think that the rankings are all that they could be. I don't know who leads marines to wins. I don't know who can take on several skulks in a row versus someone who can shotgun an AFKer over and over. I don't know the average time into game that marines get motion tracking compared to the average time of motion tracking <i>when marines win</i>, etc.
On top of this, a <b>lot</b> of log lines are exact duplicates sans the timestamps, especially so for gorges and commanders. Psychostats does no caching (that I could see.) I'm getting about a 50% speed increase when I use some simple caching. A stats program for NS could parse the logs much faster if it was designed <i>just for</i> NS.
Psychostats is an awesome program; much credit to them. NS, however, does not fall into storm the front/plant the bomb/aim for the head/most kills wins category that Psychostats was designed for.
Opinions?
I don't think that the rankings are all that they could be. I don't know who leads marines to wins. I don't know who can take on several skulks in a row versus someone who can shotgun an AFKer over and over. I don't know the average time into game that marines get motion tracking compared to the average time of motion tracking <i>when marines win</i>, etc.
On top of this, a <b>lot</b> of log lines are exact duplicates sans the timestamps, especially so for gorges and commanders. Psychostats does no caching (that I could see.) I'm getting about a 50% speed increase when I use some simple caching. A stats program for NS could parse the logs much faster if it was designed <i>just for</i> NS.
Psychostats is an awesome program; much credit to them. NS, however, does not fall into storm the front/plant the bomb/aim for the head/most kills wins category that Psychostats was designed for.
Opinions?
Comments
So if you were thinking of hacking up more of his code i don't think he'd be happy. But your right, NS does need its own type of stats programs.
Just for some extremely rough and probably completely negligible gauging ..
Right now I'm parsing at approx. 1MB/s. I'm not doing nearly as much as Psychostats does, as my current level of script qualifies as "parser" with a test method for stats gathering.
"1230 log files (28.07 MB) scanned in 2:42 minutes" I don't know if that includes file generation time, but we'll assume that it does, giving 120 seconds for 30MB of logs. Four seconds a meg.
1:4. Though in a complete state, I could easily see it being 1:2 or maybe even 3:4. Less time <b>and</b> more information. *shrug*
With caching: Total runtime: 20.419362s
Without: Total runtime: 27.229154s
Odd. I seem to have been owned by my own numbers.
/me goes to try to figure out what is going on.
edit:
caching on
<!--QuoteBegin--></span><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td><b>QUOTE</b> </td></tr><tr><td id='QUOTE'><!--QuoteEBegin-->Parsed 1483 logs.
Caching saved 100568/279838 parses. (35.94%)
Total runtime: 17.044509/s<!--QuoteEnd--></td></tr></table><span class='postcolor'><!--QuoteEEnd-->
caching off
<!--QuoteBegin--></span><table border='0' align='center' width='95%' cellpadding='3' cellspacing='1'><tr><td><b>QUOTE</b> </td></tr><tr><td id='QUOTE'><!--QuoteEBegin-->Parsed 1483 logs.
Caching saved /279838 parses. (0.00%)
Total runtime: 26.137584/s<!--QuoteEnd--></td></tr></table><span class='postcolor'><!--QuoteEEnd-->
Ehh, it helps, but it isn't as groundbreaking as I originally thought. Oh well.
I'm not sure wtf i'm talking about because i've never messed with caching, i'm just using my pseudo thinking here.
how is that possible