|
|
|
Member 46 posts
Registered: Jan 2006
I was curious when taking a look at the results so far in the LG competition to see that some of the top scores belong to some very unsuspecting players. Is there some secret to these super high scores, or are these just extremely talented players, head and shoulders above the rest. When I began examining demos of the top entries in the LG competition I spotted one peculiar artifact. It seems if you hold TAB and watch the ping and packetloss (PL) throughout the match, the bots seem to have constant packet loss of one, that spikes up to as high as 4 or more. When I compared these demos to my own, and the demos from some other reputable players such as Interceptor, Jukka, Joulukuu, and Gamer, there was no packet loss. Zero. This got me thinking, what is the cause of this PL and does it have any impact on the scores? Through some testing I have disocvered a way to replicate this packetloss on my own setup. Setting /sys_highpriority -1 and /sys_yieldcpu 1 in console creates the aformentioned scenario where the bot possesses packetloss. In essence by lowering the priority of your client / server , the amount of cpu cycles dedicated to the bot are decreased, effectively lowering the bot's skill level, primarilly movement and aim. I've verified the effects of low priority myself by playing a single round vs the bots and shafted a 56% lg on my first try (5% over my best legitimiate record which took dozens of tries). The bot's movement was very linear, exhibited lack of air movement, and his aim was in many ways sub standard. Which results exhibit packetloss in the demos? I have not examined every single demo, but I did go through the top entries. Entries 1 through 5 belonging to RST-Thornado (Bra), ANM-Zip (Bra), [6]Mur (Bra), React (Aus), and Renzo (?) respectively all contain packet loss. In my opinion, all results that exhibit packetloss should be disqualified from the competition. You can argue back and forth regarding whether the players that submitted these records had knowledge of the effects of low priority. You may verify any of my statements by downloading the demos from the LGC website.
Member 810 posts
Registered: Jan 1970
so basicly they are just shit players?
Member 116 posts
Registered: Mar 2006
Member 46 posts
Registered: Jan 2006
They aren't shit players, but their lg% and scores are exaggerated. The advantage in using low priority is massive, that much is certain. In any case, their results are inadmissable.
Member 1011 posts
Registered: Feb 2006
its just a case of hangovers from their standard qw configs
these settings just need to be added to the lgc.cfg and checked
Member 810 posts
Registered: Jan 1970
yeah, but basicly they are no better than ilf, fnosk, paradoks,qmole, doser and SD. Trying to get an unfair advantage over fellow legit players?
Member 693 posts
Registered: Jan 2006
yeah, but basicly they are no better than ilf, fnosk, paradoks,qmole, doser and SD. Trying to get an unfair advantage over fellow legit players? What a ridiculous statement. It's a valid config option to use and is not a cheat. I have it in my config and have had for a long time because it was the only way I could stop some mouse lag issues when running ventrilo. If you ban this command then you will effectively ban anyone who needs that in their config to get a non-laggy mouse.
Member 810 posts
Registered: Jan 1970
yeah, i added 2 a2k names there. but here you go add this aswell ---->
Member 46 posts
Registered: Jan 2006
What a ridiculous statement. It's a valid config option to use and is not a cheat. I have it in my config and have had for a long time because it was the only way I could stop some mouse lag issues when running ventrilo.
If you ban this command then you will effectively ban anyone who needs that in their config to get a non-laggy mouse. Incorrect. /sys_highpriority 0 and /sys_yieldcpu 0 (aka normal priority) will not introduce any latency or lag to your mouse. Setting /sys_highpriority 1 and /sys_yieldcpu 1 might. If you have a lagged mouse at normal priority, you should increase priority via task manager to your mouse drivers process, and not set your Quake to low priority. This only hurst your aim (especially online where AI is not a part of the equation). When low priority is set for the server, the bot plays for the lack of a better phrase "Like Ilf, except without the aimbot."
Member 11 posts
Registered: Apr 2006
actually i find harder to hit players with PL, and i was wondering why i wasnt able to get 170+ frags and 57 lg like i was able to do with ibc bots, maybe that is the reason, btw, i use sys_yieldcpu 1, but i use highpriority 0, im not a newcomer..i play quake since 1997 and i have no reason to " cheat ", altough i'm not at my prime form right now since i stoped playing quake a while ago, well anyway, in season 2 we'll see. []'s
Member 810 posts
Registered: Jan 1970
OMG def is jealous of brazilian shaft skills. L-O-S-E-R
Just fix these parameters and watch ZIP, MUR, THORN doing even better.
Member 7 posts
Registered: Jan 2006
Well i just tested Def`s variables of sys_highpriority and sys_yieldcpu and shafting is alot easier B(
Member 810 posts
Registered: Jan 1970
my high and yeld are set to 0
Member 46 posts
Registered: Jan 2006
[6]Mur: I'm not acusing anyone of cheating. There is alot of variables that can influence your system to cause the bot to packetloss.
However, the fact remains, when the bot loses packets it moves alot worse, and aims alot worse, because its being deprived of CPU in one form or another. The problem revolves around priority.
AnM Zip is being very helpful at the moment helping me to narrow down the problem and look for a solution. So far everything we've tried has hit a brickwall. No matter what Zip does, his bot still has pl. Regardless of setting priority to high or realtime with no yieldcpu, or even as far as running a dedicated mvd server with high priority, and not being connected to the server (only myself connected remotely and observing the bot), the bot still exhibits packetloss.
It would be helpful if you guys could post your system info if possible. Start > Run > DxDiag.exe and attach it in here. Also if you find any solution to getting rid of the PL please post it in here.
Two lines of thought we are following at the moment:
1) Tweaking services to remove all background services / processes (this seems to have reduced Zip's PL from constant 2-4 to 0-1 but not removed the problem entirely.)
2) It could be an ATI Issue (unlikely though possible). Zip and I have identical hardware except for NV and ATI graphics cards. Do you guys have ATI hardware?
Member 11 posts
Registered: Apr 2006
------------------ System Information ------------------ Time of this report: 4/21/2006, 19:37:54 Machine name: *************** Operating System: Windows XP Professional (5.1, Build 2600) Service Pack 2 (2600.xpsp_sp2_gdr.050301-1519) Language: English (Regional Setting: Portuguese) System Manufacturer: NVIDIA System Model: AWRDACPI BIOS: Phoenix - AwardBIOS v6.00PG Processor: Unknown CPU Typ, MMX, 3DNow, ~2.2GHz Memory: 768MB RAM Page File: 379MB used, 1498MB available Windows Dir: E:\WINDOWS DirectX Version: DirectX 9.0c (4.09.0000.0904) DX Setup Parameters: Not found DxDiag Version: 5.03.2600.2180 32bit Unicode
if u want the full report dxdiag.txt, give ur email and i'll send
Member 810 posts
Registered: Jan 1970
if someone wins a big prize at the lottery i have a good idea how to spend the money do a worldwide qw lan
Member 693 posts
Registered: Jan 2006
What a ridiculous statement. It's a valid config option to use and is not a cheat. I have it in my config and have had for a long time because it was the only way I could stop some mouse lag issues when running ventrilo.
If you ban this command then you will effectively ban anyone who needs that in their config to get a non-laggy mouse. Incorrect. /sys_highpriority 0 and /sys_yieldcpu 0 (aka normal priority) will not introduce any latency or lag to your mouse. Setting /sys_highpriority 1 and /sys_yieldcpu 1 might. If you have a lagged mouse at normal priority, you should increase priority via task manager to your mouse drivers process, and not set your Quake to low priority. This only hurst your aim (especially online where AI is not a part of the equation). When low priority is set for the server, the bot plays for the lack of a better phrase "Like Ilf, except without the aimbot." Well I arrived at the value for sys_highpriority for some reason - I wouldn't just set it to -1 for shits and giggles
Member 46 posts
Registered: Jan 2006
Anm, Mur, and I tried alot of stuff on my localhost, and theirs, and we werent able to solve it. The pl comes and goes, its a bitch. I give up for now. It's definitely tied into priority, and it seems that it requires /restart console command for the priority to take effect on the bot. One thing is for sure, the effect on the bot between 0pl and even an ocassional 1pl spike seems to make all the difference in the bots ability to aim and move.
Member 11 posts
Registered: Apr 2006
i was able to record a demo with PL 0..i uploaded to lgc. the trick, is keep restarting, until the bot gets in the game with PL 0 ( after the yellow pl comes off ) []'s
Moderator 1329 posts
Registered: Apr 2006
I have /sys_highpriority 0 and /sys_yieldcpu 1 and I've never even thought of them affecting "botmatch" since I don't play it. If you think those affect the bot's behaviour, just omit the correct settings to lgc.cfg. You can argue back and forth regarding whether the players that submitted these records had knowledge of the effects of low priority. I've always seen that packet loss with bot skill 10 but I didn't give it a much thought since it didn't appear on them with skill 20 (ping 13).
Moderator 1329 posts
Registered: Apr 2006
Yup, it's the /sys_yieldcpu 1 that makes packet loss on bot, just tested it. Now I'll try to play some with value of "0" and see if it really affects as def claimed.
Edit: No, it's not. I still got some pl in the middle of the match (I used restart command after changing the values).
Member 401 posts
Registered: Mar 2006
BTW def who removed your 70% demo? It was awesome It looked timescaled tho
Member 46 posts
Registered: Jan 2006
BTW def who removed your 70% demo? It was awesome It looked timescaled tho http://quakeworld.nu/forum/viewtopic.php?id=522&p=2 ready my post on that thread, it explains.
Member 810 posts
Registered: Jan 1970
sent a new demo with pl0 to the comp...
score: ~10300..
cya..
Member 810 posts
Registered: Jan 1970
Member 6 posts
Registered: Apr 2006
exaggerated boy
Member 11 posts
Registered: Apr 2006
new scores from me and zip =) []'s
Member 6 posts
Registered: Apr 2006
omg.. 58% and 60% lg Ban def! He is contributing to noone else besides you guys win this competition.
News Writer 646 posts
Registered: Mar 2006
Member 810 posts
Registered: Jan 1970
Def is trying 24/7 to get #1... poor guy. Zip has the best lg on the planet, period.
|
|
|
|