1 2011-01-28 00:00:45 <Kiba> AAA_awright: dude, do you know how many projects I have going?
  2 2011-01-28 00:04:13 <AAA_awright> Kiba: List em off
  3 2011-01-28 00:04:22 <Kiba> lee see
  4 2011-01-28 00:04:25 <Kiba> let see*
  5 2011-01-28 00:04:26 <AAA_awright> Can't possibly be more than me
  6 2011-01-28 00:04:28 <Kiba> bitcoin-chrome
  7 2011-01-28 00:04:36 <Kiba> bitcointo freelance assignment
  8 2011-01-28 00:04:39 <Kiba> writing assignment
  9 2011-01-28 00:04:44 <Kiba> writing for The Bitcoin Times
 10 2011-01-28 00:04:55 <Kiba> a freelance project outside of bitcoin
 11 2011-01-28 00:05:00 <Kiba> two sites I am working on
 12 2011-01-28 00:05:11 <Kiba> trying to pipeline more projects
 13 2011-01-28 00:05:28 <Kiba> err
 14 2011-01-28 00:05:29 <Kiba> jobs
 15 2011-01-28 00:05:45 <Kiba> art project in bitcoin
 16 2011-01-28 00:16:27 <ducki2p> Kiba: learn CSS and I got more for you
 17 2011-01-28 00:22:08 <OneFixt> when commentary is added in the "sendfrom" command in bitcoind, does it show up in the block?
 18 2011-01-28 00:34:57 <AAA_awright> kabo69: Yeah so my idea was abandoning the Times thing in favor of the journal
 19 2011-01-28 00:34:59 <AAA_awright> er
 20 2011-01-28 00:35:06 <AAA_awright> Kiba: Yeah so my idea was abandoning the Times thing in favor of the journal
 21 2011-01-28 00:44:32 <Kiba> why?
 22 2011-01-28 00:53:35 <OneFixt> ping
 23 2011-01-28 01:15:17 <AAA_awright> pong?
 24 2011-01-28 01:15:31 <Kiba> AAA_awright: why?
 25 2011-01-28 01:16:16 <AAA_awright> Kiba: To produce something with journal-level quality
 26 2011-01-28 01:16:36 <Kiba> so you want to get rid of interview/review/etc
 27 2011-01-28 01:16:37 <Kiba> ?
 28 2011-01-28 01:16:39 <AAA_awright> I don't even care what it's about I just want to order authors around and edit stuff
 29 2011-01-28 01:16:55 <lfm> hehe
 30 2011-01-28 01:17:43 <lfm> seems you should give the time thing a chance
 31 2011-01-28 01:19:14 <Kiba> the bitcoin economy is so small and tiny ya know?
 32 2011-01-28 01:19:35 <Kiba> we maybe the world's fastest growing economy but we're also the smallest
 33 2011-01-28 01:19:48 <Kiba> may be*
 34 2011-01-28 01:20:23 <AAA_awright> Depends on what an economy is?
 35 2011-01-28 01:21:22 <Kiba> the bitcoin economy
 36 2011-01-28 01:21:35 <Kiba> encompassing all bitcoin accepting business and bitcoiners
 37 2011-01-28 01:32:33 <riush> ;)
 38 2011-01-28 01:35:49 <noagendamarket> Maybe you can start posting to witcoin and people can  support it by voting
 39 2011-01-28 01:36:20 <Kiba> noagendamarket: I have no idea how it works
 40 2011-01-28 01:36:26 <Kiba> please enlighten me how I can post content
 41 2011-01-28 01:36:43 <Kiba> also, I will  be delivering you the finished extension
 42 2011-01-28 01:36:47 <Kiba> with the key and stuff
 43 2011-01-28 01:38:29 <noagendamarket> Kiba just sign up and start posting
 44 2011-01-28 01:38:41 <Kiba> that's it, I am done
 45 2011-01-28 01:38:46 <Kiba> noagendamarket: I don't see the post button
 46 2011-01-28 01:38:59 <Kiba> ya know, how I do post new content?
 47 2011-01-28 01:39:06 <noagendamarket> just go to the category section
 48 2011-01-28 01:39:43 <noagendamarket> http://witticisms.witcoin.com/witticise
 49 2011-01-28 01:40:08 <Kiba> why do I need to have money to post?
 50 2011-01-28 01:40:29 <doublec> makes spam less profitable
 51 2011-01-28 01:40:32 <Kiba> also, I am not a graphic designer
 52 2011-01-28 01:40:50 <Kiba> ya see, you guys need to explain how it's all work
 53 2011-01-28 01:40:59 <Kiba> because the information on how it all works IS ABSENT!
 54 2011-01-28 01:44:02 <noagendamarket> it stops spam
 55 2011-01-28 01:44:26 <Kiba> noagendamarket: so..bitcointo good or no?
 56 2011-01-28 01:45:12 <noagendamarket> yes its good\n197097
 57 2011-01-28 01:47:09 <noagendamarket> address
 58 2011-01-28 01:48:42 <molecular> http://mining.bitcoin.cz/stats/ block history is too short, slush1
 59 2011-01-28 01:49:51 <noagendamarket> Kiba: sent
 60 2011-01-28 01:52:40 <Kiba> noagendamarket: post on http://www.bitcoin.org/smf/index.php?topic=2815.0 how great a job I am doing, ya?
 61 2011-01-28 02:12:13 <afed> http://www.hardocp.com/article/2011/01/24/epower_juice_box_450w_auxiliary_power_supply_review
 62 2011-01-28 02:12:19 <afed> anyone evaluated that
 63 2011-01-28 02:12:26 <afed> for powering excess GPUs?
 64 2011-01-28 02:12:57 <afed> n/m i see it failed their testing
 65 2011-01-28 02:13:06 <afed> considered unsafe to connect to an expensive GPU
 66 2011-01-28 02:35:32 <molecular> slush just hit 30ghps
 67 2011-01-28 02:47:25 <doublec> it's finding a fair number of the blocks
 68 2011-01-28 02:47:59 <doublec> 104932,934,935,936,938,940,942 - nice run there
 69 2011-01-28 02:48:10 <andrew12> ...shares?
 70 2011-01-28 02:48:28 <molecular> blocks!
 71 2011-01-28 02:48:31 <andrew12> also does anyone know if its possible to run bitcoin over i2p?
 72 2011-01-28 02:48:35 <andrew12> ohhh
 73 2011-01-28 02:48:35 <molecular> how did you do that??
 74 2011-01-28 02:48:42 <andrew12> those are the block numbers.
 75 2011-01-28 02:49:15 <molecular> doublec, what is finding a fair amount of blocks
 76 2011-01-28 02:49:24 <andrew12> molecular: slush's pool
 77 2011-01-28 02:49:36 <doublec> molecular: I'm saying that slush's pool had a nice run of finding the blocks in the bitcoin network
 78 2011-01-28 02:50:11 <molecular> yeah
 79 2011-01-28 02:50:14 <molecular> true
 80 2011-01-28 02:51:01 <doublec> he can't be too far off being able to exploit that somehow if he was evil
 81 2011-01-28 02:51:15 <andrew12> hehe
 82 2011-01-28 02:51:21 <molecular> slush is not evil
 83 2011-01-28 02:51:26 <molecular> yet
 84 2011-01-28 02:51:47 <andrew12> wait till he owns most of the network, he'll use all of this to make a new block chain! muahahaha
 85 2011-01-28 02:51:56 <molecular> and I don't think he will become evail
 86 2011-01-28 02:52:09 <molecular> artforz will kick slush's ass pretty soon
 87 2011-01-28 02:52:12 <andrew12> >evail
 88 2011-01-28 02:52:51 <molecular> evil
 89 2011-01-28 02:53:14 <molecular> becoming eval makes only sense if you want to destroy the currenc
 90 2011-01-28 02:53:15 <molecular> y
 91 2011-01-28 02:53:52 <citizen> does a gpu's memorize size come into play at all?
 92 2011-01-28 02:54:05 <citizen> on the big ATI cards 1gB vs 2gB for example
 93 2011-01-28 02:54:11 <citizen> im guessing it doesn't, but i just want to verify
 94 2011-01-28 02:54:46 <molecular> memory is irrelevant
 95 2011-01-28 02:55:13 <molecular> even memory speed is irrelevant
 96 2011-01-28 02:56:20 <gribble> Error: "bc,difficulty" is not a valid command.
 97 2011-01-28 02:56:20 <molecular> ;;bc,difficulty
 98 2011-01-28 02:56:40 <andrew12> ;;bc,stats
 99 2011-01-28 02:56:43 <gribble> Current Blocks: 104957 | Current Difficulty: 22012.4941572 | Next Difficulty At Block: 106847 | Next Difficulty In: 1890 blocks | Next Difficulty In About: 1 week, 5 days, 9 hours, 9 minutes, and 0 seconds | Next Difficulty Estimate: 23521.07777741
100 2011-01-28 02:56:59 <molecular> ;;bc,stats
101 2011-01-28 02:57:01 <gribble> Current Blocks: 104957 | Current Difficulty: 22012.4941572 | Next Difficulty At Block: 106847 | Next Difficulty In: 1890 blocks | Next Difficulty In About: 1 week, 5 days, 9 hours, 9 minutes, and 0 seconds | Next Difficulty Estimate: 23521.07777741
102 2011-01-28 02:57:52 <molecular> that's why my shares/s rate dropped
103 2011-01-28 02:58:22 <andrew12> ;;bc,poolstats
104 2011-01-28 02:58:24 <gribble> {"round_probability": 5.4400000000000004, "hashes_ps": 30918740177, "shares": 1231, "active_workers": 370, "round_duration": "0:02:51", "round_started": "2011-01-28 03:55:32", "getwork_ps": 116}
105 2011-01-28 02:58:39 <andrew12> night
106 2011-01-28 03:00:13 <andrew12> ;;bc,poolstats
107 2011-01-28 03:00:23 <gribble> {"round_probability": 8.8200000000000003, "hashes_ps": 30838773984, "shares": 2032, "active_workers": 369, "round_duration": "0:04:43", "round_started": "2011-01-28 03:55:32", "getwork_ps": 115}
108 2011-01-28 03:00:36 <andrew12> just started the miner on my desktop
109 2011-01-28 03:01:15 <luke-jr> lol
110 2011-01-28 03:02:22 <andrew12> funny because the active_workers went down
111 2011-01-28 03:02:36 <andrew12> ok nini
112 2011-01-28 03:03:44 <Keefe> pci-e bandwidth is also irrelevant (or almost so)
113 2011-01-28 03:04:10 <Keefe> i have some of my 5970's connected to the x1 slots using modded risers
114 2011-01-28 03:04:25 <hacim> hm, no block found for many days beyond average
115 2011-01-28 03:05:22 <luke-jr> hacim: difficulty just went up 25%
116 2011-01-28 03:05:25 <Cusipzzz> hacim: rigged, imo :)
117 2011-01-28 03:05:26 <andrew12^mac> hm, no block found *ever*
118 2011-01-28 03:05:37 <hacim> 25% is quite a jump in difficulty
119 2011-01-28 03:07:09 <Cusipzzz> too many people jumping on the mining gravy train
120 2011-01-28 03:09:50 <doublec> not enough on the spending gravy train
121 2011-01-28 03:10:11 <doublec> by which I really mean "giving to doublec" gravy train
122 2011-01-28 03:17:01 <dirtyfilthy> is there some kind of equation that relates difficulty to total approx network hashes a second? i mean, i guess there is, but does anybody know it?
123 2011-01-28 03:18:23 <Cusipzzz> yes, and yes. who, i'm not sure :)
124 2011-01-28 03:19:20 <andrew12^droid> Haha.
125 2011-01-28 03:35:51 <hacim> since it has been far more than 25% time, i'm more worried that the new miner code isn't working :)
126 2011-01-28 03:48:37 <Diablo-D3> what did you break now?
127 2011-01-28 03:51:23 <Keefe> only 25% and you're thinking it's not working??
128 2011-01-28 03:51:46 <luke-jr> far more != only
129 2011-01-28 03:52:28 <Keefe> if he said far more than 90% time, i'd agree the code should be checked
130 2011-01-28 04:46:48 <nanotube> OneFixt: no comments don't show up in the blockchain
131 2011-01-28 04:47:54 <Kiba> hey nanotube
132 2011-01-28 04:48:00 <nanotube> heya Kiba :)
133 2011-01-28 04:48:12 <Kiba> I kick butt this month
134 2011-01-28 04:48:42 <nanotube> oh? made a bunch of btc? :)
135 2011-01-28 04:50:15 <Kiba> oh yeah, I think I have 200 ish btc at the beginning of the month
136 2011-01-28 04:50:20 <Kiba> now I have 600 BTC
137 2011-01-28 04:50:53 <ducki2p> whats your goal?
138 2011-01-28 04:51:00 <Kiba> along the way, I acquired a steady revenue source that work for me 24/7 days a week
139 2011-01-28 04:51:19 <Kiba> ducki2p: prove to my family that I am not a bum
140 2011-01-28 04:51:41 <nanotube> Kiba: nice work :)
141 2011-01-28 04:51:59 <Kiba> well, the larger goal is to establish independence
142 2011-01-28 04:52:27 <nanotube> just wait until 1btc= 1000usd. heh.
143 2011-01-28 04:52:41 <Kiba> that takes forever
144 2011-01-28 04:52:44 <Kiba> that's like...years
145 2011-01-28 04:56:15 <ducki2p> Kiba: have you considered being even more 'aggressive' and directly approach bitcoin merchants and forum posters and tell them what you can do exactly for them?
146 2011-01-28 04:56:22 <OneFixt> nanotube:thanks
147 2011-01-28 04:56:34 <Kiba> I have not really thought about that
148 2011-01-28 04:56:41 <Kiba> I got several job offers already
149 2011-01-28 04:56:57 <ducki2p> nice lineup!
150 2011-01-28 04:57:50 <Kiba> they are leads only, not maturing yet
151 2011-01-28 04:57:59 <ducki2p> eq, I am shocked to see top posters without a forum avatar
152 2011-01-28 04:59:10 <Kiba> like me?
153 2011-01-28 04:59:15 <ducki2p> mail them with "you need an avatar and I have XYZ in mind for you, and it will cost you X"
154 2011-01-28 05:00:13 <ducki2p> :)
155 2011-01-28 05:50:32 <Diablo-D3> http://www.youtube.com/watch?v=bRaWoN6lRiI
156 2011-01-28 05:50:36 <Diablo-D3> this is a very good fucking movie
157 2011-01-28 06:11:40 <MT`AwAy> wow
158 2011-01-28 06:11:43 <MT`AwAy> http://www.renesys.com/blog/2011/01/egypt-leaves-the-internet.shtml
159 2011-01-28 06:11:48 <MT`AwAy> did we have bitcoin users in egypt ?
160 2011-01-28 06:11:49 <Diablo-D3> MT`AwAy: yeah
161 2011-01-28 06:11:52 <Diablo-D3> and also, yes
162 2011-01-28 06:11:55 <Diablo-D3> I believe we did
163 2011-01-28 06:12:12 <MT`AwAy> would be a great way to see some large scale chain re-merging
164 2011-01-28 06:12:21 <Diablo-D3> Ive heard rumors from this channel that there are a handful of bitcoin users in arab nations
165 2011-01-28 06:12:26 <MT`AwAy> assuming people in egypt have enough cpu power to handle current difficulty
166 2011-01-28 06:12:34 <Diablo-D3> so I imagine we have at least one or two egyptian friends.
167 2011-01-28 06:23:22 <echelon> "Your entire order can't be filled at that price."
168 2011-01-28 06:23:28 <echelon> what does that mean
169 2011-01-28 06:23:49 <sgornick> echelon: On Mt. Gox?
170 2011-01-28 06:24:06 <echelon> ya
171 2011-01-28 06:25:40 <sgornick> means even though there might have been some bitcoins available at a certain price, you didn't give a high enough bid  (if buying) or a low enough ask (if selling) such that part of your order was left unfilled.
172 2011-01-28 06:26:51 <echelon> oh
173 2011-01-28 06:27:56 <sgornick> There are two good views of prices: http://www.taters.net/cgi-bin/btc/matrix.pl?axisinc=0.0025   and
174 2011-01-28 06:28:32 <sgornick> http://bitcoincharts.com/markets/mtgoxUSD.html  (assuming you are looking at Mt. Gox trades.)
175 2011-01-28 06:29:17 <echelon> ok thanks :)
176 2011-01-28 06:31:43 <echelon> so i have to wait until someone's offer gets low enough to mine?
177 2011-01-28 06:36:38 <sgornick> right, ... the lowest ask right now on Mt Gox is $0.43, so if you are only willing to pay 0.42 something ..., nobody willing to sell that low.  You can always cancel the remaining unfilled part of the order and place a new order at a higher amount.
178 2011-01-28 07:39:31 <jaromil> win 53
179 2011-01-28 07:39:35 <jaromil> ops sry
180 2011-01-28 09:52:25 <Diablo-D3> ArtForzZz: hey
181 2011-01-28 09:52:52 <Diablo-D3> ArtForzZz: someone is reporting a 15 mhash boost on their gtx460 with your kernel... which is a 68 mhash part.
182 2011-01-28 09:54:26 <Diablo-D3> nvidia's compiler must be absolute shit
183 2011-01-28 09:58:25 <zygf> is stream sdk 2.1 still the preferred version?
184 2011-01-28 09:59:01 <Diablo-D3> yes
185 2011-01-28 10:02:17 <jwalck> ;;bc,stats
186 2011-01-28 10:02:19 <gribble> Current Blocks: 105002 | Current Difficulty: 22012.4941572 | Next Difficulty At Block: 106847 | Next Difficulty In: 1845 blocks | Next Difficulty In About: 1 week, 5 days, 0 hours, 1 minute, and 30 seconds | Next Difficulty Estimate: 23660.11160997
187 2011-01-28 10:02:30 <jwalck> 105k blocks, thats 1/4 of all coins right?
188 2011-01-28 10:09:25 <andrew12^droid> Damn, we have school today.
189 2011-01-28 10:20:20 <andrew12^droid> *crickets*
190 2011-01-28 10:28:12 <andrew12^droid> ;;bc,stats
191 2011-01-28 10:28:14 <gribble> Current Blocks: 105004 | Current Difficulty: 22012.4941572 | Next Difficulty At Block: 106847 | Next Difficulty In: 1843 blocks | Next Difficulty In About: 1 week, 5 days, 0 hours, 13 minutes, and 29 seconds | Next Difficulty Estimate: 23598.87997538
192 2011-01-28 10:28:16 <andrew12^droid> ;;bc,poolstats
193 2011-01-28 10:28:17 <gribble> {"round_probability": 93.0, "hashes_ps": 30244091244, "shares": 58524, "active_workers": 356, "round_duration": "2:18:31", "round_started": "2011-01-28 09:09:45", "getwork_ps": 116}
194 2011-01-28 10:45:31 <andrew12^droid> Ding!
195 2011-01-28 10:54:07 <andrew12^droid> Noone?
196 2011-01-28 11:02:20 <sipa> no, sorry
197 2011-01-28 11:02:51 <andrew12^droid> Lol
198 2011-01-28 11:34:31 <andrew12^droid> Sneaky.
199 2011-01-28 12:25:18 <comboy> what temperature do you get with stock cooling on non-oveclocked 5970 full usage?
200 2011-01-28 12:28:22 <tcatm> 52..53 ???C on my 2x 5970 miner @ 800 MHz
201 2011-01-28 12:28:49 <Diablo-D3> comboy: thats dependent on your case's airflow
202 2011-01-28 12:28:54 <sipa> fans at how much, tcatm?
203 2011-01-28 12:29:01 <tcatm> 100%
204 2011-01-28 12:42:12 <cosurgi> damn. no block for 5 days, and my average time is 1d10h,  770 Mh.
205 2011-01-28 12:42:42 <cosurgi> I will wait a bit, then switch to pool :/ And lose 5 days.
206 2011-01-28 12:43:27 <sipa> there is no such thing as losing 5 days
207 2011-01-28 12:43:42 <sipa> your expected time to find a block doesn't change
208 2011-01-28 12:44:14 <sipa> your average time to find a block is 1d10h from *now* , even if your previous block was a week ago
209 2011-01-28 12:47:22 <cosurgi> tell me more :)
210 2011-01-28 12:47:35 <comboy> Diablo-D3: sorry connection issues, well currently it's running aut of the box, just this one card so it's not that it is inside some warm non-moving air
211 2011-01-28 12:47:58 <comboy> Diablo-D3: what temp do you get in case with nice airflow?
212 2011-01-28 12:48:13 <sipa> cosurgi: it's called statistical independency
213 2011-01-28 12:48:23 <sipa> cosurgi: every has you try has the same chance of being a good one
214 2011-01-28 12:48:26 <Diablo-D3> comboy: I dont have one, but you can keep them down below 60
215 2011-01-28 12:48:28 <sipa> *hash
216 2011-01-28 12:48:37 <Diablo-D3> comboy: you're fine as long as its below 80.
217 2011-01-28 12:49:18 <cosurgi> ok, but 95% in 4 days, 6 hours, 10 minutes.
218 2011-01-28 12:49:48 <cosurgi> so I'm now close to 98% probability of finding a block. And still not happening.
219 2011-01-28 12:50:14 <luke-jr> &
220 2011-01-28 12:50:39 <luke-jr> probabilities don't work like that :P
221 2011-01-28 12:51:06 <comboy> Diablo-D3: then I guess I'm not, 87 currently still rising slowly, I've bought used one :/
222 2011-01-28 12:51:14 <sipa> cosurgi: you have 95% chance to find at least one block in *any* 4 days, 6 hours, 10 minutes
223 2011-01-28 12:51:18 <luke-jr> the only factor influencing the probability is the difficulty :P
224 2011-01-28 12:51:25 <tcatm> comboy: both cores?
225 2011-01-28 12:51:31 <comboy> yeah
226 2011-01-28 12:51:31 <sipa> cosurgi: but every minute you don't find a hash, you lost that minute
227 2011-01-28 12:51:38 <luke-jr> the time since your last block is irrelevant
228 2011-01-28 12:51:40 <tcatm> comboy: what fanspeed?
229 2011-01-28 12:51:46 <comboy> tcatm: ok no I dont know if both
230 2011-01-28 12:51:51 <comboy> I just use aticonfig --odgt
231 2011-01-28 12:52:00 <tcatm> aticonfig --adapter=all --odgt
232 2011-01-28 12:52:18 <comboy> oh, now I get two of them, thanks
233 2011-01-28 12:52:20 <luke-jr> what Linux GPU miner can someone recommend?
234 2011-01-28 12:52:22 <comboy> so 85 and 87 :/
235 2011-01-28 12:52:36 <tcatm> luke-jr: both, poclbm and Diablo-D3's
236 2011-01-28 12:52:50 <luke-jr> tcatm: can Diablo-D3's be compiled for Linux, rather than JVM?
237 2011-01-28 12:52:55 <tcatm> comboy: DISPLAY=:0.0 aticonfig --pplib-cmd 'get fanspeed 0'
238 2011-01-28 12:53:27 <sipa> luke-jr: you mean to binary code?
239 2011-01-28 12:53:28 <comboy> 44%
240 2011-01-28 12:53:36 <luke-jr> sipa: yes, I have no intention of installing a JVM
241 2011-01-28 12:53:36 <sipa> instead of java bytecode?
242 2011-01-28 12:53:40 <comboy> I wonder why it won't go faster if it gets so hot
243 2011-01-28 12:53:42 <tcatm> comboy: DISPLAY=:0.0 aticonfig --pplib-cmd 'set fanspeed 0 100'
244 2011-01-28 12:54:03 <sipa> luke-jr: it's probably a lot easier to run poclbm then, i think
245 2011-01-28 12:54:19 <comboy> tcatm: oh yeah, I like heavy music
246 2011-01-28 12:54:23 <luke-jr> did poclbm integrate ArtForzZz's code?
247 2011-01-28 12:54:26 <sipa> yes
248 2011-01-28 12:54:29 <luke-jr> cool
249 2011-01-28 12:59:51 <comboy> sorry I went away, my system hanged again :/ I wonter if that may be power adapter, but I have some 80+ bronze 750W, thought it's enough...
250 2011-01-28 13:01:03 <comboy> anyway what could be the reason ATI is not adjusting fanspeed according to temp? is it not using some feedbak loop to this just hardcoded speeds for temperatures ?
251 2011-01-28 13:01:42 <comboy> tcatm: and one more question, any place when I could find reference for all these nice commands you can send to card?
252 2011-01-28 13:02:00 <sipa> cosurgi: doesn't it change at all, or not enough?
253 2011-01-28 13:02:01 <tcatm> I only know of the set/get fanspeed command
254 2011-01-28 13:02:23 <sipa> typically it does adjust, but only slowly, and never to 100%
255 2011-01-28 13:02:43 <cosurgi> sipa: what?
256 2011-01-28 13:03:02 <cosurgi> oh, you're talking to comboy I guess.
257 2011-01-28 13:03:04 <sipa> cosurgi: sorry
258 2011-01-28 13:03:06 <tcatm> comboy: The typical symptom of a too small PSU is sudden power off and refusing to turn on again...
259 2011-01-28 13:03:12 <comboy> sipa, so when it adjusts, do you know what temperature is it trying to achive?
260 2011-01-28 13:03:41 <cosurgi> ok, I switched to pool now.
261 2011-01-28 13:03:44 <comboy> tcatm: yeah I went through that with my previous PSU on 5870, system was unstable on weaker PSU
262 2011-01-28 13:04:04 <cosurgi> And I bet, that I will find a block or two in the pool, today or tomorrow. heh :>
263 2011-01-28 13:05:00 <cosurgi> Diablo-D3: do you know how to assign a 128bit variable in opencl? like: ulong long x=0x12345678123456781234567812345678; ?
264 2011-01-28 13:05:14 <cosurgi> Diablo-D3: compiler complains about too bit constant.
265 2011-01-28 13:05:32 <cosurgi> ulong long x=(ulong long)(0x12345678123456781234567812345678); isn't helping.
266 2011-01-28 13:05:39 <cosurgi> *too big.
267 2011-01-28 13:05:49 <Diablo-D3> cosurgi: you cant.
268 2011-01-28 13:06:06 <Diablo-D3> there isnt a 128 bit type in opencl
269 2011-01-28 13:06:12 <comboy> tcatm: but 750W bronze 80+ ain't enough for that one card? :/
270 2011-01-28 13:06:17 <Diablo-D3> cosurgi: why do you want one?
271 2011-01-28 13:06:46 <cosurgi> I have found ulong long in the 1.1 specification, it is 128bit. I want to try vectorizing sha-256.
272 2011-01-28 13:07:01 <comboy> and it's weird my otherwise very stable system hangss randomly even when this card is not in use :/
273 2011-01-28 13:07:28 <cosurgi> how about some reinterpret cast? But I didn't found a working one yet. Like this: ulong long x=as_ulong_long((uint4)(0x12345678,0x12345678,0x12345678,0x12345678));
274 2011-01-28 13:10:01 <Diablo-D3> cosurgi: erm
275 2011-01-28 13:10:06 <cosurgi> yes?
276 2011-01-28 13:10:07 <Diablo-D3> thats 1.1
277 2011-01-28 13:10:13 <Diablo-D3> you require sdk 2.2 and up
278 2011-01-28 13:10:14 <Diablo-D3> AND
279 2011-01-28 13:10:20 <Diablo-D3> the hardware doesnt support such types anyhow
280 2011-01-28 13:10:24 <cosurgi> oh.
281 2011-01-28 13:10:31 <cosurgi> but why is it compiling?
282 2011-01-28 13:10:31 <Diablo-D3> so what are you using it for?
283 2011-01-28 13:10:38 <sipa> and sha256 only uses 32-bit arithmetic
284 2011-01-28 13:11:19 <cosurgi> I want to try vectorizing sha-256. Doing xors and ors on four 32 bit numbers at a single operation.
285 2011-01-28 13:11:56 <Diablo-D3> you're an idiot
286 2011-01-28 13:11:58 <cosurgi> there's one trick in addition involved, and another trick with rotating. But a single run of sha-256 woulc calculate four hashes.
287 2011-01-28 13:12:08 <Diablo-D3> cosurgi: opencl already has vectorized types!
288 2011-01-28 13:12:24 <Diablo-D3> not only that
289 2011-01-28 13:12:32 <Diablo-D3> I already had that in my miner and removed it
290 2011-01-28 13:12:37 <cosurgi> I'm listening.
291 2011-01-28 13:12:48 <Diablo-D3> AMD's impl already auto vectorizes correctly.
292 2011-01-28 13:12:51 <sipa> cosurgi: you know about uint4, why not use that?
293 2011-01-28 13:13:41 <cosurgi> sipa: will have to try. I just need to make sure that xor and or works corerctly on uint4.
294 2011-01-28 13:14:04 <cosurgi> I mean - operates on all components.
295 2011-01-28 13:14:08 <Diablo-D3> cosurgi: yes.
296 2011-01-28 13:14:16 <sipa> cosurgi: look at poclbm's code
297 2011-01-28 13:14:22 <Diablo-D3> vec1 operator vec4 does exactly what you think it does.
298 2011-01-28 13:14:22 <sipa> it already has uint2-based version
299 2011-01-28 13:14:23 <cosurgi> Diablo-D3: so you removed it - it wasn't faster?
300 2011-01-28 13:14:31 <sipa> s/already/still/
301 2011-01-28 13:14:41 <Diablo-D3> cosurgi: it was faster in exactly one case, and then AMD fixed that
302 2011-01-28 13:14:51 <cosurgi> which case?
303 2011-01-28 13:15:01 <Diablo-D3> radeon 4xxx, older drivers wernt autovec correctly
304 2011-01-28 13:15:09 <Diablo-D3> they fixed that around 10.9
305 2011-01-28 13:16:10 <cosurgi> but, "fixed" as in - it was faster but calculating wrong, or "fixed" and we want to use older drivers, so it's faster?
306 2011-01-28 13:16:34 <Diablo-D3> before they fixed it, it chose vec1 on 4xxx, the correct is vec2
307 2011-01-28 13:16:37 <Diablo-D3> it now chooses vec2
308 2011-01-28 13:17:09 <sipa> cosurgi: fixed as in: the autovectorizer now does a better job than a programming hardcoding vectorization
309 2011-01-28 13:17:16 <sipa> *programmer
310 2011-01-28 13:17:24 <cosurgi> I see...
311 2011-01-28 13:17:26 <Diablo-D3> sipa: mostly
312 2011-01-28 13:17:39 <Diablo-D3> I saw the same exact performance as vec2 as autovec on radeon 5xxx to begin with
313 2011-01-28 13:17:48 <Diablo-D3> it was 4xxx that was doing it wrong, and it does it right now
314 2011-01-28 13:17:51 <Diablo-D3> so I removed the code
315 2011-01-28 13:18:09 <sipa> on my 4870 vec2 seems slower than no vec
316 2011-01-28 13:18:13 <Diablo-D3> and this did nothing on geforces anyhow, since they're not vectorized hardware
317 2011-01-28 13:18:18 <Diablo-D3> sipa: it does now
318 2011-01-28 13:18:25 <Diablo-D3> the new drivers fight stupid vec2 usage
319 2011-01-28 13:18:36 <Diablo-D3> vectors wernt meant to be used to SIMD shit
320 2011-01-28 13:18:43 <Diablo-D3> they were meant to hold tuples
321 2011-01-28 13:20:21 <Diablo-D3> the compiler and driver is pretty intelligent on how to properly fill the 4 ALUs wrt register usage
322 2011-01-28 13:21:01 <Diablo-D3> vec2 is the proper balance, and it works right
323 2011-01-28 13:21:19 <cosurgi> vec2 are two 32bit numbers?
324 2011-01-28 13:21:39 <Diablo-D3> literally? two 32 bit floats
325 2011-01-28 13:21:45 <Diablo-D3> but I mean two wide vectorization
326 2011-01-28 13:22:01 <cosurgi> ok. And you removed the code based on vec2 ?
327 2011-01-28 13:22:05 <Diablo-D3> yes
328 2011-01-28 13:22:16 <Diablo-D3> it performed worse after AMD fixed radeon 4xxx.
329 2011-01-28 13:22:32 <cosurgi> and a single 32bit become faster
330 2011-01-28 13:22:41 <Diablo-D3> no
331 2011-01-28 13:22:49 <Diablo-D3> the driver autovectorizes it
332 2011-01-28 13:22:58 <Diablo-D3> internally, its doing vec2 correctly
333 2011-01-28 13:23:06 <Diablo-D3> forcing it using a uint2 was a bad idea.
334 2011-01-28 13:23:18 <cosurgi> ok...
335 2011-01-28 13:23:39 <Diablo-D3> like I said, this is not how you expose SIMD usage in opencl
336 2011-01-28 13:24:00 <Diablo-D3> the registers are 4xxx/5xxx are 32bit*4
337 2011-01-28 13:24:01 <cosurgi> but the loop is outside of the kernel. How it can be autovectorized, if a single call is with just one nonce?
338 2011-01-28 13:24:17 <Diablo-D3> cosurgi: because it just runs multiple loops.
339 2011-01-28 13:24:37 <Diablo-D3> the whole kernel execution state is SIMD'ed.
340 2011-01-28 13:24:50 <cosurgi> so four loops are ran sunchronized? With four different nonces, and the driver takes care of it?
341 2011-01-28 13:25:00 <Diablo-D3> four different everythinged
342 2011-01-28 13:25:00 <sipa> cosurgi: probably a lot more
343 2011-01-28 13:25:06 <Diablo-D3> sipa: no, less
344 2011-01-28 13:25:08 <Diablo-D3> its doing 2
345 2011-01-28 13:25:14 <sipa> only two? wow
346 2011-01-28 13:25:22 <Diablo-D3> runs out of registers first
347 2011-01-28 13:25:36 <Diablo-D3> and its optimum anyhow, enough of sha256 can be ran superscalar
348 2011-01-28 13:25:53 <sipa> not two per stream processor?
349 2011-01-28 13:25:54 <Diablo-D3> so you're executing multiple things at once even in a single threaded context
350 2011-01-28 13:25:59 <Diablo-D3> sipa: hell no.
351 2011-01-28 13:26:33 <Diablo-D3> if you have a radeon 5xxx with, say, 800 pipes, you're running 1600 kernel executions at once
352 2011-01-28 13:26:49 <Diablo-D3> (or 800, or 2400, or whatever is optimum for that kernel)
353 2011-01-28 13:26:49 <sipa> ok, so two per stream processor
354 2011-01-28 13:27:05 <Diablo-D3> sipa: last time I noticed, stream processor referred to the compute unit
355 2011-01-28 13:27:11 <Diablo-D3> which you'd have, say, 24 or whatever of
356 2011-01-28 13:27:15 <sipa> hmm ok
357 2011-01-28 13:27:22 <cosurgi> ok, thanks.
358 2011-01-28 13:27:24 <sipa> my terminology can be wrong
359 2011-01-28 13:27:29 <Diablo-D3> cosurgi: btw
360 2011-01-28 13:27:36 <cosurgi> yes? :)
361 2011-01-28 13:27:37 <Diablo-D3> the interesting part with that vec4 register is
362 2011-01-28 13:27:43 <Diablo-D3> vec1 usage uses only a fourth of it
363 2011-01-28 13:27:58 <sipa> cosurgi: anyway, the host programs calls opencl to run a single kernel on a large number of different inputs at once
364 2011-01-28 13:28:09 <Diablo-D3> on a properly working driver, trying to vec2 your kernel stupidly means you have register read/write contention
365 2011-01-28 13:28:10 <sipa> and therefore it can vectorize
366 2011-01-28 13:28:20 <Diablo-D3> since the registers belong to the CU, not the pipes
367 2011-01-28 13:29:23 <mtgox> hi
368 2011-01-28 13:29:27 <cosurgi> ok, so it will be slower...
369 2011-01-28 13:29:29 <cosurgi> interesting.
370 2011-01-28 13:29:32 <Diablo-D3> cosurgi: yes.
371 2011-01-28 13:29:39 <Diablo-D3> cosurgi: the compiler is smart enough to use the hardware.
372 2011-01-28 13:32:55 <luke-jr> .
373 2011-01-28 13:34:44 <Kiba> yo
374 2011-01-28 13:35:08 <Diablo-D3> cosurgi: this is a highly parallel lockstep arch
375 2011-01-28 13:35:37 <Diablo-D3> cosurgi: each group of 4 (well, 5, but the T unit isnt too useful here) ALUs work on 20 items at a time
376 2011-01-28 13:36:36 <ArtForzZz> errr... not quite
377 2011-01-28 13:36:51 <Diablo-D3> or was it 16
378 2011-01-28 13:36:54 <ArtForzZz> 16
379 2011-01-28 13:37:04 <Diablo-D3> yeah its 16
380 2011-01-28 13:37:12 <ArtForzZz> each CU is 16 * VLIW5 SIMD
381 2011-01-28 13:38:01 <ArtForzZz> 5870 has 20CU
382 2011-01-28 13:39:47 <Diablo-D3> you mean
383 2011-01-28 13:39:49 <Diablo-D3> over 9000
384 2011-01-28 13:40:20 <ArtForzZz> and now ... why is multiple of 64 localworksize optimal?
385 2011-01-28 13:40:33 <Diablo-D3> because AMD said so.
386 2011-01-28 13:40:43 <ArtForzZz> well, benchmarks also show it
387 2011-01-28 13:40:57 <Diablo-D3> its the wavefront size for the CU
388 2011-01-28 13:41:02 <ArtForzZz> yep
389 2011-01-28 13:41:09 <Diablo-D3> but it still boils down to
390 2011-01-28 13:41:11 <Diablo-D3> AMD said so.
391 2011-01-28 13:41:35 <ArtForzZz> the whole arch is really quite complicated
392 2011-01-28 13:41:52 <Diablo-D3> ArtForzZz: cosurgi wanted to reintroduece vec2 use through long longs.
393 2011-01-28 13:42:02 <ArtForzZz> meh, wont really help
394 2011-01-28 13:42:08 <Diablo-D3> I dont know what they're putting in the school lunchrooms anymore, but whatever it is, its nasty shit
395 2011-01-28 13:42:23 <cosurgi> actually I was thinking about 128bit, so that would be vec4
396 2011-01-28 13:42:25 <ArtForzZz> you're hitting the register limit with vec2s
397 2011-01-28 13:43:40 <Diablo-D3> yeah which is what I said beofre
398 2011-01-28 13:43:50 <Diablo-D3> and I also said trying to force SIMD using vecs is wrong
399 2011-01-28 13:44:00 <cosurgi> ... and if not enough registers it starts to use slow GPU memory
400 2011-01-28 13:44:00 <Diablo-D3> vecs are meant to hold tuples, not multiple work items
401 2011-01-28 13:44:21 <ArtForzZz> yup
402 2011-01-28 13:44:23 <Diablo-D3> and cramming multiple unrelated work items into a single register causes contention
403 2011-01-28 13:44:25 <Diablo-D3> which is slow
404 2011-01-28 13:44:31 <Diablo-D3> and it was a stupid fucking hack to begin with
405 2011-01-28 13:44:44 <ArtForzZz> iirc with localworksize 64 you get 64 4*32-bit registers
406 2011-01-28 13:45:09 <ArtForzZz> thanks to load/store restrictions thats not really much
407 2011-01-28 13:45:09 <Diablo-D3> ArtForzZz: you know, Im wondering if thats even right
408 2011-01-28 13:45:19 <Diablo-D3> the 16384 seems to imply thats 128 bit registers
409 2011-01-28 13:45:30 <ArtForzZz> 16384 32 bit registers
410 2011-01-28 13:45:33 <Diablo-D3> not 4096 128
411 2011-01-28 13:45:40 <Diablo-D3> s/the/they/
412 2011-01-28 13:45:49 <luke-jr> davout: so what is this for? I don't see it documented& :prompt => "--"
413 2011-01-28 13:45:50 <ArtForzZz> = really 4096 * 128
414 2011-01-28 13:45:55 <Diablo-D3> hrm
415 2011-01-28 13:45:59 <Diablo-D3> AMD should make that more clear
416 2011-01-28 13:46:06 <Diablo-D3> because they're not addressed that way
417 2011-01-28 13:46:14 <Diablo-D3> they're in groups of 4*32
418 2011-01-28 13:46:18 <ArtForzZz> yep
419 2011-01-28 13:46:23 <Diablo-D3> putting a vec1 in a register uses all 4 slots
420 2011-01-28 13:46:42 <ArtForzZz> well, the compiler is pretty good at using partial registers
421 2011-01-28 13:46:56 <Diablo-D3> yeah, if it can guarantee there isnt conflicts
422 2011-01-28 13:47:00 <ArtForzZz> otherwise even vec1 would be a damn tight fit
423 2011-01-28 13:47:15 <Diablo-D3> sha256 seems to only be 2x superscalar
424 2011-01-28 13:47:22 <ArtForzZz> ?
425 2011-01-28 13:47:35 <cosurgi> and what is the local workgroup size in Diablo's miner?
426 2011-01-28 13:47:37 <Diablo-D3> as in, you're only executing two things at once inside a normal execution
427 2011-01-28 13:47:54 <cosurgi> that is the -w 64 parameter ?
428 2011-01-28 13:47:57 <sipa> yes
429 2011-01-28 13:47:59 <ArtForz> things?
430 2011-01-28 13:47:59 <cosurgi> oh, right. of course.
431 2011-01-28 13:48:06 <Diablo-D3> cosurgi: yes
432 2011-01-28 13:48:09 <Diablo-D3> ArtForz: operations
433 2011-01-28 13:48:21 <Diablo-D3> ArtForz: as in, if I ran this on a pentium, it'd happily fill both ALUs
434 2011-01-28 13:48:31 <ArtForz> yep
435 2011-01-28 13:48:31 <cosurgi> so with -w 256 I get 256 4*32 bit registers?
436 2011-01-28 13:48:39 <Diablo-D3> ArtForz: so, 2*2 works well here.
437 2011-01-28 13:48:42 <ArtForz> well, actually check the generated ASM
438 2011-01-28 13:48:43 <Diablo-D3> cosurgi: no.
439 2011-01-28 13:49:08 <Diablo-D3> cosurgi: it executes one wavefront at a time... what multiples of 64 do is queue multiple wavefronts into the CU pipeline
440 2011-01-28 13:49:13 <ArtForz> seems it can make decent use of all 5 ALUs even with 2-wide autovectorization
441 2011-01-28 13:49:14 <Diablo-D3> cosurgi: which does NOT help us here
442 2011-01-28 13:49:41 <Diablo-D3> ArtForz: yeah, but what Im saying is, sha256 seems to be mostly all dual ALU usage normally
443 2011-01-28 13:49:54 <Diablo-D3> ArtForz: it wasnt designed for highly parallel calculation on a single sha256
444 2011-01-28 13:49:55 <ArtForz> yeah, pretty much
445 2011-01-28 13:49:57 <cosurgi> oh, ok. Se regardless of local workgroup the maximum number of registers is always 64 4*32-bit ?
446 2011-01-28 13:49:59 <ArtForz> nope
447 2011-01-28 13:50:02 <Diablo-D3> so 2*2 works well
448 2011-01-28 13:50:06 <davout> luke-jr: so, any success ?
449 2011-01-28 13:50:12 <Diablo-D3> cosurgi: thats how many is in the CU, 16384
450 2011-01-28 13:50:20 <ArtForz> sha256 was designed to be fast on a 2 or 3-wide superscalar 32 bit processor
451 2011-01-28 13:50:27 <cosurgi> ok.
452 2011-01-28 13:50:36 <Diablo-D3> cosurgi: if you had a kernel that used, say, 1 register, you could run a shitload in parallel if the hardware allowed it
453 2011-01-28 13:51:12 <Diablo-D3> cosurgi: to fit a 2x vectorized program in, it needs to use less than 32 registers.
454 2011-01-28 13:51:23 <cosurgi> so in fact, reducing the number of used registers in sha-256 is a way to go faster?
455 2011-01-28 13:51:29 <ArtForz> which it barely manages for sha256
456 2011-01-28 13:51:30 <Diablo-D3> cosurgi: "no"
457 2011-01-28 13:51:45 <Diablo-D3> cosurgi: you'd need to go the whole way down to less than 16 to fit 4 in
458 2011-01-28 13:51:50 <Diablo-D3> which is impossible
459 2011-01-28 13:51:55 <ArtForz> yep
460 2011-01-28 13:52:15 <luke-jr> davout: still waiting on you for an answer :P
461 2011-01-28 13:52:27 <ArtForz> well, possible, but slow as fuck because you get shitloads of data conflicts
462 2011-01-28 13:52:34 <luke-jr> davout: so what is this for? I don't see it documented& :prompt => "--"
463 2011-01-28 13:52:42 <Diablo-D3> yeah
464 2011-01-28 13:52:50 <Diablo-D3> which is why my forced vec4 usage fucked over shit badly
465 2011-01-28 13:52:56 <ArtForz> yep
466 2011-01-28 13:53:04 <Diablo-D3> hilariously so, too
467 2011-01-28 13:53:28 <Diablo-D3> at that time I was getting 72 on vec2 (before AMD fixed the driver), vec4 was like 50 something
468 2011-01-28 13:53:46 <Diablo-D3> and vec1 was upper 60s
469 2011-01-28 13:53:52 <cosurgi> and after the fix?
470 2011-01-28 13:53:58 <Diablo-D3> after the fix, I was nailing 75 all the time on vec1
471 2011-01-28 13:54:10 <cosurgi> I see :)
472 2011-01-28 13:54:34 <Diablo-D3> after the fix it was also matching the already existing radeon 5xxx behavior
473 2011-01-28 13:54:39 <ArtForz> IL compiler autovectorizing vec1 makes better use of units than vec2
474 2011-01-28 13:54:44 <Diablo-D3> yeah
475 2011-01-28 13:54:51 <Diablo-D3> because of that weird 5th unit
476 2011-01-28 13:54:55 <ArtForz> yep
477 2011-01-28 13:55:06 <ArtForz> who the fuck thought that was a good idea
478 2011-01-28 13:55:11 <Diablo-D3> well
479 2011-01-28 13:55:18 <Diablo-D3> I wouldnt have done what the 69xx did
480 2011-01-28 13:55:19 <ArtForz> 128-bit datapaths everywhere, let's use 5 32-bit units!
481 2011-01-28 13:55:36 <Diablo-D3> if I needed a T unit that badly, I would have just made 1 of the 4 be one
482 2011-01-28 13:55:53 <Diablo-D3> AMD must have seen a lot of T usage in common shit or something
483 2011-01-28 13:56:11 <ArtForz> well, T can also do quite a few "normal" ops
484 2011-01-28 13:56:17 <Diablo-D3> thats not what I mean
485 2011-01-28 13:56:26 <Diablo-D3> the 69xx ALUs are _all_ T units
486 2011-01-28 13:56:32 <Diablo-D3> thats just nonsensical
487 2011-01-28 13:56:35 <Diablo-D3> why make them all do it
488 2011-01-28 13:56:38 <ArtForz> ... not really
489 2011-01-28 13:56:45 <Diablo-D3> well, they can execute T unit shit
490 2011-01-28 13:56:53 <Diablo-D3> theres no reason to have a dedicated T unit anymore
491 2011-01-28 13:56:55 <cosurgi> what is 'T unit'?
492 2011-01-28 13:56:59 <cosurgi> some special ops ?
493 2011-01-28 13:57:04 <ArtForz> Transcendental
494 2011-01-28 13:57:11 <cosurgi> what it means?
495 2011-01-28 13:57:19 <Diablo-D3> it means it worships Jesus.
496 2011-01-28 13:57:20 <Diablo-D3> clearly.
497 2011-01-28 13:57:23 <ArtForz> trig functions, sqrt, double precision FP
498 2011-01-28 13:57:38 <cosurgi> hmm... useless :)
499 2011-01-28 13:57:40 <Diablo-D3> ArtForz: certain 32 bit int ops too
500 2011-01-28 13:57:46 <ArtForz> yep
501 2011-01-28 13:57:54 <Diablo-D3> cosurgi: actually, very useful
502 2011-01-28 13:58:08 <ArtForz> while other 32 bit int ops can only be done on the vector units...
503 2011-01-28 13:58:20 <ArtForz> looks like in 69xx xyz is the T unit
504 2011-01-28 13:58:28 <cosurgi> Diablo-D3: why useful? But you are both complaining on T units
505 2011-01-28 13:59:21 <Diablo-D3> cosurgi: it doesnt make sense to have a 5th unit where your pipeline is 4 units wide
506 2011-01-28 13:59:34 <cosurgi> obvioulsy.
507 2011-01-28 13:59:37 <Diablo-D3> it WOULD make sense to have a ALU that does both normal and T unit stuff
508 2011-01-28 13:59:47 <Diablo-D3> but it DOESNT make sense to have all four do it
509 2011-01-28 13:59:59 <cosurgi> ok
510 2011-01-28 14:00:18 <Diablo-D3> now, the biggest part of the circuity was being able to do doubles
511 2011-01-28 14:00:26 <ArtForz> yeah... I also dont quite get why they didnt simply merge 1 vector unit with T
512 2011-01-28 14:00:30 <Diablo-D3> 69xx does combo mode doubles
513 2011-01-28 14:00:36 <ArtForz> ?
514 2011-01-28 14:00:44 <Diablo-D3> as in, takes 2 alus to do one double op
515 2011-01-28 14:01:00 <ArtForz> I don't think so
516 2011-01-28 14:01:03 <Diablo-D3> ArtForz: it does
517 2011-01-28 14:01:07 <ArtForz> looks like it uses 3
518 2011-01-28 14:01:13 <Diablo-D3> it looks like 2
519 2011-01-28 14:01:19 <Diablo-D3> because its doing 1/2 of SP performance
520 2011-01-28 14:01:26 <ArtForz> where?
521 2011-01-28 14:01:36 <Diablo-D3> erm, doesnt 69xx do double at half the speed of single?
522 2011-01-28 14:01:40 <ArtForz> nope
523 2011-01-28 14:01:41 <ArtForz> 1/4
524 2011-01-28 14:01:44 <Diablo-D3> huh.
525 2011-01-28 14:01:47 <ArtForz> 58xx was 1/5
526 2011-01-28 14:01:47 <Diablo-D3> goddamnit AMD
527 2011-01-28 14:01:49 <Diablo-D3> quit fucking shit up.
528 2011-01-28 14:02:01 <Diablo-D3> so the only fucking thing they did
529 2011-01-28 14:02:09 <Diablo-D3> was add a few integer and useless ops to the ALus
530 2011-01-28 14:02:17 <Diablo-D3> and somehow fuck up double calculation
531 2011-01-28 14:02:21 <ArtForz> they intermingled T with XYZ
532 2011-01-28 14:02:29 <Diablo-D3> not even intermingled
533 2011-01-28 14:02:33 <ArtForz> you can do a T and a V op on XYZ + W
534 2011-01-28 14:02:34 <Diablo-D3> the most complex part of T was the doubles
535 2011-01-28 14:03:03 <cosurgi> ok, thank you guys, I gotta go now, see you later.
536 2011-01-28 14:03:13 <Diablo-D3> stuff like 32bit integer MADs and shit could easily be added to the XYZW units
537 2011-01-28 14:03:21 <Diablo-D3> since they already did 24
538 2011-01-28 14:03:28 <ArtForz> yep, integer shit is usually quite small
539 2011-01-28 14:03:37 <Diablo-D3> so maybe 69xx isnt as bad as I thought it was
540 2011-01-28 14:03:41 <ArtForz> at least compared to FP
541 2011-01-28 14:03:44 <Diablo-D3> 7xxx should be genius fucking work
542 2011-01-28 14:03:53 <x6763> is there any special order that a series of inputs or outputs of a transaction must follow?
543 2011-01-28 14:03:54 <Diablo-D3> cosurgi: seeya
544 2011-01-28 14:04:00 <ArtForz> and iirc some used-to-be-T ops are now V ops
545 2011-01-28 14:04:09 <Diablo-D3> ArtForz: yeah
546 2011-01-28 14:04:14 <Diablo-D3> ArtForz: btw, did you get a 69xx yet?
547 2011-01-28 14:04:18 <ArtForz> yup
548 2011-01-28 14:04:24 <ArtForz> 6970
549 2011-01-28 14:04:25 <Diablo-D3> how did it fair?
550 2011-01-28 14:04:42 <ArtForz> not too bad, a bit slower than 5870
551 2011-01-28 14:04:57 <ArtForz> about 2% slower
552 2011-01-28 14:05:06 <Diablo-D3> thats... not really bad at all
553 2011-01-28 14:05:11 <Diablo-D3> given you're stuck on a broken SDK
554 2011-01-28 14:05:29 <ArtForz> well, since I dont have to use OCL anymore... not really
555 2011-01-28 14:05:34 <Diablo-D3> heh
556 2011-01-28 14:05:44 <Diablo-D3> still
557 2011-01-28 14:05:50 <Diablo-D3> 2% might as well be ignored
558 2011-01-28 14:05:54 <ArtForz> yup
559 2011-01-28 14:06:04 <Diablo-D3> for typical apps, 69xx should be fast as fuck
560 2011-01-28 14:06:12 <x6763> i'm reading a transaction from blk0001.dat, but it seems the transaction inputs are all mixed up as compared to blockexplorer, and this is affecting the transaction hash, as well as the verification
561 2011-01-28 14:06:27 <ArtForz> well, it uses more power than 5870
562 2011-01-28 14:07:00 <ArtForz> so I kinda fear that 6990 will be even more crippled by the 300W limit than 5970
563 2011-01-28 14:07:02 <Diablo-D3> yeah, but it has a more 3D power
564 2011-01-28 14:07:24 <ArtForz> sure, it'll be faster for 3D mode, but we're a power virus :P
565 2011-01-28 14:07:28 <Diablo-D3> I mean, its a fucking video card, people do simple SP ops on it
566 2011-01-28 14:07:37 <Diablo-D3> it does this exceedingly well
567 2011-01-28 14:07:49 <Diablo-D3> the fact it also has a very powerful integer engine is just extra awesomery
568 2011-01-28 14:07:57 <ArtForz> yup
569 2011-01-28 14:07:57 <Diablo-D3> ArtForz: you know what I was thinking?
570 2011-01-28 14:08:13 <Diablo-D3> you know how powervr chips like you to use lower computation int ops when you can?
571 2011-01-28 14:08:20 <Diablo-D3> like lowp in glsl shaders and shit
572 2011-01-28 14:08:32 <Diablo-D3> imagine if catalyst knew how to do that
573 2011-01-28 14:08:43 <Diablo-D3> it'd be EVEN FASTER
574 2011-01-28 14:09:19 <ArtForz> yup
575 2011-01-28 14:09:51 <Diablo-D3> although, to be fair
576 2011-01-28 14:09:58 <Diablo-D3> thats not the bottleneck
577 2011-01-28 14:10:19 <Diablo-D3> the ONLY bottleneck left (besides the obvious "theres never enough memory bandwidth ever") is how the triangle setup engine works
578 2011-01-28 14:10:26 <ArtForz> yep
579 2011-01-28 14:10:28 <Diablo-D3> it still prefers 16 pixel triangles
580 2011-01-28 14:10:49 <ArtForz> what I dont get, where's the HUGE bottleneck in ATIs arch
581 2011-01-28 14:11:05 <Diablo-D3> I mean, it does everything else wonderfully
582 2011-01-28 14:11:11 <Diablo-D3> you can feed it crazy complex pixel shaders
583 2011-01-28 14:11:17 <Diablo-D3> and it just chugs right through them
584 2011-01-28 14:11:22 <ArtForz> raw shader throughput of a 5870 is >4x of a GTX580 ...
585 2011-01-28 14:11:30 <Diablo-D3> almost disproportionately so
586 2011-01-28 14:11:40 <ArtForz> errr... 3x
587 2011-01-28 14:11:42 <Diablo-D3> it can run through normal shaders faster than it can run through tiny triangles
588 2011-01-28 14:12:09 <ArtForz> err... wait. nevermind
589 2011-01-28 14:12:11 <Diablo-D3> which also fucks itself because it has a fancy new tess engine
590 2011-01-28 14:12:20 <comboy> you guys were progamming GPUs before bitcoin?
591 2011-01-28 14:12:31 <Diablo-D3> raw shader throughput is more like 2x geforce
592 2011-01-28 14:12:34 <ArtForz> yep
593 2011-01-28 14:12:36 <Diablo-D3> on normal glsl shaders
594 2011-01-28 14:12:52 <ArtForz> on anything that makes good use of the VLIW-SIMD arch
595 2011-01-28 14:12:55 <Diablo-D3> do a shitload of textures concurrently, or do exceedingly complex ones, then 5xxx really shines
596 2011-01-28 14:13:09 <Diablo-D3> on most games, its only around 2x
597 2011-01-28 14:13:10 <comboy> may I ask what for? some science computation, security or what?
598 2011-01-28 14:13:22 <Diablo-D3> comboy: I do it has a hobby
599 2011-01-28 14:13:31 <Diablo-D3> ArtForz: now, you have games running the other direction
600 2011-01-28 14:13:39 <Diablo-D3> very simple shaders
601 2011-01-28 14:14:00 <ArtForz> which is why 6xxx got less shadery :P
602 2011-01-28 14:14:16 <Diablo-D3> well, not only less, it seems to handle simple ones faster
603 2011-01-28 14:14:18 <Diablo-D3> not sure wh
604 2011-01-28 14:14:20 <Diablo-D3> *why
605 2011-01-28 14:14:28 <Kiba> Diablo-D3 have a hobby?
606 2011-01-28 14:14:33 <Kiba> I thought he make very little!
607 2011-01-28 14:14:40 <ArtForz> iirc they beefed up the schedulting units for 68xx
608 2011-01-28 14:14:42 <Diablo-D3> my hobbies are free.
609 2011-01-28 14:14:43 <ArtForz> *scheduling
610 2011-01-28 14:14:46 <Diablo-D3> ArtForz: probably
611 2011-01-28 14:14:56 <Diablo-D3> per shader/triangle/whatever setup cost seems to be less
612 2011-01-28 14:15:06 <Diablo-D3> which also seems to be where half that awesome tess performance is coming from
613 2011-01-28 14:15:13 <ArtForz> yep
614 2011-01-28 14:15:31 <Diablo-D3> which is where it fucks itself
615 2011-01-28 14:15:36 <Diablo-D3> if I cant have a triangle per pixel
616 2011-01-28 14:15:39 <Diablo-D3> whats the point
617 2011-01-28 14:15:48 <Diablo-D3> thats the last true flaw.
618 2011-01-28 14:15:54 <luke-jr> davout: +    <%= f.collection_select :bitcoin_number_system, nil, ["Automatic", "Decimal", "Tonal"], :prompt => "--" %>
619 2011-01-28 14:15:57 <luke-jr> davout: that look good?
620 2011-01-28 14:15:58 <ArtForz> *shrug*
621 2011-01-28 14:16:02 <Diablo-D3> I might as well render the screen 16x bigger, with AA off
622 2011-01-28 14:16:06 <Diablo-D3> and then manually shrink it
623 2011-01-28 14:16:16 <Diablo-D3> instant triangle per pixel
624 2011-01-28 14:16:22 <Diablo-D3> NO ONE WILL EVER NOTICE
625 2011-01-28 14:17:59 <ArtForz> might be a reason why ATI is pushing res
626 2011-01-28 14:19:58 <ArtForz> nvidia starts to relatively lose ground once beyond 4MP or so
627 2011-01-28 14:25:25 <x6763> nevermind my question earlier...at this point it looks like a serialization bug
628 2011-01-28 14:29:28 <davout> luke-jr: does it work ? :)
629 2011-01-28 14:29:40 <luke-jr> davout: nfc
630 2011-01-28 14:29:47 <luke-jr> davout: got some Ruby mess to test it on? :P
631 2011-01-28 14:30:15 <davout> luke-jr: what ?
632 2011-01-28 14:30:19 <luke-jr> is Ruby flexible enough to, like PHP, let you throw it in some subdir on bitcoin-central.com?
633 2011-01-28 14:32:39 <davout> oh, do you mean do i have a test server ?
634 2011-01-28 14:32:49 <davout> well easiest is to just move to the rails dir
635 2011-01-28 14:32:53 <davout> and do rails s
636 2011-01-28 14:33:10 <davout> you might want to install the rails stack if you have some ruby on rails development to do :)
637 2011-01-28 14:37:42 <luke-jr> I don't want to install Ruby
638 2011-01-28 14:38:00 <luke-jr> nor a webserver, for that matter
639 2011-01-28 14:40:03 <davout> luke-jr: yea, testing code is for faggots anyway
640 2011-01-28 14:40:35 <luke-jr> :P
641 2011-01-28 14:40:40 <luke-jr> nah, just Ruby
642 2011-01-28 14:41:19 <luke-jr> I wouldn't know how to set it up anyway
643 2011-01-28 14:51:40 <davout> what system are you running ?
644 2011-01-28 14:51:58 <luke-jr> Gentoo
645 2011-01-28 14:52:14 <luke-jr> desktop, not server
646 2011-01-28 14:53:10 <davout> installing rubygems
647 2011-01-28 14:53:14 <davout> shouldn't be that hard
648 2011-01-28 14:53:33 <davout> from there you can just go gem install rails
649 2011-01-28 14:54:37 <Kiba> timmmmmmme to work on my job
650 2011-01-28 15:17:52 <grondilu> yesterday I sent my cheque and credit card back to my bank, asking for the closure of my account.  Now, who's the greatest bitcoin fan :)  ??
651 2011-01-28 15:20:50 <nanotube> grondilu: haha nice
652 2011-01-28 15:21:19 <edcba> grondilu: where is your money now ?
653 2011-01-28 15:22:17 <grondilu> well, honnestly I have an other bank account, with shareholdings.  But no more credit card.
654 2011-01-28 15:23:12 <grondilu> I was spending 60 EUR a year for it and I almost never used it anyway.
655 2011-01-28 15:24:55 <edcba> yes cc are evils
656 2011-01-28 15:25:16 <edcba> but now it seems cc companies are starting to push anonym cc
657 2011-01-28 15:25:16 <grondilu> Moreover, I will try morpheus service (http://www.bitcoin.org/smf/index.php?topic=2627.0) if  really need a VISA card.
658 2011-01-28 15:25:47 <grondilu> anonym cc would be a huge help for bitcoins.
659 2011-01-28 15:25:59 <edcba> why ?
660 2011-01-28 15:26:15 <grondilu> because it would be a great gateway system.
661 2011-01-28 15:26:48 <edcba> i don't follow you
662 2011-01-28 15:26:59 <grondilu> just like morpheus proposal.
663 2011-01-28 15:28:00 <luke-jr> lol
664 2011-01-28 15:28:11 <luke-jr> I don't pay for credit cards
665 2011-01-28 15:28:16 <luke-jr> just have crappy low limits
666 2011-01-28 15:29:27 <afed> any of you deploy a 69X0 card
667 2011-01-28 15:29:55 <afed> i have two 6950 cards that won't run stable more than a few hours
668 2011-01-28 15:30:05 <afed> even without mods or overclocks, and on a 850 watt psu
669 2011-01-28 15:30:47 <afed> new drivers though, i had better try them
670 2011-01-28 16:24:25 <gribble> 24010.90024766
671 2011-01-28 16:24:25 <molecular> ;;bc,estimate
672 2011-01-28 16:24:39 <gribble> 22012.4941572
673 2011-01-28 16:24:39 <molecular> ;;bc,diff
674 2011-01-28 16:25:08 <molecular> ;;bc,stats
675 2011-01-28 16:25:10 <gribble> Current Blocks: 105047 | Current Difficulty: 22012.4941572 | Next Difficulty At Block: 106847 | Next Difficulty In: 1800 blocks | Next Difficulty In About: 1 week, 4 days, 12 hours, 30 minutes, and 0 seconds | Next Difficulty Estimate: 24010.90024766
676 2011-01-28 16:25:28 <newsham> in mtgox when you queue an order, it doesnt seem to reserve the money from your account to cover the order.
677 2011-01-28 16:25:47 <newsham> does that mean you can place many orders in the queue that cant be filled due to insufficient funds?
678 2011-01-28 16:26:10 <newsham> couldnt you use that to game the market depth data?
679 2011-01-28 16:27:11 <gribble> 5252350.00000000
680 2011-01-28 16:27:11 <molecular> ;;bc,totalbc
681 2011-01-28 16:33:04 <molecular> newsham, I think what you're saying is true.
682 2011-01-28 16:34:52 <newsham> :(
683 2011-01-28 16:34:59 <newsham> so market depth data is an illusion :(
684 2011-01-28 16:35:04 <newsham> (not to mention dark pools)
685 2011-01-28 16:35:38 <molecular> well, that one big ask (16500 for 0.51) is not an illusion
686 2011-01-28 16:35:55 <citizen> anyone here use m0mchil's miner on windows?
687 2011-01-28 16:39:25 <molecular> probably not so many here use win. maybe your question can be answered anyway
688 2011-01-28 16:39:45 <citizen> im a bit confused as to the steps to get the miner running on a windows machine
689 2011-01-28 16:39:54 <citizen> 1) download m0mchil's package
690 2011-01-28 16:40:16 <citizen> 2) download python
691 2011-01-28 16:40:23 <citizen> 3) download pyopencl
692 2011-01-28 16:40:37 <citizen> then does something need to be compiled...
693 2011-01-28 16:41:30 <citizen> http://www.newslobster.com/random/how-to-get-started-using-your-gpu-to-mine-for-bitcoins-on-windows
694 2011-01-28 16:41:57 <citizen> the instruction link doesn't even say anything about downloading python or pyopencl
695 2011-01-28 16:42:32 <[Noodles]> if you just use m0mchils pre-compiled binaries, there's no need for python
696 2011-01-28 16:42:49 <citizen> oh ok, so it is set to go then
697 2011-01-28 16:43:03 <[Noodles]> you only need to install python and pyopencl if you want to build it yourself
698 2011-01-28 16:43:13 <citizen> excellent
699 2011-01-28 16:43:17 <[Noodles]> just unpack m0mchils miner and start it
700 2011-01-28 16:43:20 <citizen> thank you so much
701 2011-01-28 16:43:34 <[Noodles]> and yes, me runs it on windows too
702 2011-01-28 16:43:39 <citizen> last question - is it a common practice to run bitcoin -server with an open port
703 2011-01-28 16:43:50 <citizen> on a routable interface
704 2011-01-28 16:44:00 <citizen> and have machines at remote locations dumping into it
705 2011-01-28 16:44:11 <citizen> or are there any security concerns doing this
706 2011-01-28 16:44:53 <[Noodles]> dont think so
707 2011-01-28 16:44:58 <molecular> shouldn't be a problem
708 2011-01-28 16:45:00 <citizen> does bitcoin have to be running on the machine that is running the miner
709 2011-01-28 16:45:06 <molecular> just protect your wallet.dat
710 2011-01-28 16:45:17 <[Noodles]> no, you dont need to run bitcoin to run a miner
711 2011-01-28 16:45:18 <molecular> no, you can do it like you described above
712 2011-01-28 16:45:27 <citizen> cool!
713 2011-01-28 16:45:32 <molecular> just use good rpc password
714 2011-01-28 16:45:46 <citizen> roger that
715 2011-01-28 16:46:08 <molecular> and protect the machine from other intrusions as you normally would
716 2011-01-28 16:46:28 <citiz3n> yeah i would only have 8333 forwarded, nothing else
717 2011-01-28 16:46:48 <citiz3n> thanks much - can't wait to try this business out :)
718 2011-01-28 17:00:44 <citiz3n> so the bitcoin client does not create the bitcoin.conf file even after first run?
719 2011-01-28 17:00:45 <citiz3n> you must create this yourself?
720 2011-01-28 17:01:03 <molecular> I think so
721 2011-01-28 17:01:13 <molecular> at least it was like that with my linux client
722 2011-01-28 17:01:59 <molecular> mine now contains 2 lines: "rpcuser=yyy" and "rpcpassword=xxx"
723 2011-01-28 17:03:49 <citiz3n> any idea if there are size/character limitations on either?
724 2011-01-28 17:04:23 <gavinandresen> Carriage returns in your username or password would be a bad idea....
725 2011-01-28 17:04:48 <gavinandresen> (but otherwise bitcoin doesn't care)
726 2011-01-28 17:05:58 <gavinandresen> Oh, # might cause problems, since that's the comment character for .conf files
727 2011-01-28 17:06:23 <citiz3n> cool
728 2011-01-28 17:09:18 <molecular> hmm, I have 2 bitcoin clients running behind a NAT. the first one is good, 63 connection, but the other one doesn't seem to connect.
729 2011-01-28 17:09:49 <molecular> should I use a different port on the other one or something?
730 2011-01-28 17:09:59 <molecular> or should I just make a single connection to the first one?
731 2011-01-28 17:10:15 <gavinandresen> single connection to the first one (-connect=IP) is the most efficient
732 2011-01-28 17:11:03 <molecular> makes sense, thank. but still the second one should be able to connect to other nodes in theory. it seems it's loading peers from irc (debug.log)
733 2011-01-28 17:11:57 <gavinandresen> Yup, it should be able to connect out.
734 2011-01-28 17:12:30 <gavinandresen> Maybe it is just unlucky and hasn't been able to find a peer that has an open port.
735 2011-01-28 17:12:41 <molecular> possible
736 2011-01-28 17:12:48 <molecular> will just use -connect to the other, good enough
737 2011-01-28 17:19:09 <citiz3n> why am i getting "error: couldn't connect to server" when trying to RUN the server
738 2011-01-28 17:19:16 <citiz3n> im running bitcoin.exe with the -server switch
739 2011-01-28 17:20:21 <gavinandresen> It thinks you're trying to send a command to an already-running bitcoin.  What's the command line exactly?
740 2011-01-28 17:22:33 <citiz3n> ah, i figured it out
741 2011-01-28 17:22:53 <citiz3n> i was having trouble with the syntax in the batch file, so I tried a ) at the end
742 2011-01-28 17:22:57 <citiz3n> like the instructions showed
743 2011-01-28 17:23:04 <citiz3n> when i removed the ) it works now
744 2011-01-28 17:23:06 <citiz3n> :)
745 2011-01-28 17:23:45 <citiz3n> i thought that ) was weird - it was just part of the instructions not part of the command.
746 2011-01-28 17:23:54 <citiz3n> looks like im up and running now
747 2011-01-28 17:24:54 <gavinandresen> That'd do it... Hey, I'm about to go grab lunch, would you mind filing an issue at github bitcoin/bitcoin asking that the "cannot connect to server" error message say what command it was trying to send?  You're not the first person to waste time trying to figure that problem out....
748 2011-01-28 17:33:22 <molecular> newsham: actually I just tried to do what you feared would be possible. and actually now I think mtgox _does_ reserve the funds.
749 2011-01-28 17:34:05 <molecular> newsham, my order got split into one with status "active" and one with status "no enough funds", when I tried to sell more than I had minus what I alread had in orders
750 2011-01-28 17:35:04 <molecular> newsham, I assume the "not enough funds" order will not show up "depth of market"
751 2011-01-28 17:36:33 <molecular> newsham, I just verified this, it doesn't show up in depth of market table
752 2011-01-28 17:45:34 <molecular> someone actually bought 24000 btc about half a day ago on mtgox, am I seeing this right?
753 2011-01-28 17:58:16 <citiz3n> what's the command switch to connect the client to a server on a specified IP?
754 2011-01-28 17:59:09 <citiz3n> -connect=xx.xx.xx.xx
755 2011-01-28 17:59:11 <citiz3n> i think
756 2011-01-28 18:00:25 <citiz3n> or not
757 2011-01-28 18:00:33 <citiz3n> hello, mtgox
758 2011-01-28 18:01:31 <afed> lol difficulty going WAY up
759 2011-01-28 18:04:53 <citiz3n> again?
760 2011-01-28 18:05:05 <EvanR> bah.
761 2011-01-28 18:05:19 <EvanR> i ran bitcoin gui on my new computer, and its taking forever to load
762 2011-01-28 18:05:57 <helmut> EvanR: you mean it is downloading the block chain slowly?
763 2011-01-28 18:05:59 <EvanR> it does network traffic but doesnt show the gui
764 2011-01-28 18:06:09 <EvanR> it has to be finished downloading the block chain, i left it on all night
765 2011-01-28 18:08:17 <afed> reboot and try again?
766 2011-01-28 18:12:57 <gavinandresen> EvanR: windows, mac or linux?
767 2011-01-28 18:13:55 <ne0futur> EvanR: check your firewall ?
768 2011-01-28 18:18:23 <gavinandresen> EvanR: if windows, check for over-active virus software;  see http://www.bitcoin.org/smf/index.php?topic=323.0
769 2011-01-28 18:18:36 <molecular> maybe the net is low on available peers with open ports?
770 2011-01-28 18:18:41 <Kiba> so we got a big fricking protest in Egypt
771 2011-01-28 18:19:08 <Kiba> start up this decade with a big bang, eh
772 2011-01-28 18:19:23 <Kiba> that how 2000s start that way too
773 2011-01-28 18:19:28 <Kiba> 9/11 attack
774 2011-01-28 18:19:29 <gavinandresen> molecular: possibly.  That reminds me, "we" need to update the list of hard-coded nodes in the client...
775 2011-01-28 18:20:12 <molecular> when is 0.3.20 due?
776 2011-01-28 18:21:20 <gavinandresen> When it has had more testing; we need people who are willing to compile and test and report back their results.
777 2011-01-28 18:21:22 <citiz3n> can anyone help me by letting me know how to connect the windows client bitcoin.exe to another windows client running in -server mode
778 2011-01-28 18:21:30 <citiz3n> eg. how to specify the IP for the client to connect to
779 2011-01-28 18:21:50 <molecular> citiz3n, probably use "-connect <ip>"
780 2011-01-28 18:21:51 <citiz3n> does the client even have this capability, or must you run a mining client?
781 2011-01-28 18:21:58 <gavinandresen> citiz3n: both inside a NAT firewall?
782 2011-01-28 18:22:06 <citiz3n> in this case, yes
783 2011-01-28 18:22:31 <gavinandresen> Lets call the machines A and B.  Where A is running normally, connected to outside machines.
784 2011-01-28 18:23:02 <citiz3n> ok
785 2011-01-28 18:23:09 <gavinandresen> On A, you need to set rpcpassword/rpcuser, and set rpcallowip=192.168.*.*  (or whatever your LAN ip addresses are)
786 2011-01-28 18:23:26 <gavinandresen> On B, you need to -connect=A.IP.Address
787 2011-01-28 18:23:46 <citiz3n> ah, rcpallowip...
788 2011-01-28 18:23:49 <citiz3n> i didn't know about that one
789 2011-01-28 18:23:52 <gavinandresen> rPc
790 2011-01-28 18:23:58 <citiz3n> yes
791 2011-01-28 18:23:59 <gavinandresen> (I always accidently type rcp, too)
792 2011-01-28 18:24:00 <molecular> if I don't use rpcallowip, is the rpc port closed completely? it seems I can connect to my "A-Machine" withough using rpcallowip
793 2011-01-28 18:24:08 <citiz3n> rcp sounds better ;)
794 2011-01-28 18:24:17 <citiz3n> but rather than buck the system, ill go with rpc :D
795 2011-01-28 18:24:33 <gavinandresen> Without rpcallowip, only 127.0.0.1 (aka localhost) is allowed to connect.
796 2011-01-28 18:25:35 <molecular> actuall, gavinandresen, that doesn't seem to be true. I started bitcoin on machine A, then on B "bitcoin -connect A".. it's working
797 2011-01-28 18:25:40 <gavinandresen> Oh, wait, hang on, you're not sending rpc commands from B to A....
798 2011-01-28 18:25:47 <gavinandresen> (so you don't actually need rpcallowip)
799 2011-01-28 18:25:55 <molecular> ah!
800 2011-01-28 18:26:04 <molecular> it's a "normal p2p" connect
801 2011-01-28 18:26:05 <molecular> ?
802 2011-01-28 18:26:29 <citiz3n> well, the idea was to have it get/dump work to the instance running with the -server switch
803 2011-01-28 18:26:38 <citiz3n> like a miner would do
804 2011-01-28 18:26:41 <gavinandresen> Yeah, -connect says connect to port 8333 (the bitcoin network port).
805 2011-01-28 18:26:56 <gavinandresen> citiz3n: "it" is bitcoin or a miner?
806 2011-01-28 18:27:04 <citiz3n> bitcoin
807 2011-01-28 18:27:06 <citiz3n> .exe
808 2011-01-28 18:27:20 <citiz3n> the regular GUI windows client
809 2011-01-28 18:27:37 <citiz3n> does this have this functionality, or should i just forget it
810 2011-01-28 18:27:39 <gavinandresen> Last I checked, bitcoin.exe can't act as an external miner.
811 2011-01-28 18:27:58 <gavinandresen> (you can send it getwork, but it won't send completed work out)
812 2011-01-28 18:28:01 <citiz3n> ah ok, so that feature was put in there expressly for 3rd party miners
813 2011-01-28 18:28:14 <gavinandresen> Yup.
814 2011-01-28 18:28:20 <citiz3n> ok cool, well that is cleared up at least :)
815 2011-01-28 18:28:56 <molecular> what's the default rpcport?
816 2011-01-28 18:29:11 <gavinandresen> 8332
817 2011-01-28 18:29:19 <gavinandresen> (change it with -rpcport=...)
818 2011-01-28 18:29:22 <molecular> and 8333 is the p2p-port?
819 2011-01-28 18:29:26 <gavinandresen> yup
820 2011-01-28 18:30:01 <citiz3n> seems like there's a real lack of documentation for all of this
821 2011-01-28 18:30:14 <citiz3n> the forums are a bit hard to navigate and find answers to easy questions
822 2011-01-28 18:30:18 <molecular> stupid me, forgot the -server flag
823 2011-01-28 18:30:26 <citiz3n> how long has bitcoin been active?
824 2011-01-28 18:32:21 <molecular> http://nullvoid.org/bitcoin/difficultiez.php
825 2011-01-28 18:32:32 <molecular> Block 0 was created on Sat, 03 Jan 2009
826 2011-01-28 18:33:59 <luke-jr> gavinandresen: hi
827 2011-01-28 18:34:03 <molecular> http://nullvoid.org/bitcoin/ <- that's a nice site. who's running it?
828 2011-01-28 18:34:13 <gavinandresen> hi luke-jr
829 2011-01-28 18:34:35 <ArtForz> molecular: iirc necrodearia
830 2011-01-28 18:34:35 <luke-jr> gavinandresen: do you concur with dooglus that not rounding BTCents is dangerous? :P
831 2011-01-28 18:34:53 <ArtForz> dangerous how?
832 2011-01-28 18:35:13 <luke-jr> molecular: I think mizerydearia
833 2011-01-28 18:35:17 <luke-jr> ArtForz: I don't know
834 2011-01-28 18:35:18 <ArtForz> right
835 2011-01-28 18:35:25 <ArtForz> too many dearias in here :P
836 2011-01-28 18:36:03 <ArtForz> normal clients are already sending transactions with unrounded subcent values around
837 2011-01-28 18:36:11 <gavinandresen> luke-jr: I looked at your patches, and my concern is that they don't fix the whole problem in a consistent way-- e.g. using your patches, somebody using the GUI might get a message that says "This transaction needs to pay a fee of 0.00 bitcoins; OK?"  .....
838 2011-01-28 18:36:22 <gavinandresen> (because the GUI would continue to round)
839 2011-01-28 18:36:36 <citiz3n> what's the connection ip address switch
840 2011-01-28 18:36:44 <citiz3n> trying to dig through forums to find it
841 2011-01-28 18:36:50 <citiz3n> address?
842 2011-01-28 18:36:56 <ArtForz> imo rounding for displaying the value in that case is stupid
843 2011-01-28 18:37:09 <gavinandresen> ArtForz: I completely agree.
844 2011-01-28 18:37:12 <molecular> citiz3n, did you try "bitcoin -h" ?
845 2011-01-28 18:37:13 <luke-jr> gavinandresen: the GUI doesn't use RPC
846 2011-01-28 18:37:34 <citiz3n> im firing up m0mchil's miner now
847 2011-01-28 18:37:41 <citiz3n> so want to see if it will connect to my server
848 2011-01-28 18:37:41 <luke-jr> IMO displaying any rounded values is a bug
849 2011-01-28 18:37:44 <citiz3n> server is running fine
850 2011-01-28 18:37:59 <gavinandresen> luke-jr:  Right, but one of your patches adds sub-cent "throwaway" change to nFee returned... which will affect the GUI
851 2011-01-28 18:38:10 <necrodearia> ;;bc,stats
852 2011-01-28 18:38:12 <gribble> Current Blocks: 105062 | Current Difficulty: 22012.4941572 | Next Difficulty At Block: 106847 | Next Difficulty In: 1785 blocks | Next Difficulty In About: 1 week, 4 days, 9 hours, 42 minutes, and 0 seconds | Next Difficulty Estimate: 24025.27191971
853 2011-01-28 18:38:16 <citiz3n> i see you have to use --user=youruser -pass=yourpw
854 2011-01-28 18:38:27 <citiz3n> but the instructions don't say how to specify an IP
855 2011-01-28 18:38:36 <necrodearia> mm, my site's estimated difficulty shows est. new difficulty: 23881.47951016981
856 2011-01-28 18:38:51 <necrodearia> it must be wrong
857 2011-01-28 18:38:58 <molecular> are you trying to use bitcoin.exe as a miner connection to another bitcoin.exe? that won't work
858 2011-01-28 18:39:06 <molecular> *connecting
859 2011-01-28 18:39:11 <citiz3n> nope, im using m0mchil's miner now
860 2011-01-28 18:39:15 <ArtForz> my code currently says ... 24025.3
861 2011-01-28 18:39:15 <molecular> ah, ok
862 2011-01-28 18:39:18 <gavinandresen> luke-jr:  I agree that displaying rounded values in the GUI aught to be fixed-- it aught to display 1.00  for 1 bitcoin, or 1.00000001  if that's the exact amount.
863 2011-01-28 18:39:57 <luke-jr> gavinandresen: it's better to show 0.00 BTC fee, than to not show a fee at all, IMO
864 2011-01-28 18:40:00 <Kiba> I realize I been procrastinating quite a bit
865 2011-01-28 18:40:06 <ArtForz> yep
866 2011-01-28 18:40:22 <molecular> citiz3n, maybe you have to also use "--host"
867 2011-01-28 18:40:35 <molecular> citiz3n, and also, did you start bitcoin.exe with the "-server" flag?
868 2011-01-28 18:40:41 <citiz3n> yup :)
869 2011-01-28 18:40:47 <citiz3n> i will try --host
870 2011-01-28 18:40:57 <citiz3n> i didn't know if it was --address --ip --host etc
871 2011-01-28 18:41:04 <citiz3n> so --host=xx.xx.xx.xx
872 2011-01-28 18:41:14 <molecular> if you set the rpcport to sth other than 8332 you also need "--port"
873 2011-01-28 18:41:25 <molecular> yeah
874 2011-01-28 18:41:32 <ArtForz> hrrrm, or dont report unavoidable sub-cent change throwaway at all
875 2011-01-28 18:41:47 <gavinandresen> ArtForz: that's what the GUI does now.
876 2011-01-28 18:41:56 <ArtForz> errr... no
877 2011-01-28 18:42:06 <ArtForz> right now it doesnt even try to avoid sub-cent change
878 2011-01-28 18:42:42 <molecular> citiz3n, if you have rather low mining power, I would suggest using pooled mining (e.g. http://mining.bitcoin.cz) anyway
879 2011-01-28 18:42:57 <luke-jr> gitorious master now only includes the undisputable bugfix: avoiding subcent throwaway
880 2011-01-28 18:43:10 <ArtForz> ahhh, k
881 2011-01-28 18:43:14 <ArtForz> well, that sounds ok
882 2011-01-28 18:43:20 <citiz3n> a window is popping open then closing right away
883 2011-01-28 18:43:24 <citiz3n> i can't see what it's saying
884 2011-01-28 18:43:24 <luke-jr> IMO, reporting the subcent throwaway as a fee, is also a bugfix
885 2011-01-28 18:43:29 <gavinandresen> luke-jr:  I suppose my, and dooglus' meta-point is that it's not a simple straightforward bug fix.  So it aught to be discussed in the forums.
886 2011-01-28 18:43:33 <luke-jr> but if it's disputed, I can make it another branch
887 2011-01-28 18:43:40 <ArtForz> except if you see it, that means you can't avoid it
888 2011-01-28 18:43:40 <molecular> citiz3n, when doing what?
889 2011-01-28 18:43:46 <luke-jr> gavinandresen: well, dooglus's point was about something different
890 2011-01-28 18:44:01 <luke-jr> ArtForz: but you might opt to send the recipient it instead
891 2011-01-28 18:44:05 <citiz3n> i created a batch file to start the poclbm.exe
892 2011-01-28 18:44:05 <luke-jr> eg, Fee or Send All
893 2011-01-28 18:44:12 <gavinandresen> Am I misremembering?  I thought he asked you to create a separate thread to discuss....
894 2011-01-28 18:44:22 <citiz3n> with --user=username --pass=mypw --host=xx.xx.xx.xx
895 2011-01-28 18:44:23 <molecular> try starting it from a cmd-prompt (start->execute->"cmd")
896 2011-01-28 18:44:27 <luke-jr> gavinandresen: no, that's right. Just not about the subcent fee bit
897 2011-01-28 18:44:29 <citiz3n> yeah i did that too
898 2011-01-28 18:44:44 <citiz3n> it's definitely running the poclbm.exe
899 2011-01-28 18:44:44 <luke-jr> gavinandresen: his concept was allowing RPC commands to give subcent floats
900 2011-01-28 18:44:45 <ArtForz> I'd like to avoid that currently
901 2011-01-28 18:44:50 <molecular> then drop your .bat-file into the black cmd-window
902 2011-01-28 18:44:52 <citiz3n> but i can't tell what it's doing after that it just pops open then close
903 2011-01-28 18:44:58 <citiz3n> yeah i did that too molecular
904 2011-01-28 18:45:14 <citiz3n> it pops open a new black window that closes right away
905 2011-01-28 18:45:21 <citiz3n> it prints some text then closes
906 2011-01-28 18:45:30 <citiz3n> maybe i can screenshot it really quick :P
907 2011-01-28 18:45:35 <molecular> yeah, to avoid closing: do what I said above (use "cmd")
908 2011-01-28 18:45:39 <ArtForz> sending transactions that include unexpected sub-cent values to recipient around could cause problems
909 2011-01-28 18:45:44 <molecular> no, it's simpler than that
910 2011-01-28 18:46:03 <molecular> you open a terminal by running "cmd" (start->execute, enter "cmd")
911 2011-01-28 18:46:24 <citiz3n> yeah that's what i tried
912 2011-01-28 18:46:30 <citiz3n> the command prompt stays open just fine
913 2011-01-28 18:46:36 <molecular> that window should not close until you enter "exit"