1 2011-02-04 00:00:02 <jav> doublec, I believe the latest bitcoin code also adds the option '-nolisten' .. should be another way to avoid having people connect
  2 2011-02-04 00:00:40 <doublec> luke-jr: N900 is definitely a phone
  3 2011-02-04 00:00:46 <luke-jr> doublec: nope.
  4 2011-02-04 00:00:49 <doublec> luke-jr: I must be imagining that phone call I made on it before
  5 2011-02-04 00:00:50 <sipa> can you call with it?
  6 2011-02-04 00:00:51 <doublec> sipa: yes, it's a phone
  7 2011-02-04 00:00:54 <luke-jr> doublec: ability to make calls does not make it a phone
  8 2011-02-04 00:01:10 <luke-jr> I can make calls with my desktop
  9 2011-02-04 00:01:21 <sipa> fine
 10 2011-02-04 00:01:31 <luke-jr> sipa: earlier models didn't even have cellular connectivity
 11 2011-02-04 00:01:57 <luke-jr> that was a new feature in N900, and now all of a sudden it's a "phone" to a bunch of n00bs
 12 2011-02-04 00:02:10 <luke-jr> and N900 even was only spec'd to have it for data only originally
 13 2011-02-04 00:02:23 <sipa> question: does this matter?
 14 2011-02-04 00:02:26 <luke-jr> Nokia just said "heck, why not?" and added the minimal code for audio
 15 2011-02-04 00:02:30 <luke-jr> probably not
 16 2011-02-04 00:02:33 <sipa> ok
 17 2011-02-04 00:02:46 <luke-jr> unless you want to buy a phone
 18 2011-02-04 00:03:00 <luke-jr> then it matters, because N900 doesn't have the features people expect in a phone
 19 2011-02-04 00:03:07 <doublec> like?
 20 2011-02-04 00:03:21 <luke-jr> like MMS, or a phone-centric user interface
 21 2011-02-04 00:03:30 <sipa> mms?
 22 2011-02-04 00:03:34 <doublec> I can send and receive MMS fine
 23 2011-02-04 00:03:35 <luke-jr> sipa: picture SMS
 24 2011-02-04 00:03:43 <sipa> ill send an email if i want
 25 2011-02-04 00:03:47 <luke-jr> doublec: with third party software that only works by random chance
 26 2011-02-04 00:03:54 <doublec> same way as the original iPhone didnt have MMS - but it was still a phone
 27 2011-02-04 00:03:55 <luke-jr> sipa: sure, but that's not a photo interface
 28 2011-02-04 00:04:00 <luke-jr> it didn't?
 29 2011-02-04 00:04:03 <doublec> nope
 30 2011-02-04 00:04:06 <sipa> ive never sent or received an mms in my life
 31 2011-02-04 00:04:12 <luke-jr> still, the software for iPhone is phone software
 32 2011-02-04 00:04:15 <doublec> only jailbroken original iPhones had MMS originally
 33 2011-02-04 00:04:23 <doublec> through third party software
 34 2011-02-04 00:04:37 <luke-jr> for N900, you have to go launch the Phone app to get phone functionality
 35 2011-02-04 00:04:57 <doublec> just like you have to on the iPhone
 36 2011-02-04 00:05:07 <luke-jr> I'm not complaining, I think handheld computing (that is, non-phones) is the future.
 37 2011-02-04 00:05:28 <sipa> or on my android
 38 2011-02-04 00:05:31 <luke-jr> doublec: iPhone can't run arbitrary GNU/Linux/X apps, so there :p
 39 2011-02-04 00:05:38 <luke-jr> neither can Android
 40 2011-02-04 00:10:56 <sipa> luke-jr: but an android phone is a phone, and an n900 isn't?
 41 2011-02-04 00:11:16 <luke-jr> yes, Android is a phone platform.
 42 2011-02-04 00:11:31 <luke-jr> Maemo was a general-purpose computing platform.
 43 2011-02-04 00:11:36 <sipa> but if i put debian on my phone, it stops being a phone?
 44 2011-02-04 00:11:47 <luke-jr> yes, then you've repurposed it :p
 45 2011-02-04 00:12:06 <luke-jr> like if you root an XBox360 and run it as a server, it's a server, not a game console
 46 2011-02-04 00:12:22 <luke-jr> :p
 47 2011-02-04 00:12:36 <Syke> It all depends on your definition of "is".
 48 2011-02-04 00:12:42 <sipa> haha
 49 2011-02-04 00:12:49 <ArtForz> hhehehe
 50 2011-02-04 01:06:55 <afed> ;;bc,diff
 51 2011-02-04 01:06:55 <gribble> 22012.4941572
 52 2011-02-04 01:19:31 <hacim> EvanR: yeah, its a a little scary running solo when you are trying to recoup your costs, eh?
 53 2011-02-04 01:20:49 <EvanR> im used to that though ;)
 54 2011-02-04 01:20:56 <EvanR> i havent made a profit... ever
 55 2011-02-04 01:21:03 <EvanR> some shit always happens to deplete my reserves
 56 2011-02-04 01:21:17 <EvanR> bitcoin seems like independent profit at this point ;)
 57 2011-02-04 01:24:12 <EvanR> yee bitch
 58 2011-02-04 01:24:17 <EvanR> look at those blocks fly
 59 2011-02-04 01:24:20 <EvanR> lol
 60 2011-02-04 01:26:14 <hacim> hehe
 61 2011-02-04 01:27:36 <EvanR> they havent found a block in 2:22:22
 62 2011-02-04 01:27:49 <EvanR> that means the next block outta not take very long at all!
 63 2011-02-04 01:28:14 <EvanR> </gamblersfallacy>
 64 2011-02-04 01:28:39 <lfm> bad time to join pool
 65 2011-02-04 01:28:58 <EvanR> why?
 66 2011-02-04 01:29:29 <lfm> better pay within first hour or two after a block has been found
 67 2011-02-04 01:29:56 <EvanR> well better join now than later
 68 2011-02-04 01:30:05 <lfm> maybe
 69 2011-02-04 01:33:08 <EvanR> this wallet i have has balance zero, one +50 generate transaction and -50 spend transaction
 70 2011-02-04 01:33:13 <EvanR> can i just throw it away?
 71 2011-02-04 01:34:04 <hacim> if you want
 72 2011-02-04 01:34:12 <EvanR> i cant combine it with my main wallet right
 73 2011-02-04 01:34:13 <doublec> if you've published any addresses from the wallet then any coins sent to that address will be gone forever if you throw it away
 74 2011-02-04 01:34:27 <hacim> it has zero balance
 75 2011-02-04 01:34:46 <EvanR> but theres history, and as doublec said, someone may send to it
 76 2011-02-04 01:35:01 <doublec> a tool to merge wallets would be useful
 77 2011-02-04 01:35:02 <EvanR> well in principle. i didnt public any addresses from it
 78 2011-02-04 01:35:15 <doublec> I had the same situation when I shut a vps down that had a wallet that had addresses I'd used for donations
 79 2011-02-04 01:36:03 <lfm> just merge the balances with a send
 80 2011-02-04 01:36:19 <hacim> if you think there might be coin coming in you'd want to save the private key material and reuse it elsewhere
 81 2011-02-04 01:36:20 <lfm> oh nm i c
 82 2011-02-04 01:36:22 <EvanR> but future payments would be lost
 83 2011-02-04 01:36:32 <doublec> yeah it's not the balances, it's the private keys for the addresses
 84 2011-02-04 01:36:55 <hacim> then you gotta run the private keys on the p2p net until yuou dont think anymore coin will come in
 85 2011-02-04 01:37:31 <EvanR> make sure you change all your public addresses, and dont send to random addresses you wrote down without checking the primary source ;)
 86 2011-02-04 01:37:44 <lfm> well you dont have to run them, just save em
 87 2011-02-04 01:38:03 <presence> there are a couple of people in the virtual hosting game for BTC are there not?
 88 2011-02-04 01:38:08 <EvanR> might suck to have to periodically cycle through all your old wallets to see if anything showed up in them ;)
 89 2011-02-04 01:38:16 <lfm> ya
 90 2011-02-04 01:38:41 <lfm> its also bug prone
 91 2011-02-04 01:39:26 <presence> How many btc/month for hosting
 92 2011-02-04 01:39:46 <doublec> presence: https://www.autovps.net/?Currency=BTC
 93 2011-02-04 01:39:51 <EvanR> i ran my miners for like 2 days with no blocks. i just switched to the pool so now i feel like i wasted work ;)
 94 2011-02-04 01:40:01 <EvanR> if i had just continued mining a little longer id get 50... right?! ;)
 95 2011-02-04 01:40:21 <lfm> or maybe nothing for years
 96 2011-02-04 01:40:31 <EvanR> 2 days was my average
 97 2011-02-04 01:40:37 <doublec> "I've put so much money into this slot machine, I can't stop now, it has to pay out soon, right?"
 98 2011-02-04 01:40:43 <EvanR> lol
 99 2011-02-04 01:40:58 <EvanR> better to stop earlier than later, i guess
100 2011-02-04 01:42:00 <lfm> if you're losing and your odd predict you will continue to lose on average then yes it is better to stop sooner than later
101 2011-02-04 01:42:36 <lfm> odds
102 2011-02-04 01:42:38 <EvanR> so whats the optimal configuration of fan speed and clock speed to heat my room. we have a freezing rain severe weather warning tonight
103 2011-02-04 01:42:42 <EvanR> xD
104 2011-02-04 01:43:24 <EvanR> is slow fan and high temperature better? or high speed fan and cooler gpus
105 2011-02-04 01:44:20 <lfm> depends how hi is hi
106 2011-02-04 01:45:14 <lfm> up to about 70-80c you should be fine higher than that you may be asking for trouble
107 2011-02-04 01:46:00 <EvanR> and blowing faster is better?
108 2011-02-04 01:46:09 <EvanR> at that temperature
109 2011-02-04 01:46:24 <lfm> if you're too hot then ya speed up your fan if oyu can
110 2011-02-04 01:47:16 <EvanR> no, im too cold ;)
111 2011-02-04 01:47:21 <lfm> remember fans use electricity too so dont use em if you dont need em. below 70c you can slow down the fan to save power
112 2011-02-04 01:47:36 <EvanR> no im trying to use more power
113 2011-02-04 01:47:56 <lfm> you have free power? or you need more heat?
114 2011-02-04 01:47:59 <EvanR> i disabled the central heater to save money
115 2011-02-04 01:48:04 <EvanR> but its freezing outside
116 2011-02-04 01:48:05 <EvanR> lol
117 2011-02-04 01:48:39 <EvanR> freezing outside this room too, but i blocked the cracks in the door and the window
118 2011-02-04 01:49:01 <lfm> what does your central heat run on? electricity? or gas or oil? sometimes eletricity is not a good choice for heating
119 2011-02-04 01:49:10 <EvanR> yes its electricity
120 2011-02-04 01:49:15 <EvanR> costs a lot
121 2011-02-04 01:49:29 <lfm> ok just turn up your fans to 100% then
122 2011-02-04 01:49:42 <EvanR> k
123 2011-02-04 01:49:47 <lfm> it still might not be enuf heat of course
124 2011-02-04 01:49:55 <EvanR> now this will cause the gpus to run at like 57C
125 2011-02-04 01:50:17 <EvanR> wouldnt it be nice if i could overclock to like 80C ;)
126 2011-02-04 01:50:27 <lfm> ya thats fine, they are still producing the same heat, its just being blown into the room
127 2011-02-04 01:50:29 <EvanR> but im already around the limit before it crashes :(
128 2011-02-04 01:50:31 <ArtForz> so?
129 2011-02-04 01:50:45 <ArtForz> higher exhaust velocity is better, gets more air movement
130 2011-02-04 01:50:57 <ArtForz> otherwise you get *very* nice layering
131 2011-02-04 01:51:07 <EvanR> yes thats fine, but id like it to be hotter
132 2011-02-04 01:51:08 <lfm> and higher fan power produces heat in the fan
133 2011-02-04 01:51:26 <EvanR> overclocking does that, and makes more coins... but crashes X
134 2011-02-04 01:51:30 <EvanR> :(
135 2011-02-04 01:51:59 <lfm> EvanR did you try increasing the voltage? (its dangerous tho)
136 2011-02-04 01:52:03 <EvanR> heh
137 2011-02-04 01:52:04 <EvanR> no
138 2011-02-04 01:53:00 <EvanR> how about cpu mining at the same time ;)
139 2011-02-04 01:53:13 <lfm> yup, turn up the cpus too
140 2011-02-04 01:53:21 <EvanR> m0's work on the cpu?
141 2011-02-04 01:53:30 <EvanR> i saw it listed, but i thought it was a gpu miner
142 2011-02-04 01:53:49 <lfm> just use main bitcoin or cpuminer from jgarzik
143 2011-02-04 01:54:32 <lfm> I think puddinpoop has some cpu miners too
144 2011-02-04 01:55:06 <EvanR> ive used jgarzik before so ill use that
145 2011-02-04 01:55:07 <presence> so dont run x
146 2011-02-04 01:55:12 <presence> why do you need x anyway
147 2011-02-04 01:55:20 <EvanR> because
148 2011-02-04 01:55:32 <EvanR> ;)
149 2011-02-04 01:55:32 <lfm> the opencl support is iin the X drivers
150 2011-02-04 01:55:44 <presence> you can load thr drivers without running x
151 2011-02-04 01:56:06 <EvanR> hehe, disable X, and you can overclock to whatever!
152 2011-02-04 01:56:24 <lfm> presence: ok, go ahead, and if you get the chance detail for the rest of us how you do it
153 2011-02-04 01:56:59 <presence> hell, I have the video card passed through to the VM on my esx4i server and it works fine :D
154 2011-02-04 01:57:23 <lfm> ok esx eh, not for me
155 2011-02-04 01:57:38 <presence> I use it for other reasons.  Why not put a card in it and mine?
156 2011-02-04 01:57:51 <presence> better than standing up a host solely to use the gpu
157 2011-02-04 02:01:51 <EvanR> god dammit
158 2011-02-04 02:02:30 <EvanR> again with cpuminer automake crap
159 2011-02-04 02:03:19 <EvanR> does anyone have binaries
160 2011-02-04 02:03:57 <lfm> automake is good, whats wrong with your system?
161 2011-02-04 02:04:19 <EvanR> nothing
162 2011-02-04 02:04:28 <EvanR> this is the first times its failed
163 2011-02-04 02:04:32 <EvanR> i always build my own stuff
164 2011-02-04 02:04:35 <lfm> its missing some libs eh?
165 2011-02-04 02:04:55 <EvanR> ill pastebin the error
166 2011-02-04 02:07:41 <EvanR> http://codepad.org/yp280REl
167 2011-02-04 02:08:56 <lfm> you cant just run ./configure instead?
168 2011-02-04 02:09:27 <EvanR> oh, it comes with configure now.
169 2011-02-04 02:09:28 <EvanR> ill try
170 2011-02-04 02:09:37 <lfm> doh
171 2011-02-04 02:09:40 <EvanR> no
172 2011-02-04 02:09:47 <EvanR> let me start over
173 2011-02-04 02:10:05 <ne0futur> re install imlib-devel
174 2011-02-04 02:10:06 <EvanR> no, it doesnt come with configure
175 2011-02-04 02:10:08 <ne0futur> clean all
176 2011-02-04 02:10:23 <EvanR> i saw a half built version, which failed
177 2011-02-04 02:10:36 <EvanR> all 'packages' come with no configure, configured
178 2011-02-04 02:10:58 <EvanR> autoconf throws that error when configure attempts to generate
179 2011-02-04 02:11:11 <EvanR> 2.65
180 2011-02-04 02:11:16 <EvanR> :(
181 2011-02-04 02:11:51 <EvanR> last time i complained jgarzik gave me a package with stuff configured and i just have to configure make
182 2011-02-04 02:16:13 <jgarzik> EvanR: download cpuminer 0.6 tarball from http://yyz.us/bitcoin/
183 2011-02-04 02:16:40 <EvanR> ah
184 2011-02-04 02:16:51 <EvanR> bookmarked
185 2011-02-04 02:17:07 <EvanR> i tried cryptopp_asm32 on my 0.3 binary on my new miner... segfaulted
186 2011-02-04 02:17:11 <EvanR> ill try .6
187 2011-02-04 02:17:33 <lfm> EvanR I just downloaded cpuminer 0.6 and it has configure
188 2011-02-04 02:17:54 <lfm> tar.gz
189 2011-02-04 02:18:11 <EvanR> my browser doesnt show me a tar.gz for .6 in github
190 2011-02-04 02:18:18 <EvanR> just master
191 2011-02-04 02:18:30 <lfm> https://www.bitcoin.org/smf/index.php?topic=1925.0
192 2011-02-04 02:19:55 <EvanR> cryptopp_asm32 works!
193 2011-02-04 02:20:04 <EvanR> 150% increase in speed
194 2011-02-04 02:20:13 <EvanR> 1.2 Mhps
195 2011-02-04 02:20:22 <EvanR> sempron ;)
196 2011-02-04 02:20:39 <lfm> wtg
197 2011-02-04 02:21:24 <EvanR> 3hours block...
198 2011-02-04 02:21:36 <EvanR> any minute now
199 2011-02-04 02:23:31 <EvanR> ;;bc,calc 30000000
200 2011-02-04 02:23:32 <gribble> The average time to generate a block at 30000000 Khps, given current difficulty of 22012.4941572 , is 52 minutes and 31 seconds
201 2011-02-04 02:24:06 <jgarzik> EvanR: you don't want the .tar.gz from github; that is not a distribution tarball, so you'd have to deal with autoconf/automake nastiness
202 2011-02-04 02:24:15 <jgarzik> distribution tarball has configure, and saves you a lot of work
203 2011-02-04 02:24:23 <EvanR> yeah i figured
204 2011-02-04 02:24:32 <EvanR> is there a link to that distribution site on github?
205 2011-02-04 02:24:44 <jgarzik> yyz.us/bitcoin is the official download site for cpuminer
206 2011-02-04 02:24:52 <EvanR> tell google that
207 2011-02-04 02:25:09 <jgarzik> EvanR: I don't know how to set that up on github. If that's possible, I should do that
208 2011-02-04 02:25:32 <EvanR> i dont know github
209 2011-02-04 02:25:58 <lfm> blind leading blind eh
210 2011-02-04 02:26:04 <andrew12> heh
211 2011-02-04 02:26:45 <andrew12> jgarzik: are you trying to figure out if it's possible to make a site with github?
212 2011-02-04 02:27:02 <andrew12> or are you trying to figure out how to upload tars and shit on github
213 2011-02-04 02:27:03 <EvanR> a link to the project homepage on the github page
214 2011-02-04 02:27:09 <andrew12> because both are possible
215 2011-02-04 02:27:16 <andrew12> oh
216 2011-02-04 02:27:39 <andrew12> sec
217 2011-02-04 02:27:40 <lfm> andrew12: they just want people who stumble on github to find the official downloads
218 2011-02-04 02:28:02 <andrew12> jgarzik: on https://github.com/jgarzik/cpuminer, click admin,
219 2011-02-04 02:28:06 <andrew12> ...damn you enter key
220 2011-02-04 02:28:33 <andrew12> er
221 2011-02-04 02:28:48 <EvanR> so, as this 3+ hour long block on the pool continues, my expected payout is growing. was 0, then .1 .2 now .3
222 2011-02-04 02:28:51 <andrew12> jgarzik: on https://github.com/jgarzik/cpuminer, hover over the empty space under the description. it's pretty obvious from there
223 2011-02-04 02:28:58 <EvanR> is it going to continue to grow?
224 2011-02-04 02:29:04 <andrew12> EvanR: maybe!
225 2011-02-04 02:29:09 <EvanR> well it cant go over 50
226 2011-02-04 02:29:17 <andrew12> what if it does?! :p
227 2011-02-04 02:29:24 <EvanR> ._.
228 2011-02-04 02:30:49 <lfm> whats your khash/s?
229 2011-02-04 02:32:04 <hacim> when you run a cpu miner, do you want to run one thread per physical CPU/
230 2011-02-04 02:32:15 <luke-jr> hacim: one thread per core, like with anything
231 2011-02-04 02:32:41 <hacim> luke-jr: so if you have 2 cpus hyperthreaded  you launch 4
232 2011-02-04 02:32:52 <luke-jr> no
233 2011-02-04 02:32:56 <luke-jr> hyperthreading is not cores
234 2011-02-04 02:33:07 <lfm> oh you have to try it, hyperthreaded cpus may not count as full cpus
235 2011-02-04 02:33:25 <luke-jr> hacim: you have 2 single-core CPUs?
236 2011-02-04 02:33:27 <lfm> sometimes they work and sometimes they dont
237 2011-02-04 02:33:51 <luke-jr> lfm: spawning cores + 1 may be of minute benefit, but that's about it really
238 2011-02-04 02:34:02 <lfm> depends
239 2011-02-04 02:34:05 <EvanR> lfm: 1.2M on the cpu 620 on the gpu
240 2011-02-04 02:34:47 <gribble> The average time to generate a block at 621200 Khps, given the supplied difficulty of 1, is 6 seconds
241 2011-02-04 02:34:47 <lfm> ;;bc,calcd 621200 1
242 2011-02-04 02:34:57 <luke-jr> lol
243 2011-02-04 02:35:02 <EvanR> $_$
244 2011-02-04 02:35:05 <hacim> i'm getting 475.52 per thread
245 2011-02-04 02:35:20 <luke-jr> can you imagine, if ArtForz had his GPU miner at the start? XD
246 2011-02-04 02:35:36 <presence> the difficulty would have went up fast
247 2011-02-04 02:35:37 <EvanR> hed have 21 million!
248 2011-02-04 02:35:37 <luke-jr> he'd be insane rich
249 2011-02-04 02:35:42 <presence> nah
250 2011-02-04 02:35:47 <presence> it would auto-scale
251 2011-02-04 02:35:48 <luke-jr> 51k bitcoins off the bat
252 2011-02-04 02:35:49 <lfm> luke well difficulty would have been a lot higher
253 2011-02-04 02:35:57 <luke-jr> lfm: only after 1020 blocks
254 2011-02-04 02:36:03 <luke-jr> the first 1020 would have been easy money
255 2011-02-04 02:36:12 <lfm> 2016 block you mean?
256 2011-02-04 02:36:15 <luke-jr> yeah
257 2011-02-04 02:36:27 <luke-jr> $100k if he saved it until now
258 2011-02-04 02:36:39 <luke-jr> in under 4 hours
259 2011-02-04 02:36:41 <lfm> luke-jr:  ya I saw what happened when he turned on testnet
260 2011-02-04 02:36:50 <luke-jr> $25k/hr
261 2011-02-04 02:36:55 <luke-jr> wish I could get paid that
262 2011-02-04 02:36:57 <luke-jr> XD
263 2011-02-04 02:37:16 <lfm> hindsight
264 2011-02-04 02:37:19 <EvanR> 100k meh
265 2011-02-04 02:37:26 <EvanR> few years salary lol
266 2011-02-04 02:37:33 <EvanR> chump change
267 2011-02-04 02:37:52 <luke-jr> Warning: overflow encountered in uint_scalars
268 2011-02-04 02:37:55 <luke-jr> so anyone else get that?
269 2011-02-04 02:38:05 <luke-jr> I'd really like some peace of mind that my mining is working :x
270 2011-02-04 02:38:18 <lfm> if he had done that it might have ruined bitcoin from the start and it might never have been worth anything, you cant say for sure
271 2011-02-04 02:38:19 <EvanR> you got coins from it?
272 2011-02-04 02:38:24 <luke-jr> EvanR: not so far
273 2011-02-04 02:38:25 <EvanR> its working xD
274 2011-02-04 02:38:27 <EvanR> oh
275 2011-02-04 02:38:55 <EvanR> lfm: like satoshi?
276 2011-02-04 02:38:58 <epicurus> i'll wager 500 quatloos on the newcomer
277 2011-02-04 02:38:58 <EvanR> lol
278 2011-02-04 02:38:59 <luke-jr> but I'm only supposed to find a blokc every 4 days or so
279 2011-02-04 02:39:07 <luke-jr> wtf are quatloos
280 2011-02-04 02:39:21 <luke-jr> is that a new bitcoin unit?
281 2011-02-04 02:39:23 <lfm> ditto
282 2011-02-04 02:39:26 <EvanR> luke-jr: yeah so it might be 8 days, maybe even 16!
283 2011-02-04 02:39:29 <EvanR> :S
284 2011-02-04 02:39:40 <hacim> i get pretty much the same hashrate with cryptopp as SHA256
285 2011-02-04 02:39:56 <EvanR> gotta use cyptopp_asm32
286 2011-02-04 02:40:13 <lfm> i use via
287 2011-02-04 02:40:29 <luke-jr> does anyone here by chance have Deadly Games?
288 2011-02-04 02:40:33 <hacim> via gives me an illegal instruction
289 2011-02-04 02:40:44 <hacim> i dont see any improvement with cyptopp_asm32
290 2011-02-04 02:40:47 <luke-jr> hacim: pretty sure via requires special hw
291 2011-02-04 02:40:51 <lfm> well you have to have a VIA cpu, not intel or amd
292 2011-02-04 02:41:08 <luke-jr> lfm: you should add it to the wiki
293 2011-02-04 02:41:15 <luke-jr> under CPUs/Nvidia
294 2011-02-04 02:41:26 <lfm> nvidia isnt via
295 2011-02-04 02:41:41 <luke-jr> no?
296 2011-02-04 02:41:45 <lfm> what wiki?
297 2011-02-04 02:41:50 <gribble> (bc,calcd <an alias, 2 arguments>) -- Alias for "echo The average time to generate a block at $1 Khps, given the supplied difficulty of $2, is [time elapsed [math calc 1/((2**224-1)/$2*$1*1000/2**256)]]".
298 2011-02-04 02:41:50 <hacim> ;;bc,calcd 2743
299 2011-02-04 02:41:50 <luke-jr> bitocin wiki
300 2011-02-04 02:42:06 <luke-jr> nvidia didn't buy via?
301 2011-02-04 02:42:19 <lfm> dont think so
302 2011-02-04 02:42:46 <luke-jr> O.O
303 2011-02-04 02:42:51 <luke-jr> weird maths
304 2011-02-04 02:43:07 <luke-jr> lfm: https://en.bitcoin.it/wiki/Mining_Hardware_Comparison
305 2011-02-04 02:43:54 <luke-jr> I guess put it under VIA then
306 2011-02-04 02:44:12 <presence> I wanna know why teslas suck ass
307 2011-02-04 02:44:16 <presence> 1.x Tflops each
308 2011-02-04 02:44:27 <presence> the drivers have to be balls
309 2011-02-04 02:44:33 <EvanR> lol
310 2011-02-04 02:44:35 <presence> which is lame since it doesnt drive video
311 2011-02-04 02:44:39 <luke-jr> presence: because it's nvidia
312 2011-02-04 02:44:39 <presence> it JUST DOES MATH
313 2011-02-04 02:44:47 <luke-jr> all nvidia stuff sucks
314 2011-02-04 02:45:07 <lfm> teslas are good for double precision fp and also good for ecc memory
315 2011-02-04 02:46:10 <lfm> altho ati DP may be has caught up now
316 2011-02-04 02:46:48 <ArtForz> they did
317 2011-02-04 02:47:00 <ArtForz> 6970 is faster than C20xx for DP
318 2011-02-04 02:47:19 <ArtForz> well, C/M/whatever 20xx
319 2011-02-04 02:47:41 <epicurus> http://www.youtube.com/watch?v=umbqfa3XyTA#t=4m54s
320 2011-02-04 02:47:53 <epicurus> to learn about quatloos
321 2011-02-04 04:45:42 <frewsxcv> if there is no central bitcoin server, where does the data come to generate coins?
322 2011-02-04 04:46:04 <luke-jr> frewsxcv: from the person who generates it
323 2011-02-04 04:46:12 <luke-jr> frewsxcv: it's just a bruteforce mechanism
324 2011-02-04 04:46:42 <luke-jr> the SHA256 hash of the block must be less than the difficulty
325 2011-02-04 04:47:05 <luke-jr> so miners are just bruteforcing different data to try to find a hash that fits
326 2011-02-04 04:50:04 <andrew12> was just about to ask why BCBot sends its message after you identify, then i remembered that the server makes it look like you reconnected
327 2011-02-04 04:51:45 <luke-jr> it has to
328 2011-02-04 04:55:11 <frewsxcv> luke-jr: how does the person who sends the data know where to send the data?
329 2011-02-04 04:55:31 <luke-jr> frewsxcv: it sends it everywhere
330 2011-02-04 04:55:54 <draginx> its a& torrent.. network lol =x
331 2011-02-04 04:56:19 <draginx> Upload/Download
332 2011-02-04 04:57:56 <frewsxcv> draginx: torrents have a tracker
333 2011-02-04 05:01:18 <luke-jr> do clients care what the Timestamp is for blocks?
334 2011-02-04 05:01:33 <luke-jr> eg, will they notice if a miner is off by several years?
335 2011-02-04 05:04:37 <nanotube> yes, there is a validity window.
336 2011-02-04 05:05:50 <luke-jr> how big is it?
337 2011-02-04 05:06:06 <luke-jr> what if it just takes a long time to cross the network?
338 2011-02-04 05:06:15 <luke-jr> what if someone's timezone is just set wrong?
339 2011-02-04 05:06:38 <draginx> isn't the irc a torrent tracker? doesnt bitcoind look for irc if it's not found then to a common list? Isn't that common list open sourced?
340 2011-02-04 05:07:09 <luke-jr> draginx: that list cannot be copyrighted
341 2011-02-04 05:07:20 <luke-jr> and Bitcoin shares no connection to bittorrent
342 2011-02-04 05:07:58 <draginx> luke-jr: sorry, i meant open sourced as in anyone can see where it goes too through the source of your code (you can do MD5 checksums or just git it :P)
343 2011-02-04 05:08:51 <draginx> https://github.com/bitcoin/bitcoin here you go :)
344 2011-02-04 05:14:57 <Diablo-D3> wee
345 2011-02-04 05:15:04 <Diablo-D3> I've found 3 blocks on BPM
346 2011-02-04 05:16:56 <luke-jr> heh, I was just pondering pre-processing block hashes, but it looks like bitcoind already does that :
347 2011-02-04 05:17:14 <luke-jr> can the GPU miners use the midstate, or do they start from scratch every hash?
348 2011-02-04 05:17:23 <Diablo-D3> luke-jr: they use the midstate already
349 2011-02-04 05:17:37 <Diablo-D3> mine just happens to double check before submit by fully hashing
350 2011-02-04 05:18:07 <luke-jr> no offense, but I'm not using yours :P
351 2011-02-04 05:18:14 <Diablo-D3> well, you said gpu miners
352 2011-02-04 05:18:18 <Diablo-D3> mine is a gpu miner
353 2011-02-04 05:18:26 <luke-jr> yeah
354 2011-02-04 05:18:31 <luke-jr> there are others :
355 2011-02-04 05:18:39 <luke-jr> Diablo-D3: I did try yours on my MBP, but it didn't owrk
356 2011-02-04 05:18:47 <luke-jr> probably becuase I don't have *Snow* Leopard
357 2011-02-04 05:18:53 <Diablo-D3> Yes, thats why
358 2011-02-04 05:18:57 <Diablo-D3> no GPU miner will work on yours
359 2011-02-04 05:19:13 <luke-jr> oh well
360 2011-02-04 05:19:24 <Diablo-D3> and if yours has hybrid chips (which it probably dosent because it didnt some with 10.6), mine wont work either
361 2011-02-04 05:26:24 <jgarzik> does the diablo miner check more than high 32 bits of target?
362 2011-02-04 05:26:36 <Diablo-D3> in the kernel? no
363 2011-02-04 05:26:39 <Diablo-D3> in the host? yes
364 2011-02-04 05:26:44 <jgarzik> And it appears that m0mchil checks high 64 bits of target, if I'm reading the code correctly.
365 2011-02-04 05:27:02 <Diablo-D3> m0's kernel checks H in the kernel, G in the host
366 2011-02-04 05:27:32 <Diablo-D3> I check H in the kernel, H and G in the host, and also check if the whole hash actually works in the host
367 2011-02-04 05:33:33 <luke-jr> does m0's miner tell me if it finds a block? :P
368 2011-02-04 05:35:00 <luke-jr> hmmm
369 2011-02-04 05:37:01 <Diablo-D3> yes
370 2011-02-04 05:39:21 <luke-jr> it's cute how all the possible cheats are secured properly :p
371 2011-02-04 05:39:29 <luke-jr> and/or implemented
372 2011-02-04 05:39:50 <luke-jr> 'midstate' is already done
373 2011-02-04 05:40:32 <luke-jr> setting my clock 1.5 hours into the future won't cause others' blocks to be rejected for that long because they're "old", since the clients use a minimum of lastblock-timestamp + 1 for their own timestamps&
374 2011-02-04 05:43:51 <luke-jr> Diablo-D3: since I was on ignore when I implicitly asked the first time: do you think there's a chance your miner could beat 265 MH/s on a 5850?
375 2011-02-04 05:45:58 <Diablo-D3> have you tried it recently?
376 2011-02-04 05:46:11 <luke-jr> never.
377 2011-02-04 05:46:20 <luke-jr> would be an awful pain to get some kind of Java setup
378 2011-02-04 05:46:40 <Diablo-D3> what, apt-get install openjdk6-jre?
379 2011-02-04 05:47:08 <luke-jr> my VM is Gentoo :p
380 2011-02-04 05:48:06 <Diablo-D3> then... emerge... dongs. I dont know.
381 2011-02-04 05:48:09 <Diablo-D3> whatever the gentoo does.
382 2011-02-04 05:48:23 <luke-jr> it probably fails.
383 2011-02-04 05:49:05 <Diablo-D3> ... but its open source GPL'ed software!
384 2011-02-04 05:49:13 <Diablo-D3> you run one command to build it!
385 2011-02-04 05:49:16 <luke-jr> is it? :p
386 2011-02-04 05:49:19 <Diablo-D3> how hard can this be!
387 2011-02-04 05:49:27 <luke-jr> dunno, Java doesn't work like most software
388 2011-02-04 05:49:39 <luke-jr> and Gentoo tries to support multiple JDKs at once
389 2011-02-04 05:49:52 <Diablo-D3> well you dont need MULTIPLE jdks
390 2011-02-04 05:49:54 <Diablo-D3> you need one
391 2011-02-04 05:50:02 <luke-jr> but Gentoo still does the same crap for one
392 2011-02-04 05:50:31 <luke-jr> maybe I'll clone my VM someday to try it&
393 2011-02-04 05:50:55 <luke-jr> wanna setup another VM to try PCSX2 in anyway
394 2011-02-04 05:52:50 <OneFixt> who here knows m0m's miner inside-out?
395 2011-02-04 05:54:32 <gribble> m0mchil was last seen in #bitcoin-dev 1 week, 0 days, 13 hours, 55 minutes, and 55 seconds ago: <m0mchil> pool is diff = 1
396 2011-02-04 05:54:32 <lfm> ;;seen m0mchil
397 2011-02-04 05:54:49 <lfm> you missed him
398 2011-02-04 06:15:09 <OneFixt> Diablo-D3: ping
399 2011-02-04 06:24:35 <Diablo-D3> what
400 2011-02-04 06:31:52 <andrew12> heh
401 2011-02-04 06:32:23 <Diablo-D3> OneFixt: what
402 2011-02-04 06:33:54 <OneFixt> Diablo-D3: I have a question about your miner
403 2011-02-04 06:34:11 <Diablo-D3> it helps if you ask it
404 2011-02-04 06:34:25 <OneFixt> you said it does H==0 checks in the kernel + winning hash check in the program?
405 2011-02-04 06:34:32 <Diablo-D3> yes
406 2011-02-04 06:34:54 <OneFixt> can a h==0 hash overwrite another h==0 hash, like in m0m's?
407 2011-02-04 06:35:09 <OneFixt> ie, do you loop through many threads before checking whether the hash is a winner?
408 2011-02-04 06:35:18 <Diablo-D3> yes, but the chances of seeing two interesting h==0 at the same time is incredibly low
409 2011-02-04 06:35:36 <OneFixt> how many threads run before you check the h==0?
410 2011-02-04 06:35:49 <Diablo-D3> whatever the current global run length is
411 2011-02-04 06:36:09 <OneFixt> i'll take m0m's as an example, and use a 350MH card
412 2011-02-04 06:36:19 <Diablo-D3> yeah, but its a function of time
413 2011-02-04 06:36:34 <OneFixt> let's take 1 sec kernel run which is default i think
414 2011-02-04 06:36:42 <Diablo-D3> -f 60 is the default
415 2011-02-04 06:36:57 <ArtForz> back
416 2011-02-04 06:36:59 <Diablo-D3> but on -f 1, its either 350 million or 350/3
417 2011-02-04 06:37:00 <Diablo-D3> I forget
418 2011-02-04 06:37:03 <OneFixt> ah, art!
419 2011-02-04 06:37:04 <luke-jr> I use -f 1
420 2011-02-04 06:37:05 <ArtForz> OneFixt: checked the math, you're right
421 2011-02-04 06:37:09 <OneFixt> phew
422 2011-02-04 06:37:14 <Diablo-D3> math on what?
423 2011-02-04 06:37:17 <ArtForz> about 3.5% loss with 350M trys/run
424 2011-02-04 06:37:18 <OneFixt> yeah the sim confirmed it, but no point in looking at it
425 2011-02-04 06:37:43 <Diablo-D3> OneFixt: art's miner has large output buffers to store this
426 2011-02-04 06:38:13 <ArtForz> my original kernel used one "space" per 1024 nonces
427 2011-02-04 06:38:23 <ArtForz> so chances of a dupe were pretty damn low
428 2011-02-04 06:38:35 <Diablo-D3> I probably should switch to something similar for one reason
429 2011-02-04 06:38:39 <OneFixt> yep, true
430 2011-02-04 06:38:45 <Diablo-D3> pool share stuffing.
431 2011-02-04 06:39:46 <ArtForz> and even with a 2* H==0 in one spot I don't lose nonces
432 2011-02-04 06:40:08 <OneFixt> yeah, 7 weeks of runtime to get one h==0 at 1024
433 2011-02-04 06:40:11 <ArtForz> I recheck all 1024 in that area on the host to find the H==0 ones
434 2011-02-04 06:40:23 <OneFixt> ah
435 2011-02-04 06:40:47 <Diablo-D3> pool share stuffing should be hilarious
436 2011-02-04 06:40:48 <ArtForz> becasuse for some reason flag=1 is faster than flag=cur_nonce
437 2011-02-04 06:41:13 <Diablo-D3> ArtForz: because there are two magical registers, one holds 0, one holds 1
438 2011-02-04 06:41:17 <OneFixt> funny, but that does remove the inefficiency
439 2011-02-04 06:41:19 <Diablo-D3> since they are so fucking common
440 2011-02-04 06:41:33 <Diablo-D3> luke-jr: yes
441 2011-02-04 06:41:36 <Diablo-D3> luke-jr: but not as shares
442 2011-02-04 06:41:41 <luke-jr> oh, I suppose
443 2011-02-04 06:41:44 <ArtForz> rechecking 1024 out of 2**32 nonces on the host is less slowdown than the hit from flag=nonce
444 2011-02-04 06:42:20 <OneFixt> yeah that makes sense somewhat
445 2011-02-04 06:42:24 <ArtForz> 1 CPU hash for every 4M GPU hashes
446 2011-02-04 06:43:01 <OneFixt> with the miners currently being used by most people though, this thing is a problem
447 2011-02-04 06:43:07 <ArtForz> yep
448 2011-02-04 06:43:10 <ArtForz> especially m0s
449 2011-02-04 06:43:11 <OneFixt> i've noticed slush's pool underperforming lately
450 2011-02-04 06:43:13 <OneFixt> yep
451 2011-02-04 06:43:15 <ArtForz> diablos is less affected
452 2011-02-04 06:43:17 <luke-jr> this radeon is heating up my PC :/
453 2011-02-04 06:43:29 <Diablo-D3> hmm
454 2011-02-04 06:43:34 <Diablo-D3> a 8mb buffer could do like
455 2011-02-04 06:43:35 <luke-jr> watch it screw up my motherboard's SATA sockets 9
456 2011-02-04 06:43:39 <ArtForz> with -f 60 its only like 6M nonces/kernel
457 2011-02-04 06:43:52 <Diablo-D3> 8/4*1024 hashes?
458 2011-02-04 06:44:32 <OneFixt> so .058%
459 2011-02-04 06:44:49 <ArtForz> I use 1MB buffer size
460 2011-02-04 06:44:59 <Diablo-D3> ArtForz: yeah, but I use variably sized kernels
461 2011-02-04 06:45:05 <Diablo-D3> so the cap has to be big enough
462 2011-02-04 06:45:29 <OneFixt> still, we should fix the code up for those who run -f 1
463 2011-02-04 06:45:33 <ArtForz> 1MB = 256M nonces per kernel run
464 2011-02-04 06:45:48 <Diablo-D3> heh
465 2011-02-04 06:46:01 <Diablo-D3> well wait
466 2011-02-04 06:46:05 <Diablo-D3> thats still awfully low
467 2011-02-04 06:46:08 <ArtForz> 1024 nonces = 4 byte, so 1 byte = 256 nonces
468 2011-02-04 06:46:08 <Diablo-D3> since
469 2011-02-04 06:46:27 <Diablo-D3> -f 1 on a 256 mhash card maxes that out
470 2011-02-04 06:46:32 <ArtForz> yes, thats about -f 1.5 on a 5970
471 2011-02-04 06:46:57 <Diablo-D3> so I could easily do a 4 MB buffer
472 2011-02-04 06:47:01 <ArtForz> yep
473 2011-02-04 06:47:02 <Diablo-D3> and future proof it for the next few years
474 2011-02-04 06:47:16 <Diablo-D3> I cant remember what the max is on 4xxx
475 2011-02-04 06:47:23 <Diablo-D3> because I ran into that on my split kernel shit
476 2011-02-04 06:47:27 <Diablo-D3> I think its 16
477 2011-02-04 06:47:36 <ArtForz> that should be good for up to 2**30 nonces/clenqueue
478 2011-02-04 06:48:11 <ArtForz> = about 1Gh/s at -f 1
479 2011-02-04 06:48:22 <ArtForz> yeah, should be good enough for a while
480 2011-02-04 06:49:05 <ArtForz> I'd limit max workitems just in case
481 2011-02-04 06:49:42 <ArtForz> and btw, I dont like dynamically adjusting #workitems, seems to like to cause weird oscillations
482 2011-02-04 06:51:36 <ArtForz> = tried it, caused some issues - "fuck this, just use a cmdline param"
483 2011-02-04 06:53:02 <OneFixt> agreed
484 2011-02-04 06:57:09 <prax> is freenode like an all linux thing?
485 2011-02-04 06:58:09 <prax> errr no huh
486 2011-02-04 07:01:00 <luke-jr> just majority
487 2011-02-04 07:01:16 <prax> k
488 2011-02-04 07:01:30 <prax> I used some flavor of linux only once heh
489 2011-02-04 07:01:43 <prax> thats what most of you guys here are running eh?
490 2011-02-04 07:03:08 <Diablo-D3> ArtForz: it doesnt seem to oscillate much for me
491 2011-02-04 07:04:23 <ArtForz> try taunting it
492 2011-02-04 07:04:37 <Diablo-D3> hurr
493 2011-02-04 07:06:02 <ArtForz> run at lowish -f, then drag windows around, run flash, ..., you know, the usual candidates for stressing X
494 2011-02-04 07:06:50 <Diablo-D3> yeah which is why I use -f 1000
495 2011-02-04 07:07:01 <Diablo-D3> it wildly jumps all over the place to fill in any extra time
496 2011-02-04 07:51:01 <Diablo-D3> huh
497 2011-02-04 07:56:05 <CIA-98> DiabloMiner: Patrick McFarland master * ra7d67f2 / src/main/java/com/diablominer/DiabloMiner/DiabloMiner.java :
498 2011-02-04 08:00:38 <genjix> im sending my laptop in for servicing... im paranoid of them setting up a rootkit or something
499 2011-02-04 08:00:54 <genjix> should i hash my entire hardrive to check for changes before and after?
500 2011-02-04 08:00:56 <Diablo-D3> genjix: how do you know they didnt already?
501 2011-02-04 08:00:58 <genjix> how can i do that?
502 2011-02-04 08:01:10 <genjix> i haven't sent it yet... i will do it.
503 2011-02-04 08:01:13 <Diablo-D3> and no, you should just make a duplicate of your entire hd
504 2011-02-04 08:01:38 <Diablo-D3> since they could accidently destroy your original drive
505 2011-02-04 08:02:05 <genjix> don't think so... it's just a simple fix (solder new power jack)
506 2011-02-04 08:02:27 <Diablo-D3> yeah but you know how stupid techs are
507 2011-02-04 08:02:42 <Diablo-D3> copy the entire drive and remove any sensitive data from the drive first.
508 2011-02-04 08:03:03 <genjix> ok thanks
509 2011-02-04 08:03:34 <genjix> ill copy sentitive info, delete them then reinstall once it's back
510 2011-02-04 08:04:00 <xelister> hey genjix I have a solution for you, want to hear it?
511 2011-02-04 08:04:30 <genjix> tell tell
512 2011-02-04 08:06:06 <Diablo-D3> heh good going slush
513 2011-02-04 08:06:28 <genjix> xelister: what for? :)
514 2011-02-04 08:06:42 <Diablo-D3> oh right I still have him on ignore
515 2011-02-04 08:07:08 <Diablo-D3> xelister: you may now speak
516 2011-02-04 08:09:26 <xelister> genjix: ok
517 2011-02-04 08:09:28 <xelister> They're climbin in your windows, They're snatchin your income tax
518 2011-02-04 08:09:34 <xelister> tryna enslave you so y'all need to
519 2011-02-04 08:09:52 <xelister> 1) hide your kids, hide your wife, buy a gun , buy a knife
520 2011-02-04 08:09:54 <xelister> 2) watch my show, buy your gold, and buy your silver cuz they enslavin errbody out here
521 2011-02-04 08:09:55 <xelister> 3) We don't have to look for no facts, we know this is true ...
522 2011-02-04 08:09:57 <xelister> 4) they gon chip you, they gon tag you
523 2011-02-04 08:10:10 <Diablo-D3> zomg xel has turned black
524 2011-02-04 08:10:14 <genjix> ahh you think i'm overly paranoid
525 2011-02-04 08:10:20 <xelister> :>
526 2011-02-04 08:10:42 <genjix> except i read on 4chan a techie posting that they copy everybody's harddrives and read all their stuff
527 2011-02-04 08:10:58 <genjix> and tbh it seems like something i'd do
528 2011-02-04 08:11:10 <xelister> lolz..
529 2011-02-04 08:11:11 <xelister> :)
530 2011-02-04 08:11:17 <genjix> out of curiousity, not malice
531 2011-02-04 08:11:35 <genjix> (look for nudey pics, gossip, email passwords .etc)
532 2011-02-04 08:11:53 <xelister> ok and seriously though.  Right cleaning hdd before is fine I guess. Also you could want to have laptop fixed in front of you, depending on type of shop-service and so on
533 2011-02-04 08:12:00 <genjix> esp if you're a bored techie sitting doing nothing for hours
534 2011-02-04 08:12:16 <xelister> "out of curiousity, not malice (look for nudey pics, gossip, email passwords .etc)" .... lololol
535 2011-02-04 08:12:21 <genjix> kk, i'll do that too
536 2011-02-04 08:13:01 <xelister> then what is malice, installing some plutonium there to give the owner a cancer? ;)
537 2011-02-04 08:13:51 <genjix> stealing their credit cards, uploading child porn
538 2011-02-04 08:14:40 <genjix> those actually cause real harm (loss of fundz, prison time)
539 2011-02-04 08:14:41 <Diablo-D3> xelister: wernt you asking for a double meter anyhow?
540 2011-02-04 08:14:41 <xelister> a techie "just" browsing through my hdd without my conesnt would find himself quickly missing left right, and would hight his left, arm, in his anus area
541 2011-02-04 08:15:02 <xelister> woah typos. need coffey
542 2011-02-04 08:15:09 <xelister> a techie "just" browsing through my hdd without my conesnt would find himself quickly missing left, and would find his right, arm, in his anus area
543 2011-02-04 08:15:17 <genjix> everything else is just "oh noes my reputation is damaged" or "oh noes my secret plans are exposed"
544 2011-02-04 08:15:21 <Diablo-D3> typos continue.
545 2011-02-04 08:15:23 <genjix> i.e nobody cares
546 2011-02-04 08:15:52 <xelister> speaking of reputation ruined
547 2011-02-04 08:16:10 <xelister> anyone knows if Ati/AMD plans to release a non-shit drivers for 5xxx in near future?
548 2011-02-04 08:16:35 <genjix> xelister: expect big changes in near future
549 2011-02-04 08:16:41 <genjix> X is being dumped now
550 2011-02-04 08:16:52 <genjix> new minimal interface is coming in the kernel :)
551 2011-02-04 08:16:53 <xelister> like, ones that do not crash in 75% of cases of switching virtual displays / terminals while using 2 video outputs (2 X desktops) on both 5970 and 5770 (wtf ati)
552 2011-02-04 08:17:08 <genjix> it'll be sweet. performance++
553 2011-02-04 08:17:15 <xelister> ot that do not 100% of the case show garbled mouse cursor on 2nd display
554 2011-02-04 08:17:20 <genjix> at last we'll see input redirection
555 2011-02-04 08:17:22 <xelister> and that do not hang each 3-5 days
556 2011-02-04 08:17:34 <xelister> genjix: hm??
557 2011-02-04 08:17:45 <xelister> Diablo-D3: what double meter?
558 2011-02-04 08:17:55 <Diablo-D3> xelister: short avg/forever meter
559 2011-02-04 08:17:58 <genjix> xelister: X is being deprecated as a project
560 2011-02-04 08:18:01 <xelister> ah yea
561 2011-02-04 08:18:05 <genjix> *has been
562 2011-02-04 08:18:25 <xelister> well I heavly modified it anyway to have various stats too.. is there anything very important to merge sine 2010.11's versions?
563 2011-02-04 08:18:35 <xelister> like important security / stability / performance (>1%) fix?
564 2011-02-04 08:18:37 <Diablo-D3> xelister: yes
565 2011-02-04 08:18:45 <Diablo-D3> xelister: I added art's kernel, for one
566 2011-02-04 08:18:50 <Diablo-D3> 3% increase
567 2011-02-04 08:18:53 <xelister> oh
568 2011-02-04 08:19:09 <Diablo-D3> and I made it more slush-proof
569 2011-02-04 08:19:20 <xelister> what was needed for slush?
570 2011-02-04 08:19:28 <Diablo-D3> nothing much
571 2011-02-04 08:19:34 <Diablo-D3> he just managed to often crash my miner
572 2011-02-04 08:19:46 <Diablo-D3> since he fed it shit
573 2011-02-04 08:20:06 <xelister> how often? mine doesnt ;)   unless you mean occassionall cant allocate queue but that is probably due to desktop/X usage at same time as miner runs
574 2011-02-04 08:20:44 <Diablo-D3> basically whenever he took the pool down
575 2011-02-04 08:20:50 <Diablo-D3> it wouldnt handle it correctly in rare cases
576 2011-02-04 08:21:04 <xelister> ah
577 2011-02-04 08:21:15 <Diablo-D3> oh, and I added optional compressed traffic stuff on his request
578 2011-02-04 08:21:20 <Diablo-D3> so it doesnt eat all his bandwidth
579 2011-02-04 08:21:46 <xelister> genjix: hmm... I don't get how X is deprecated... anyway, radeons 5xxx are NOT usable on linux for desktop it seems, crashes x10 more often then nvidia, literaly
580 2011-02-04 08:22:39 <Diablo-D3> radeons are fine on linux
581 2011-02-04 08:22:48 <Diablo-D3> almost everyone here owns one
582 2011-02-04 08:22:50 <Diablo-D3> and it works fine
583 2011-02-04 08:22:52 <xelister> miner <---> bitcoind/pool is basically just asking for sha256 for base of current calculation and thats all?
584 2011-02-04 08:23:07 <Diablo-D3> xelister: yes, it just does getwork
585 2011-02-04 08:23:25 <jav> quick question: in a transaction the total amount of bitcoins of all IN addresses are moved to the OUT addresses, right? .. and everything that isn't allocated to a new OUT address is considered a fee? is that correct?
586 2011-02-04 08:23:37 <xelister> Diablo-D3: but do you use radeon on desktop?  try switching X alt-ctrl-del -f6- f7 -f8 etc few times forth and back. do it 10 times - crash,  do it 2 times with dualdisaply - guaranteed crash
587 2011-02-04 08:23:44 <xelister> erm. no -del
588 2011-02-04 08:25:11 <Diablo-D3> xelister: it'll onl do that if KMS is on
589 2011-02-04 08:25:17 <ArtForz> fglrx + fbcon + kms = bad idea
590 2011-02-04 08:25:30 <Diablo-D3> fbcon alone doesnt do it from what I can tell
591 2011-02-04 08:25:31 <Diablo-D3> just kms
592 2011-02-04 08:26:17 <xelister> how to check if kms is active for me?
593 2011-02-04 08:26:25 <Diablo-D3> if you didnt turn it off, its active
594 2011-02-04 08:26:38 <xelister> perhaps ubuntu did or something... is it an boot option?
595 2011-02-04 08:27:49 <Diablo-D3> yes its a boot option, and no ubuntu hasnt disabled it
596 2011-02-04 08:28:32 <xelister> btw, I hear someone has 200 USD/card radeon's 5970 ;)  you think this could be possible if the cards would come from factory that perhaps produced a bit more then they agreed with Ati overloard or that coworker borrowed some chips home or something?
597 2011-02-04 08:28:40 <Diablo-D3> iirc its radeon.modeset=0
598 2011-02-04 08:29:28 <Diablo-D3> oh wait, its just nomodeset
599 2011-02-04 08:30:31 <joe_1> when a bitcoin client receives a pair of double-spend transactions A then B, it refuses the B transaction no matter how many more blocks B has over A?
600 2011-02-04 08:31:57 <edcba> transactions don't have blocks
601 2011-02-04 08:32:11 <edcba> blocks have transactions
602 2011-02-04 08:32:30 <joe_1> yeah but for any transaction i can say how many blocks of confirmation it has. it belongs to a block, and that block is X blocks deep.
603 2011-02-04 08:32:32 <edcba> while not acknowledged transactions aren't tied to any block
604 2011-02-04 08:33:04 <edcba> you can't have same transaction in 2 different blocks in same chain
605 2011-02-04 08:34:21 <xelister> Kernel mode-setting (KMS)  [...]  promises to make booting faster, more graphical, and less flickery.
606 2011-02-04 08:34:38 <xelister> ^---- wtf more graphical boot? sound's like "I'm an iFag" to me ;)
607 2011-02-04 08:34:53 <Diablo-D3> xelister: no
608 2011-02-04 08:35:01 <Diablo-D3> also, you do realize its ALREADY graphical, right?
609 2011-02-04 08:35:03 <Diablo-D3> its fbcon
610 2011-02-04 08:35:07 <xelister> I assumed KMS will be a good idea (one day?) because it makes the linux guys takes care of mode setting instead of nvidia/radeon noobs
611 2011-02-04 08:35:15 <Diablo-D3> you get little tux in corner of fbcon
612 2011-02-04 08:35:17 <Diablo-D3> ergo graphical
613 2011-02-04 08:35:27 <Diablo-D3> xelister: KMS is literally what it says
614 2011-02-04 08:35:34 <Diablo-D3> the kernel inits the hardware in full
615 2011-02-04 08:35:42 <Diablo-D3> no more X initing hardware
616 2011-02-04 08:35:55 <Diablo-D3> which also will lead to no more X as root
617 2011-02-04 08:35:56 <xelister> Diablo-D3: when I seen ubuntu's faggotry of showing me grahpical mode bootup console (aka "scrolling black-white text thing") I switched it off..  actually, just changing to propertiary driver changes that It seemed
618 2011-02-04 08:36:03 <Diablo-D3> ubuntu is not linux
619 2011-02-04 08:36:09 <Diablo-D3> please stop spreading that myth
620 2011-02-04 08:36:10 <joe_1> it's not?
621 2011-02-04 08:36:17 <Diablo-D3> no, its a microsoft product
622 2011-02-04 08:36:23 <Diablo-D3> it has nothing to do with linux
623 2011-02-04 08:36:32 <Diablo-D3> no more fud in #bitcoin-dev plzkthx
624 2011-02-04 08:36:47 <xelister> you are overdoing it ;)
625 2011-02-04 08:37:10 <xelister> we know it sucks and all, but still better then people running windows
626 2011-02-04 08:37:30 <Diablo-D3> dude
627 2011-02-04 08:37:31 <joe_1> ok if ubuntu is gay then what linux distribution is real linux
628 2011-02-04 08:37:34 <Diablo-D3> if I wanted to have a graphical boot
629 2011-02-04 08:37:36 <Diablo-D3> I'd run osx
630 2011-02-04 08:37:41 <Diablo-D3> joe_1: debian
631 2011-02-04 08:37:47 <Diablo-D3> although in the next year or so Im switching to alpine
632 2011-02-04 08:37:57 <xelister> wtf is alpine
633 2011-02-04 08:38:01 <xelister> joe_1: gentoo?
634 2011-02-04 08:38:37 <Diablo-D3> xelister: a new linux distro that is meant to be easy to use, fast, bloat free, and 9000% less faggotry than ubuntu
635 2011-02-04 08:39:18 <ArtForz> well, making a distro better than "let's take sid and randomly break crap" isn't exactly rocket surgery
636 2011-02-04 08:39:38 <xelister> what is appealing in ubuntu
637 2011-02-04 08:39:47 <xelister> other then graphicall faggotry (sadly)
638 2011-02-04 08:40:03 <xelister> is that so much stuff Just Works,  like lots of firmware/drivers/auto install/auto detect/auto whatever
639 2011-02-04 08:40:12 <Diablo-D3> xelister: ITS EXTRA AFRICAN
640 2011-02-04 08:40:27 <Diablo-D3> there. I said it. Ubuntu is the distro for black people.
641 2011-02-04 08:40:48 <xelister> :O
642 2011-02-04 08:40:54 <Diablo-D3> Linux is finally racist.
643 2011-02-04 08:41:01 <ArtForz> errr... the same shit generally works on sid once you enable the nonfree repo
644 2011-02-04 08:41:03 <Diablo-D3> I hope you're happy.
645 2011-02-04 08:41:04 <xelister> ArtForz: perhaps
646 2011-02-04 08:41:09 <joe_1> i felt like a thousand pound weight was lifted off my chest. i knew it was african
647 2011-02-04 08:41:10 <joe_1> that drum beat when it loads
648 2011-02-04 08:41:12 <xelister> ArtForz: debian pissed me off long ago, when its sofware was 100 years old behind upstream
649 2011-02-04 08:41:17 <ArtForz> and sid at least doesn't ship a fucking RELEASE with alpha-level fglrx
650 2011-02-04 08:41:21 <xelister> (actually 1.5 year old)
651 2011-02-04 08:41:32 <xelister> do they /finally/ improved on that?
652 2011-02-04 08:41:58 <ArtForz> ?
653 2011-02-04 08:42:19 <ArtForz> if you run stable, no
654 2011-02-04 08:42:22 <xelister> in like 2007, some libs like afair.. libboost? where 1 .. 1.5 years older then the upstream version was
655 2011-02-04 08:42:28 <xelister> 1.5 years is *really* old
656 2011-02-04 08:42:34 <ArtForz> because stable is meant to be... you know. stable.
657 2011-02-04 08:43:20 <xelister> but waiting till say 2013 to use things the developers made (e.g. on my requests/bugreports) today, or that I read about today, is meh
658 2011-02-04 08:43:23 <ArtForz> if you want current, use unstable aka sid
659 2011-02-04 08:43:33 <Diablo-D3> also, you do realize libboost is matched to the version of gcc, right?
660 2011-02-04 08:43:41 <Diablo-D3> its not recommended to mismatch libboost and gcc
661 2011-02-04 08:43:48 <Diablo-D3> random shit goes bad
662 2011-02-04 08:43:52 <Diablo-D3> because libboost coders suck dick
663 2011-02-04 08:44:17 <xelister> well then ship also not superold version of gcc
664 2011-02-04 08:44:20 <ArtForz> not to mention debian stable is pretty much what other distros would consider a LTS release
665 2011-02-04 08:44:23 <xelister> so you tell me now this is possible on SID?
666 2011-02-04 08:44:29 <xelister> My debian experience was like
667 2011-02-04 08:44:56 <Diablo-D3> xelister: you are coming very close to me putting you back on ignore
668 2011-02-04 08:45:39 <xelister> wow, linux ^_^ ... wait,  oh shit thats old   -- "update to SID, dude" ---> ok. yey new versions ^_^ ... wait, everything broken on update, cant reinstall or unsintall or upgrade now X-( what to do? --- "we told you to not use SID, fuck you" --> fuuuuuuuuuuu
669 2011-02-04 08:45:54 <sipa> haha
670 2011-02-04 08:46:08 <Diablo-D3> And there he goes.
671 2011-02-04 08:46:29 <ArtForz> yes, lets use ubuntu, where upgrading from one *release* to the next is pretty much guaranteed to cause even more issues
672 2011-02-04 08:46:32 <xelister> adn I didnt tried since, so is it nowdays possible to have more cutting endge Debian, but that will overall not break horribly?
673 2011-02-04 08:47:02 <ArtForz> and btw, I didnt have ONE SINGLE PROBLEM with sid since the ia32-libs fiasco
674 2011-02-04 08:47:09 <edcba> try it and don't annoy us with that xelister ;)
675 2011-02-04 08:47:33 <xelister> hm perahaps perhaps
676 2011-02-04 08:48:46 <xelister> so again, which version of debian makes sense... so I have rather recent versions, but also apt will not get into unusable state where basically only way to fix it, without dicking all day with it, is to reinstall
677 2011-02-04 08:49:42 <edcba> just ask in #debian
678 2011-02-04 08:51:30 <ArtForz> so, what version of ubuntu should I use that doesnt have broken sound support, NetworkMangler and fglrx drivers that arent like 2 years old or ubuntu-exclusive preview aka alpha releases?
679 2011-02-04 08:52:00 <joe_1> Alpine Linux began as a fork of the LEAF Project
680 2011-02-04 08:52:32 <Diablo-D3> joe_1: yeah, but thats a pretty bad way of describing it
681 2011-02-04 08:52:41 <Diablo-D3> it really is a debian-like distro without the faggotry
682 2011-02-04 08:52:49 <Diablo-D3> like, they got rid of glibc
683 2011-02-04 08:52:56 <Diablo-D3> no more libc bloat fest
684 2011-02-04 08:53:01 <xelister> ArtForz: Im not saying that ubuntu is better ;)
685 2011-02-04 08:53:10 <xelister> it has different kind of problems
686 2011-02-04 08:53:35 <joe_1> whats wrong with libc, don't you need that for.. everything?
687 2011-02-04 08:53:37 <xelister> hm... perhaps source based is the answer. Anyway then we can have cutting-edge & working-with-upstream-developers versions (more naturally)
688 2011-02-04 08:54:06 <ArtForz> aka gentoo
689 2011-02-04 08:54:33 <ArtForz> aka lets recompile everything because libc changed again
690 2011-02-04 08:54:33 <Diablo-D3> joe_1: yes, but they swapped it with uclibc
691 2011-02-04 08:54:50 <Diablo-D3> ArtForz: yes, it has the unfortunate side effect of breaking existing binaries
692 2011-02-04 08:54:52 <Diablo-D3> they dont care.
693 2011-02-04 08:54:58 <Diablo-D3> and honestly, I dont care either
694 2011-02-04 08:55:03 <Diablo-D3> its all legacy software
695 2011-02-04 08:55:21 <Diablo-D3> most linux devices dont even use glibc
696 2011-02-04 08:55:27 <Diablo-D3> its too bloated and slow
697 2011-02-04 08:57:29 <xelister> ArtForz: then what you use?
698 2011-02-04 08:57:52 <ArtForz> sid on my main workstation/dev box
699 2011-02-04 08:58:14 <ArtForz> stable+security on my servers
700 2011-02-04 08:58:16 <Diablo-D3> ArtForz: and no, its not like gentoo
701 2011-02-04 08:58:20 <Diablo-D3> the libc doesnt change all the time
702 2011-02-04 08:58:22 <Diablo-D3> its a one time change
703 2011-02-04 08:58:31 <Diablo-D3> its no different than how everyone had to deal with libc5->6 all those years ago
704 2011-02-04 08:59:17 <joe_1> i thought binaries don't change unless the headers change. why would the c standard library headers change. i can understand implementation changes
705 2011-02-04 08:59:24 <ArtForz> my notebook is also sid
706 2011-02-04 08:59:24 <genjix> ubuntu switched because the owner of libc is an idiot
707 2011-02-04 08:59:34 <xelister> :)
708 2011-02-04 08:59:46 <Diablo-D3> genjix: no
709 2011-02-04 08:59:54 <Diablo-D3> ubuntu switched to eglibc because debian did
710 2011-02-04 08:59:58 <Diablo-D3> its STILL glibc
711 2011-02-04 09:00:09 <genjix> oh right
712 2011-02-04 09:00:21 <genjix> but it's a different project (different maintainer) afaik
713 2011-02-04 09:01:05 <xelister> ArtForz: but can you even get the dependencies to build bitcoind on stable, and bitcoin gui on sid
714 2011-02-04 09:01:13 <ArtForz> yes
715 2011-02-04 09:01:31 <ArtForz> well, you still have to compile the custom wx for gui
716 2011-02-04 09:01:32 <xelister> was it not a problem back in 2010 end
717 2011-02-04 09:01:40 <xelister> yeah that is what I ment, wx 2.9
718 2011-02-04 09:02:01 <Diablo-D3> different maintainer, essentially same project
719 2011-02-04 09:02:05 <Diablo-D3> its not a perminent fork
720 2011-02-04 09:02:06 <xelister> ubuntu did shiped 2.9
721 2011-02-04 09:02:13 <Diablo-D3> they constantly update from glibc upstream
722 2011-02-04 09:02:33 <ArtForz> except iirc ubuntus 2.9 is configured differently than what bitcoin expects
723 2011-02-04 09:03:00 <xelister> ArtForz: bitcoin gui does build on current ubuntu with no manuall building on dependencies
724 2011-02-04 09:03:03 <genjix> tbh ubuntu is still the best linux around
725 2011-02-04 09:03:05 <xelister> *of
726 2011-02-04 09:03:06 <genjix> linux done right
727 2011-02-04 09:03:34 <genjix> gentoo is pita
728 2011-02-04 09:03:53 <genjix> arch is wip
729 2011-02-04 09:03:57 <Diablo-D3> ArtForz: ehehehe on -f 1000 Im getting 75.6 long count
730 2011-02-04 09:04:08 <genjix> fedora is 2nd rate
731 2011-02-04 09:04:12 <xelister> ArtForz: this is some advantage. Building X to build Y to build Z to finally build the project you want to co-develop / cut-edge, is not so good sometimes
732 2011-02-04 09:04:31 <joe_1> that's where windows comes in
733 2011-02-04 09:04:44 <xelister> overall I like building and all but not when I have no time to play around
734 2011-02-04 09:04:58 <genjix> lol u like building program
735 2011-02-04 09:05:05 <genjix> strange person
736 2011-02-04 09:05:17 <xelister> actually, on what satoshi develops that he used wx 2.9 ... but perhaps 2.9 was needed to correctly handle unicode?
737 2011-02-04 09:05:25 <Diablo-D3> ArtForz: I used to have to -f 1 to get that
738 2011-02-04 09:05:28 <genjix> weee look at that incomprehensible text scrolling quickly past...
739 2011-02-04 09:05:37 <Diablo-D3> ArtForz: which leads me to believe, on my box, I lose 3% by keeping interactivity
740 2011-02-04 09:06:33 <Diablo-D3> which, meh, fine
741 2011-02-04 09:06:50 <Diablo-D3> the floor is now so low after the newest modifications I think I'll be able to game without shutting the miner off
742 2011-02-04 09:07:05 <slush> ;;bc,stats
743 2011-02-04 09:07:07 <gribble> Current Blocks: 106131 | Current Difficulty: 22012.4941572 | Next Difficulty At Block: 106847 | Next Difficulty In: 716 blocks | Next Difficulty In About: 4 days, 9 hours, 12 minutes, and 44 seconds | Next Difficulty Estimate: 24615.15663506
744 2011-02-04 09:07:34 <ArtForz> also, debian is at least mostly cross-platform
745 2011-02-04 09:08:13 <da2ce7> ;;lastseen noagendamarket
746 2011-02-04 09:08:14 <gribble> Error: "lastseen" is not a valid command.
747 2011-02-04 09:08:18 <da2ce7> ;;last,seen noagendamarket
748 2011-02-04 09:08:20 <gribble> Error: "last,seen" is not a valid command.
749 2011-02-04 09:10:24 <genjix> ;;seen noagendamarket
750 2011-02-04 09:10:24 <gribble> noagendamarket was last seen in #bitcoin-dev 19 hours, 37 minutes, and 19 seconds ago: * noagendamarket is on the nod
751 2011-02-04 09:15:34 <joe_1> what do i need to do to start compiling bitcoin under mingw
752 2011-02-04 09:16:16 <genjix> boost mainly
753 2011-02-04 09:16:24 <genjix> libdb (i think)
754 2011-02-04 09:16:43 <genjix> best look in makefiles or do a test run on linux
755 2011-02-04 09:16:51 <genjix> make -f makefile.unix bitcoind
756 2011-02-04 09:17:48 <joe_1> oh ok thanks
757 2011-02-04 09:17:53 <joe_1> found makefile.mingw
758 2011-02-04 09:30:45 <m0mchil> OneFixt, ArtForz: just to clear this out - there is no loss whatsoever returning just one H per kernel run
759 2011-02-04 09:31:02 <m0mchil> regardless of using pool or not
760 2011-02-04 09:31:41 <ArtForz> yes there is
761 2011-02-04 09:31:47 <Diablo-D3> m0mchil: maximum loss would be ~3% on -f 1 with a 5970
762 2011-02-04 09:31:55 <Diablo-D3> because you're going to hit H==0 conflicts
763 2011-02-04 09:31:56 <ArtForz> actually 5870 is worse
764 2011-02-04 09:32:03 <Diablo-D3> and it'll only really matter on the pool
765 2011-02-04 09:32:08 <m0mchil> 5970 - 4 GPUs combined or?
766 2011-02-04 09:32:41 <ArtForz> you run enough globalworkitems to hit 1 sec per clenqueuendrangekernel, right?
767 2011-02-04 09:32:50 <m0mchil> right
768 2011-02-04 09:33:04 <ArtForz> and you only return one H==0 result max from that
769 2011-02-04 09:33:11 <m0mchil> go on
770 2011-02-04 09:33:13 <ArtForz> a 5870 can get like 350Mh/s
771 2011-02-04 09:33:23 <ArtForz> = 350M nonces tested in one call
772 2011-02-04 09:33:57 <Diablo-D3> yeah but 2**32 is 4294 million
773 2011-02-04 09:34:05 <ArtForz> you lose about 3.5% of H==0 solutions due to runs containing 2 or more H==0 nonces
774 2011-02-04 09:34:32 <m0mchil> but there are runs with no Hs in them, right?
775 2011-02-04 09:34:56 <Diablo-D3> yes
776 2011-02-04 09:35:08 <m0mchil> what do you lose of empty runs?
777 2011-02-04 09:35:11 <Diablo-D3> I bet 3.5% of those runs would have no H==0
778 2011-02-04 09:35:35 <Diablo-D3> it seems like it should align that way
779 2011-02-04 09:37:47 <ArtForz> ?
780 2011-02-04 09:37:57 <ArtForz> you dont lose anything of empty runs, tzhey're still empty
781 2011-02-04 09:38:07 <Diablo-D3> ArtForz: I think he means, how many runs turn up empty
782 2011-02-04 09:38:15 <ArtForz> that doesn't change
783 2011-02-04 09:38:31 <Diablo-D3> then... how many runs turn up empty?
784 2011-02-04 09:38:49 <m0mchil> let me do some statistics here... moment
785 2011-02-04 09:38:50 <ArtForz> about 88% of runs iirc
786 2011-02-04 09:39:47 <ArtForz> nm, 91.6%
787 2011-02-04 09:41:48 <ArtForz> of the remaining 8.whatever% of total runs that have at least one H==0, something like 96.6% have exactly one H==0
788 2011-02-04 09:42:47 <ArtForz> = total you only have like 0.28% runs with 2 or more H==0, which doesnt look like much
789 2011-02-04 09:43:30 <ArtForz> but *of the runs that do have a H==0* it's something like 3.4%
790 2011-02-04 09:45:01 <Diablo-D3> ahh so
791 2011-02-04 09:45:02 <Diablo-D3> it
792 2011-02-04 09:45:03 <Diablo-D3> doesnt matter
793 2011-02-04 09:45:07 <ArtForz> ?
794 2011-02-04 09:45:27 <ArtForz> = you *ignore* like 3.5% of H==0 you'd normally find
795 2011-02-04 09:45:41 <ArtForz> thats quite significant
796 2011-02-04 09:45:45 <m0mchil> 3.4% of 0.28% = 0.009%?
797 2011-02-04 09:45:58 <Diablo-D3> m0mchil: no
798 2011-02-04 09:46:07 <Diablo-D3> 0.28% of ALL runs, 3.4% of runs with at least H==0
799 2011-02-04 09:46:40 <ArtForz> 3.4% of what would be your H==0 count if you didnt overwrite a bunch of em
800 2011-02-04 09:47:02 <ArtForz> = you're cutting your effective hashrate by those same 3.whatever%
801 2011-02-04 09:47:06 <ArtForz> less with slower cards of course
802 2011-02-04 09:47:48 <ArtForz> as those do less trys/run = lower chances of >1 H==0 in a run = less lost H==0s
803 2011-02-04 09:48:41 <m0mchil> ok, let me check something... brb
804 2011-02-04 09:48:47 <ArtForz> so it is a pretty serious issue for pool clients
805 2011-02-04 09:50:20 <ArtForz> Diablo-D3: shouldnt your miner show the same behaviour at -f 1 ?
806 2011-02-04 09:53:05 <sipa> couldn't you use multiple output buffers for different kernels running in parallel, to decrease the chance of overwritten found H==0's?
807 2011-02-04 09:53:22 <ArtForz> thats what my original kernel was doing...
808 2011-02-04 09:53:28 <sipa> i see
809 2011-02-04 09:53:30 <ArtForz> used one output dword per 1024 tries
810 2011-02-04 09:53:59 <ArtForz> as the chance of two H==0 solutions in 1024 tries is pretty damn small
811 2011-02-04 09:54:08 <slush> can anybody shortly tell me what are you talking about, guys? I see 'pool' ,'serious issue', 'cutting hashrate', which makes me little nervous
812 2011-02-04 09:54:49 <ArtForz> OneFixt found a problem with how m0s mienr is handing back "good" nonces to host
813 2011-02-04 09:55:11 <ArtForz> it can only hand back one solution per kernel run, which is about 1 second worth of tries
814 2011-02-04 09:55:32 <m0mchil> well, diablo's too
815 2011-02-04 09:55:39 <ArtForz> so if you have a fast card, whenever it finds two or more sultions in one second = one run, it just ignores one
816 2011-02-04 09:55:43 <ArtForz> diablos with -f 1
817 2011-02-04 09:56:35 <sipa> seems not that hard to fix
818 2011-02-04 09:56:56 <ArtForz> with default -f 60 diablos effectively uses 1/60th your run size = a LOT less chance of 2+ H==0 in one run
819 2011-02-04 09:57:31 <tcatm> code to check for nonces within a range: http://pastebin.com/bgh0x6Zi
820 2011-02-04 09:58:09 <m0mchil> mine's -f 30, until recently it was -f 60
821 2011-02-04 09:58:10 <ArtForz> doesnt help
822 2011-02-04 09:58:20 <ArtForz> *checks*
823 2011-02-04 10:00:44 <ArtForz> hmmm... yeah
824 2011-02-04 10:00:59 <ArtForz> still a problem at low -f though
825 2011-02-04 10:01:56 <ArtForz> whats the point of running low -f if you get a few % higher hashrate but start throwing away a few % valid solutions?
826 2011-02-04 10:02:27 <m0mchil> what's the expectation of 340 MHash @ x1 difficulty?
827 2011-02-04 10:03:33 <gribble> The average time to generate a block at 340000 Khps, given the supplied difficulty of 1, is 12 seconds
828 2011-02-04 10:03:33 <lfm> ;;bc,calcd 340000 1
829 2011-02-04 10:03:41 <ArtForz> mean of 0.0791624 H==0 per run
830 2011-02-04 10:03:43 <m0mchil> thanks
831 2011-02-04 10:04:10 <sipa> so, let's find the optimum
832 2011-02-04 10:04:11 <ArtForz> 12.63 sec or so
833 2011-02-04 10:05:00 <ArtForz> thanks to -f 1 lossage it should be above 13 sec
834 2011-02-04 10:05:26 <m0mchil> ok, again, why it is a problem only against pool?
835 2011-02-04 10:05:42 <tcatm> http://pastebin.com/ar8jGTtG my miner that doesn't have the problem and *should* work with art's kernel. It just doesn't suppport getwork
836 2011-02-04 10:06:07 <ArtForz> oh wait, you dont check G in kernel anymore
837 2011-02-04 10:06:18 <ArtForz> so problem should appear always
838 2011-02-04 10:07:29 <ArtForz> well, most obvious with high hashrate per GPU and really low -f
839 2011-02-04 10:07:39 <slush> is
840 2011-02-04 10:07:52 <slush> is 'really low -f' also 5 for 1.1Ghash miner?
841 2011-02-04 10:08:01 <ArtForz> 1.1GH for a single card?
842 2011-02-04 10:08:07 <slush> 2x5970
843 2011-02-04 10:08:10 <slush> so 4 gpus
844 2011-02-04 10:08:12 <ArtForz> thats 4 GPUs
845 2011-02-04 10:08:16 <Diablo-D3> really low -f is 1.
846 2011-02-04 10:08:18 <slush> oh, this is related to gpu
847 2011-02-04 10:08:58 <ArtForz> basically problem is, with the current kernels each gpu can only pass one H==0 solution per "job"
848 2011-02-04 10:09:07 <ArtForz> and one job = 1/framerate seconds
849 2011-02-04 10:09:21 <ArtForz> = that GPUs hashpersec/framerate
850 2011-02-04 10:09:21 <sipa> on one gpu
851 2011-02-04 10:10:36 <slush> oh, see
852 2011-02-04 10:10:56 <slush> so bigger framerate solve this problem, right? Not clearly, but the probability is lower
853 2011-02-04 10:11:01 <ArtForz> yes
854 2011-02-04 10:11:07 <ArtForz> probability is a LOT lower
855 2011-02-04 10:11:28 <sipa> at 340Mh/s and -f 1, i get 3.9% loss
856 2011-02-04 10:11:31 <ArtForz> iirc reduced by more than the same factor you increase framerate
857 2011-02-04 10:11:36 <sipa> correct, ArtForz?
858 2011-02-04 10:11:45 <ArtForz> looks right
859 2011-02-04 10:12:09 <sipa> at -f 5, that becomes 0.79%
860 2011-02-04 10:12:25 <ArtForz> yep
861 2011-02-04 10:12:39 <sipa> and -f 60, 0.066%
862 2011-02-04 10:14:00 <ArtForz> yup
863 2011-02-04 10:14:47 <sipa> so, if you'd change the output buffer to let's say 256 entries, and have each kernel modify output[nonce & 0xFF], you effectively get only 0.015% loss at -f 1
864 2011-02-04 10:14:59 <ArtForz> yup
865 2011-02-04 10:15:29 <tcatm> you could even check nonce..nonce+0xFF in hostcode and have 0% loss at -f 1
866 2011-02-04 10:15:30 <m0mchil> I am still not convinced... does someone tried a long run at diff x1? I checked with my results for last 14 hours to pool, seeing 10% worse than expected results
867 2011-02-04 10:15:44 <ArtForz> I did a few simulations
868 2011-02-04 10:15:46 <sipa> there is a lot of random variation still too
869 2011-02-04 10:15:48 <m0mchil> but this could be because of network overhead
870 2011-02-04 10:16:08 <slush> m0mchil: 10% is more than expected network overhead
871 2011-02-04 10:16:12 <slush> _much_ more
872 2011-02-04 10:16:13 <ArtForz> you need a huge sample size to get anything close to being conclusive for a few % difference
873 2011-02-04 10:16:13 <sipa> network overhead shouldn't be more than a 2% loss
874 2011-02-04 10:17:09 <m0mchil> well, it's only ~4000 samples... I'd like to see something like 100 000
875 2011-02-04 10:18:07 <Diablo-D3> loss in what
876 2011-02-04 10:18:10 <Diablo-D3> performance?
877 2011-02-04 10:18:31 <ArtForz> H==0 results vs. time*hashrate
878 2011-02-04 10:18:55 <sipa> Diablo-D3: loss of submitted shares
879 2011-02-04 10:19:03 <ArtForz> it's pretty obvious when you simulate with REALLY low -f
880 2011-02-04 10:19:22 <ArtForz> like, 0.09something
881 2011-02-04 10:19:23 <sipa> submitted valid counted shares
882 2011-02-04 10:21:38 <ArtForz> basically, what happens when you have a globalworksize of 2**32?
883 2011-02-04 10:21:52 <ArtForz> you get a *mean* of 1 H==0 per run
884 2011-02-04 10:22:13 <ArtForz> = you have quite a lot of runs with more than 1 H==0
885 2011-02-04 10:23:13 <ArtForz> = you lose a lot of H==0 results because you just ignore any > 1 in a run containing multiple results
886 2011-02-04 10:24:55 <sipa> 41.8% of them
887 2011-02-04 10:26:09 <ArtForz> should be a binomial distribution, right?
888 2011-02-04 10:26:42 <sipa> yes
889 2011-02-04 10:26:48 <ArtForz> yep
890 2011-02-04 10:27:42 <sipa> i calculate 1-(P(#(H==0) == 1)/(1-P(#(H==0) == 0)
891 2011-02-04 10:27:48 <sipa> )
892 2011-02-04 10:49:27 <davout> hey all
893 2011-02-04 10:52:41 <tcatm> hey davout
894 2011-02-04 10:52:51 <tcatm> you broke the trade history api
895 2011-02-04 11:04:52 <davout> tcatm: yea, i warned about it a while ago :D
896 2011-02-04 11:05:00 <tcatm> fix it :P
897 2011-02-04 11:05:05 <davout> you fix it !
898 2011-02-04 11:05:11 <tcatm> I never received a warning
899 2011-02-04 11:05:19 <davout> i did not "break" it, I improved it
900 2011-02-04 11:05:29 <tcatm> how do I fix it?
901 2011-02-04 11:05:32 <davout> yea, I mentioned it on the forums
902 2011-02-04 11:05:40 <davout> true i didn't tell you explicitly :)
903 2011-02-04 11:05:43 <davout> hmm
904 2011-02-04 11:05:55 <davout> the paths have changed a little
905 2011-02-04 11:06:09 <davout> view-source:https://bitcoin-central.net/trades.xml
906 2011-02-04 11:06:16 <davout> or json for the matter
907 2011-02-04 11:06:27 <tcatm> same format?
908 2011-02-04 11:06:36 <davout> hmm
909 2011-02-04 11:06:38 <davout> not sure
910 2011-02-04 11:06:51 <davout> ticker is at the same place
911 2011-02-04 11:07:02 <davout> bc.net/ticker.(xml|json)
912 2011-02-04 11:07:15 <davout> lemme check the commit i'll tell you if the format changed
913 2011-02-04 11:07:25 <tcatm> looks like it fetched new trades
914 2011-02-04 11:09:09 <davout> looks like i broke the currency filter
915 2011-02-04 11:09:14 <davout> were you using it ?
916 2011-02-04 11:09:22 <tcatm> Not sure... :)
917 2011-02-04 11:09:32 <davout> just a get parameter
918 2011-02-04 11:09:34 <davout> brb food
919 2011-02-04 11:11:49 <tcatm> I'm filtering currencies for tradehistory again so that's not a problem
920 2011-02-04 11:21:59 <m0mchil> I guess poclbm should be fixed... what % loss is acceptable, is 0.25% ok (with output buffer of 32)?
921 2011-02-04 11:22:06 <m0mchil> or to go with 256?
922 2011-02-04 11:24:09 <sipa> what is the smallest worksize you use?
923 2011-02-04 11:25:02 <m0mchil> how is worksize related? or you mean threads per run?
924 2011-02-04 11:25:15 <tcatm> m0mchil: why don't you use one dword for every thread?
925 2011-02-04 11:25:50 <sipa> thread?
926 2011-02-04 11:26:01 <tcatm> gpu thread
927 2011-02-04 11:26:08 <tcatm> e.g. globalthreadsize
928 2011-02-04 11:26:19 <sipa> can't that be millions?
929 2011-02-04 11:26:34 <m0mchil> it is indeed, with -f 1
930 2011-02-04 11:26:48 <tcatm> is that a problem?
931 2011-02-04 11:26:53 <m0mchil> 285 mln on single 5970 core
932 2011-02-04 11:27:06 <sipa> = >1GiB of memory ...
933 2011-02-04 11:27:14 <tcatm> reduce globalthreadsize
934 2011-02-04 11:27:25 <sipa> yeah, by increasing -f :D
935 2011-02-04 11:27:36 <tcatm> or increase the loopsize from 1024 to 8192
936 2011-02-04 11:27:40 <m0mchil> sipa, back to your question - what is worksize?
937 2011-02-04 11:28:00 <sipa> m0mchil: i meant global thread size
938 2011-02-04 11:28:37 <m0mchil> ah, minimum in poclbm is 0x4000 (assuming minimum 'w' of 64 * 256)
939 2011-02-04 11:28:52 <tcatm> btw: on a 5870 at 350Mhash/s with -f 1 it takes 1.4 MByte
940 2011-02-04 11:29:11 <sipa> huh
941 2011-02-04 11:29:26 <sipa> that's way less than 4 bytes per global thread
942 2011-02-04 11:29:28 <tcatm> 350e6 hashes, / 1024 (kernel loop) * 4 byte
943 2011-02-04 11:29:35 <sipa> oh, you're using the loop
944 2011-02-04 11:29:36 <sipa> right
945 2011-02-04 11:30:22 <tcatm> so lots of headroom for faster GPUs
946 2011-02-04 11:30:28 <tcatm> or even smaller -f values
947 2011-02-04 11:32:31 <sipa> m0mchil: i'd say use one output per 'unit'
948 2011-02-04 11:32:40 <sipa> assuming that's a power of two
949 2011-02-04 11:32:42 <andrew12> of course, this channel is full of people at 7:32
950 2011-02-04 11:33:36 <sipa> m0mchil: at 340MH/s, -f 1, and 4096 outputs, you lose 0.0000024% :)
951 2011-02-04 11:33:48 <sipa> sorry, 16384
952 2011-02-04 11:33:54 <tcatm> why risk losing anything at all?
953 2011-02-04 11:34:00 <edcba> 13:06 <@tcatm> same format?
954 2011-02-04 11:34:12 <edcba> if you are using xml you'd better validate it...
955 2011-02-04 11:34:15 <m0mchil> output[64] 0.12%, output[128] 0.06 and so on
956 2011-02-04 11:34:38 <m0mchil> I think even 0.12 is quite acceptable