1 2010-12-27 00:00:03 <Diablo-D3> its either a 5550, a 5570, or a 5670
  2 2010-12-27 00:00:09 <devon_hillard> I want to get a block for once
  3 2010-12-27 00:00:11 <Diablo-D3> its not even good at gaming, really
  4 2010-12-27 00:00:12 <devon_hillard> 5570
  5 2010-12-27 00:00:33 <devon_hillard> 65K khashes/s
  6 2010-12-27 00:00:47 <Diablo-D3> my 4850 does 75 khash/sec
  7 2010-12-27 00:01:00 <Diablo-D3> and that has extremely poor mining performance =P
  8 2010-12-27 00:01:04 <Diablo-D3> but yeah, that line
  9 2010-12-27 00:01:08 <Diablo-D3> it just tells you what you have
 10 2010-12-27 00:01:11 <Diablo-D3> its purely informative
 11 2010-12-27 00:01:37 <devon_hillard> so the miner software is weak on my setup?
 12 2010-12-27 00:01:42 <devon_hillard> software-wise
 13 2010-12-27 00:01:56 <devon_hillard> because a 5570 should have better performance than a 4850
 14 2010-12-27 00:02:19 <Diablo-D3> it shouldnt
 15 2010-12-27 00:02:39 <Diablo-D3> 3D wise, a 4850 lines up with a 5750
 16 2010-12-27 00:02:51 <Diablo-D3> give or take, anyways
 17 2010-12-27 00:03:10 <Diablo-D3> and that does almost 120 mhash
 18 2010-12-27 00:03:27 <devon_hillard> ah, your card should be 30% more powerful overall
 19 2010-12-27 00:04:05 <Diablo-D3> here
 20 2010-12-27 00:04:09 <Diablo-D3> http://pastebin.com/AvymGnMJ
 21 2010-12-27 00:05:31 <devon_hillard> yeah, uses 30% of the power of a 4850
 22 2010-12-27 00:05:47 <devon_hillard> watt-wise
 23 2010-12-27 00:06:12 <Diablo-D3> yeah, and does a fair bit more mash
 24 2010-12-27 00:06:15 <devon_hillard> well, I can probably overclock it a bit
 25 2010-12-27 00:06:20 <Diablo-D3> 5xxx is a huge boost for mhash/watt over 4xxx
 26 2010-12-27 00:06:34 <Diablo-D3> you probably should just join the mining pool
 27 2010-12-27 00:06:42 <Diablo-D3> http://mining.bitcoin.cz/
 28 2010-12-27 00:06:56 <devon_hillard> a waterblock for this card is around $100 and I could up the frequency by another 50% I believe
 29 2010-12-27 00:07:27 <devon_hillard> Diablo-D3: what is a mining pool?
 30 2010-12-27 00:07:38 <devon_hillard> ah, reading now
 31 2010-12-27 00:07:51 <Diablo-D3> its not worth overclocking that card
 32 2010-12-27 00:08:04 <Diablo-D3> and you'd probably still require significant cooling for the vrms
 33 2010-12-27 00:08:12 <devon_hillard> vrms?
 34 2010-12-27 00:08:23 <Diablo-D3> voltage regulators
 35 2010-12-27 00:08:27 <devon_hillard> ah
 36 2010-12-27 00:08:41 <Diablo-D3> they tend to warm up faster than your gpu does when overclocking
 37 2010-12-27 00:08:49 <Diablo-D3> $100 could buy you a better card.
 38 2010-12-27 00:09:06 <devon_hillard> I'd have to get a better PSU while I'm at it
 39 2010-12-27 00:09:12 <devon_hillard> since still running off an old 400W
 40 2010-12-27 00:09:20 <Diablo-D3> heh, thats pretty low
 41 2010-12-27 00:09:31 <devon_hillard> so why I didn't bother getting a power-hungry card
 42 2010-12-27 00:12:34 <Diablo-D3> yeah, but you would have been better off updating your psu
 43 2010-12-27 00:16:08 <ArtForz> neato: http://www.techpowerup.com/137140/AMD-Radeon-HD-6950-can-be-unlocked-to-HD-6970.html
 44 2010-12-27 00:17:51 <devon_hillard> Diablo-D3: so what do I have to do to join a pool?
 45 2010-12-27 00:18:11 <Diablo-D3> follow the directions
 46 2010-12-27 00:18:18 <Diablo-D3> ArtForz: thats a VERY BAD IDEA
 47 2010-12-27 00:18:30 <Diablo-D3> ArtForz: same shit when all those fuckers unlocked _very broken_ 3 core phenoms into 4 core
 48 2010-12-27 00:18:44 <Diablo-D3> ArtForz: AMD _doesnt_ disable cores/pipes/shit just to fill orders.
 49 2010-12-27 00:19:02 <ArtForz> actually, they do
 50 2010-12-27 00:19:22 <Diablo-D3> not the way Intel does
 51 2010-12-27 00:19:36 <Diablo-D3> AMD would rather make a bunch of working quad cores just to drive the price down further
 52 2010-12-27 00:21:08 <Diablo-D3> driving the price down of a single product by selling more of it == more people buy AMD
 53 2010-12-27 00:21:34 <devon_hillard> Diablo-D3: ok, so with bitcoin.cz, I don't need to run bitcoind, just the miner script?
 54 2010-12-27 00:21:39 <ArtForz> well, lots of people have tested it, looks like about 70% got more-or-less working shaders
 55 2010-12-27 00:21:55 <Diablo-D3> devon_hillard: yes.
 56 2010-12-27 00:22:12 <Diablo-D3> ArtForz: unless they hit fab quality out of the park, that sounds wrong
 57 2010-12-27 00:22:39 <Diablo-D3> that also means they should have dropped the 6850 and just sold 6870s at 6850 prices
 58 2010-12-27 00:22:43 <Diablo-D3> just to destroy nvidia
 59 2010-12-27 00:22:59 <ArtForz> well, I guess they got higher yields than expected
 60 2010-12-27 00:23:05 <Diablo-D3> "I have an xfx 6950 on the way, hope it can unlock as well...."
 61 2010-12-27 00:23:18 <Diablo-D3> lol
 62 2010-12-27 00:23:26 <Diablo-D3> watch his xfx card blow up
 63 2010-12-27 00:23:29 <ArtForz> remember, they cut 6950 allocation and increased 6970 allocation 2 weeks before launch
 64 2010-12-27 00:24:07 <devon_hillard> Diablo-D3: what is the switch to select a server by address rather than bitcoind?
 65 2010-12-27 00:24:08 <ArtForz> = "whoops, too many full chips"
 66 2010-12-27 00:24:51 <devon_hillard> "D:Program FilesJavajdk1.6.0_23injava" -cp targetlibs*;targetDiabloMiner-0.0.1-SNAPSHOT.jar -Djava.library.path=targetlibs\nativeswindows com.diablominer.DiabloMiner.DiabloMiner -u USERNAME -p PASS -d
 67 2010-12-27 00:26:28 <Diablo-D3> ArtForz: hrrrrm
 68 2010-12-27 00:26:45 <ArtForz> not to mention 6950s could also be "hot" 6970s
 69 2010-12-27 00:26:57 <Diablo-D3> devon_hillard: -o and -r
 70 2010-12-27 00:27:05 <Diablo-D3> devon_hillard: so you want -o mining.bitcoin.cz -r 8332
 71 2010-12-27 00:27:24 <ArtForz> low Vt/high leakage chip = drop Vcore and clock a bunch
 72 2010-12-27 00:27:26 <Diablo-D3> devon_hillard: and your username is PoolUsername.MachineName
 73 2010-12-27 00:27:39 <Diablo-D3> ArtForz: yeah, thats what Im worried about
 74 2010-12-27 00:27:50 <ArtForz> well, those also tend to be very good overclockers as long as you can cool em
 75 2010-12-27 00:27:51 <Diablo-D3> ArtForz: AMD _does_ fail "working" chips because they overheat
 76 2010-12-27 00:28:11 <Diablo-D3> quite a few triple cores fell into that
 77 2010-12-27 00:28:15 <ArtForz> yep
 78 2010-12-27 00:28:19 <Diablo-D3> they far exceeded TDP but still "worked"
 79 2010-12-27 00:28:33 <devon_hillard> Diablo-D3: ok, it works, thanks
 80 2010-12-27 00:29:45 <ArtForz> power supply shouldnt be much of an issue, 5950 and 5970 have the same VRMs
 81 2010-12-27 00:30:41 <ArtForz> 6950 only has 2 6-pin power connectors, but that shouldnt be an issue, even a single 6-pin can handle >200W
 82 2010-12-27 00:31:11 <ArtForz> PCIe spec says 75W, actual connector spec says 7A/pin
 83 2010-12-27 00:31:23 <Diablo-D3> ArtForz: pci-e spec is too fucking safe, too
 84 2010-12-27 00:31:50 <Diablo-D3> every PSU manufacturer that isnt cheap chinese shit is overengineering their shit for safety reasons
 85 2010-12-27 00:32:06 <ArtForz> 6-pin has 3 power/gnd pairs, so connector can handle 21A @ 12V = 252W
 86 2010-12-27 00:32:24 <devon_hillard> I saw a really cheap 700W PSU, but it turns out it turns hot and blows up
 87 2010-12-27 00:32:35 <Diablo-D3> devon_hillard: Im _very_ picky with PSUs
 88 2010-12-27 00:32:38 <devon_hillard> "serioux", chinese stuff
 89 2010-12-27 00:32:40 <ArtForz> 8-pin uses the exact same connector pins and also only 3*12V, magically now capable of 150W
 90 2010-12-27 00:33:02 <Diablo-D3> ArtForz: yeah, its just fucking grounds to signal extra thick wires
 91 2010-12-27 00:33:07 <ArtForz> yep
 92 2010-12-27 00:33:12 <Diablo-D3> and Ive seen the connector prongs on pci-e
 93 2010-12-27 00:33:14 <Diablo-D3> they're HUEG
 94 2010-12-27 00:33:43 <ArtForz> well, I have the spec sheet for the pins, standard is 7A, high power is 13A
 95 2010-12-27 00:33:44 <Diablo-D3> Im surprised I dont have to solder copper wire coat hangers in there or some shit
 96 2010-12-27 00:36:09 <devon_hillard> Diablo-D3: do you use modular PSUs?
 97 2010-12-27 00:36:22 <sipa> Sapphire AMD 6970 2GB DDR5: 359.95 eur; can that be real?
 98 2010-12-27 00:36:34 <ArtForz> yes
 99 2010-12-27 00:36:41 <ArtForz> 6970 is barely faster than 5870 for mining
100 2010-12-27 00:36:56 <sipa> anything else is?
101 2010-12-27 00:37:19 <ArtForz> single GPU? nope
102 2010-12-27 00:37:41 <Diablo-D3> devon_hillard: I dont really care if a psu is modular or not as long as the connectors are good
103 2010-12-27 00:37:52 <sipa> ArtForz: how many Mhash/s does a 6970 do?
104 2010-12-27 00:38:06 <ArtForz> my calc says ~310
105 2010-12-27 00:38:27 <devon_hillard> [12/27/10 3:36:19 AM] DEBUG: Block found, but rejected by Bitcoin, on Redwood
106 2010-12-27 00:38:35 <sipa> and the 5970 are dual-gpu?
107 2010-12-27 00:38:40 <ArtForz> yup
108 2010-12-27 00:38:45 <Diablo-D3> devon_hillard: you sure you're using absolute newest version of my miner?
109 2010-12-27 00:39:01 <sipa> and those do over 500Mhash/s?
110 2010-12-27 00:39:04 <ArtForz> might be a few % higher because we should be able to get better VLIW usage out of VLIW4
111 2010-12-27 00:39:14 <slush> devon_hillard: Are you using latest diablo miner?
112 2010-12-27 00:39:15 <ArtForz> 5970 ~530Mh/s stock using diablominer
113 2010-12-27 00:39:34 <Diablo-D3> devon_hillard: btw, the only PSUs that dont seem to be absolute shit are Corsair's AX and HX serues
114 2010-12-27 00:39:37 <devon_hillard> Diablo-D3: Diablo-D3: I downloaded it 3-4 days ago
115 2010-12-27 00:39:42 <Diablo-D3> devon_hillard: its been updated since then
116 2010-12-27 00:39:52 <slush> devon_hillard: this is the reason
117 2010-12-27 00:39:53 <Diablo-D3> devon_hillard: I recently did a lot of fixes to deal with pool behavior
118 2010-12-27 00:40:04 <devon_hillard> ok, thanks for the tip
119 2010-12-27 00:40:26 <slush> Diablo-D3: I was scared it is here again, because I see this failure in log again :-D
120 2010-12-27 00:40:28 <ArtForz> ~556Mh/s stock with my custom miner
121 2010-12-27 00:41:25 <devon_hillard> Diablo-D3: ok, so you added the windows dlls, I had to hunt them down individually :)
122 2010-12-27 00:41:43 <Diablo-D3> devon_hillard: erm, my thing has ALWAYS had all the relevant windows dlls that arent apart of the driver
123 2010-12-27 00:41:53 <Diablo-D3> they ship with lwjgl, and I dont take lwjgl apart
124 2010-12-27 00:42:03 <ArtForz> >600Mh/s on 5970 should be possible with CAL + hand-optimized shader ASM
125 2010-12-27 00:42:15 <devon_hillard> Diablo-D3: ok, now running from the latest binary
126 2010-12-27 00:44:21 <ArtForz> >750Mh/s overvolted and OCd on water
127 2010-12-27 00:44:55 <devon_hillard> ArtForz: are you based in Europe?
128 2010-12-27 00:45:15 <devon_hillard> I have a water cooling kit for the CPU, considering expanding to cover a GPU as well
129 2010-12-27 00:45:29 <ArtForz> yep, .de
130 2010-12-27 00:45:41 <devon_hillard> where do you get your waterblocks?
131 2010-12-27 00:46:55 <ArtForz> usually whoever sells em cheapest
132 2010-12-27 00:50:31 <sipa> ArtForz: how many W does a 5970 require?
133 2010-12-27 00:51:14 <OneFixt> 300W
134 2010-12-27 00:51:40 <sipa> so you really need a decent PSU if you want more than one in the same system
135 2010-12-27 00:51:46 <OneFixt> yep
136 2010-12-27 00:52:58 <devon_hillard> looking at hashes/W, 5570 is quite efficient
137 2010-12-27 00:54:26 <devon_hillard> I think 2x5570 beats anything below 1000 EUR cards
138 2010-12-27 00:54:52 <devon_hillard> from a power efficiency standpoint
139 2010-12-27 00:54:55 <ArtForz> ?
140 2010-12-27 00:55:20 <devon_hillard> or below 500EUR, rather
141 2010-12-27 00:55:31 <ArtForz> 5850,5870,5970 have better Mh/W
142 2010-12-27 00:55:49 <devon_hillard> but are 5 times the price
143 2010-12-27 00:55:56 <OneFixt> i think 5770 is close to 5970 at Mh/W
144 2010-12-27 00:56:11 <OneFixt> but not if you count the cost and power consumption of the entire system
145 2010-12-27 00:56:46 <Diablo-D3> OneFixt: well
146 2010-12-27 00:56:49 <Diablo-D3> its sorta like
147 2010-12-27 00:56:53 <devon_hillard> true, although I only mine when I'm using the computer
148 2010-12-27 00:56:59 <devon_hillard> or not gaming
149 2010-12-27 00:57:01 <ArtForz> and 5 times the speed
150 2010-12-27 00:57:08 <ArtForz> 5570 = 60Mh/s @ 40W, 5870 = 313Mh/s @ 190W, 5970 = 535Mh/s @ 300W
151 2010-12-27 00:57:09 <Diablo-D3> performance per dollar+watt combined, 5970, 5770, 5870
152 2010-12-27 00:57:33 <Diablo-D3> and then the 69xx cards, and then the 68xx cards
153 2010-12-27 00:58:03 <devon_hillard> I need a spreadsheet to calculate amortization times
154 2010-12-27 00:58:51 <ArtForz> have fun guessing future difficulty and price
155 2010-12-27 00:58:53 <Diablo-D3> devon_hillard: thats a three way moving target
156 2010-12-27 00:59:22 <Diablo-D3> difficulty can go up unpredictably, value of btc can go up OR down as well
157 2010-12-27 00:59:35 <Diablo-D3> (and the pushes for btc going up and down are independant)
158 2010-12-27 00:59:50 <ArtForz> yep
159 2010-12-27 01:00:00 <sipa> 1.1Ghash/s = 425 eur/month (at current difficulty and 0.24 dollar/BTC)
160 2010-12-27 01:00:19 <devon_hillard> nice
161 2010-12-27 01:00:44 <devon_hillard> and you also have to consider electricity costs
162 2010-12-27 01:00:53 <sipa> yes, those not included
163 2010-12-27 01:01:09 <devon_hillard> which are around 13 $cents per kWh where I live
164 2010-12-27 01:04:19 <sipa> buying a 2-5970 system would take 3.5 month to repay itself now
165 2010-12-27 01:04:39 <sipa> if difficulty wouldn't go up, and power costs is neglected
166 2010-12-27 01:06:06 <devon_hillard> so 10 eurocents per kWh, considering 2x300W top cards (for 1k MHashes/s) + 150W for the rest of the sistem == 0.750kW x 0.1 EUR x 730.4 (hours in a month) =  54.78 EUR for running this setup for a month
167 2010-12-27 01:06:37 <devon_hillard> is that right?
168 2010-12-27 01:07:05 <sipa> looks right to me
169 2010-12-27 01:07:26 <ArtForz> looks about right
170 2010-12-27 01:08:50 <ArtForz> it's even a bit lower, box with 4*5970 on 2*1kW 80+ gold PSUs = 1280W
171 2010-12-27 01:09:21 <ArtForz> 1360W with the 5970s @ 820Mhz core
172 2010-12-27 01:09:27 <devon_hillard> there are motherboards with more than 2 full width PCI-express?
173 2010-12-27 01:09:38 <sipa> but 4 months to have it paid back... the network computation speed could have gone up *100
174 2010-12-27 01:09:54 <ArtForz> 1640W at 1.15Vcore and 900MHz
175 2010-12-27 01:10:16 <sipa> when did you buy those, ArtForz?
176 2010-12-27 01:10:33 <ArtForz> what? the 5970s?
177 2010-12-27 01:10:39 <sipa> yes
178 2010-12-27 01:11:13 <ArtForz> first 12 > 3 months ago, another 12 when they were on sale @ 380EUR
179 2010-12-27 01:12:25 <sipa> i think it's a bit late to join in on the making-real-profit-through-mining thing
180 2010-12-27 01:12:30 <ArtForz> and 1760W at 1.1625Vcore and 930MHz
181 2010-12-27 01:12:31 <ArtForz> yep
182 2010-12-27 01:12:55 <ArtForz> I dont think I'll be adding more GPUs
183 2010-12-27 01:14:52 <ArtForz> 1.1625V on air is pretty much the limit
184 2010-12-27 01:15:56 <ArtForz> even with card fans @ 100% and 25W 120mm fans VRM temps are >110???C
185 2010-12-27 01:16:27 <slush> Diablo-D3: hey, not good news for you
186 2010-12-27 01:16:43 <devon_hillard> Diablo-D3: so you're the diablo miner author? paste your receiving address so I can send you my first BTC :)
187 2010-12-27 01:16:56 <slush> I just tested my relative speed against other miners in pool with old diablo miner and new one
188 2010-12-27 01:17:11 <slush> Diablo-D3: new one is much slower in real shares count
189 2010-12-27 01:17:24 <Diablo-D3> slush: that hasnt changed.
190 2010-12-27 01:17:39 <Diablo-D3> devon_hillard: its in my forum sig =P
191 2010-12-27 01:17:54 <slush> Say I have ratio 1.4 against m0mchil with new one an 1.54 (and still rising) with old one
192 2010-12-27 01:18:14 <slush> Diablo-D3: I know _teoretically_ it hasnt changed
193 2010-12-27 01:18:16 <Diablo-D3> slush: I get a H==0 50 times in 45 minutes, this hasnt changed.
194 2010-12-27 01:18:21 <Diablo-D3> and yes, I actually measured it
195 2010-12-27 01:18:32 <sipa> is there that much difference between m0mchil's and Diablo-D3's miners?
196 2010-12-27 01:18:44 <Diablo-D3> slush: and if you arent doing it at least 50 times, your statistics will be wildly off
197 2010-12-27 01:18:44 <slush> Diablo-D3: But I was surprised that my reward from pool per block was lower for last two days. So I did this little investigation
198 2010-12-27 01:19:07 <Diablo-D3> sipa: mines more efficient
199 2010-12-27 01:19:36 <slush> Diablo-D3: i'm talking about another possibility - two executors are not working on different tasks. Say, it is even possible?
200 2010-12-27 01:19:44 <sipa> Diablo-D3: yes, i would expect that
201 2010-12-27 01:19:50 <sipa> but is it a few % or more
202 2010-12-27 01:20:02 <Diablo-D3> slush: they're not directly working on different tasks anymore
203 2010-12-27 01:20:05 <sipa> because i tuned m0mchil's, and got a few % extra by doing so as well
204 2010-12-27 01:20:22 <Diablo-D3> slush: just subchunks of the same task
205 2010-12-27 01:20:35 <Diablo-D3> sipa: try it.
206 2010-12-27 01:21:00 <slush> Diablo-D3: yes, I'm talking about this change. Is  there _any_ possibility that executors are working even on the same nonces?
207 2010-12-27 01:21:06 <sipa> yeah, i should :)
208 2010-12-27 01:21:18 <Diablo-D3> slush: no, not after the christmas day update
209 2010-12-27 01:21:53 <slush> Diablo-D3: I'm talking about it because mhash/s indicator still show correct value for me, but _real_ share ratio is lower than before
210 2010-12-27 01:22:36 <slush> I cannot attack you in any case, I'm just investigating what is behind it
211 2010-12-27 01:23:25 <Diablo-D3> slush: you're using the absolutely newest one?
212 2010-12-27 01:23:49 <slush> Diablo-D3: I updated today
213 2010-12-27 01:23:59 <Diablo-D3> slush: btw, the shares will come out much more chunky than usual
214 2010-12-27 01:24:01 <Diablo-D3> very random
215 2010-12-27 01:24:19 <Diablo-D3> with the old 1 getwork per executor, it was smoother
216 2010-12-27 01:24:23 <slush> From my (not exact) stat I'm talking about ~45 mhash lost between those versions
217 2010-12-27 01:24:41 <slush> measured on 1990 shares
218 2010-12-27 01:25:45 <da2ce7> @ stock voltage, what clock speed do you find the 5970 can reach and still be stable? Mine only get up to 800mhz.
219 2010-12-27 01:26:19 <ArtForz> varies
220 2010-12-27 01:26:31 <ArtForz> my worst card only gets 780/810
221 2010-12-27 01:26:33 <slush> Diablo-D3: I will wait for bigger data sample, but it looks weird
222 2010-12-27 01:26:45 <ArtForz> my best card does 880/900
223 2010-12-27 01:27:07 <Diablo-D3> slush: btw, if this ever ran the same nonce block twice at the same time, you'd see dupes.
224 2010-12-27 01:27:27 <slush> Diablo-D3: Even after your anti-dupes fix?
225 2010-12-27 01:27:39 <Diablo-D3> slush: I have no anti-dupes fix that fixes dupes
226 2010-12-27 01:27:52 <Diablo-D3> it only fixes atomic getwork state cloning.
227 2010-12-27 01:28:18 <slush> well, I will make one-day sample with old version and new version.
228 2010-12-27 01:28:31 <slush> On those two samples, share counts should be the same
229 2010-12-27 01:28:43 <Diablo-D3> and the mhash meter hasnt changed?
230 2010-12-27 01:28:52 <slush> no, meter is OK
231 2010-12-27 01:29:23 <Keefe> my 5970's OC to: 840/885, 870/870, 870/860
232 2010-12-27 01:30:37 <ArtForz> yeah, 5970 cypresses are mainly binned for low wattage, not high clocks
233 2010-12-27 01:31:01 <LobsterMan> mining.bitcoin.cz
234 2010-12-27 01:31:04 <LobsterMan> i can't get that to load
235 2010-12-27 01:31:21 <LobsterMan> is your site down slush?
236 2010-12-27 01:31:45 <slush> LobsterMan: no, it is in good condition now
237 2010-12-27 01:31:59 <LobsterMan> hmm
238 2010-12-27 01:32:11 <slush> LobsterMan: probably network issue
239 2010-12-27 01:32:25 <LobsterMan> http://mining.bitcoin.cz/ does not work for me
240 2010-12-27 01:32:28 <slush> LobsterMan: see counting shares in log every second...
241 2010-12-27 01:32:34 <LobsterMan> im getting some sort of dns error...
242 2010-12-27 01:33:11 <slush> LobsterMan: Thats probably government starting to block all bitcoin related sites :))
243 2010-12-27 01:33:19 <LobsterMan> i use opendns though
244 2010-12-27 01:33:22 <LobsterMan> maybe they just fail
245 2010-12-27 01:33:23 <slush> me too
246 2010-12-27 01:33:28 <sipa> Diablo-D3: i get 78900 with your miner, 79700 with m0mchil's after my own tweaking
247 2010-12-27 01:33:41 <LobsterMan> i can get to the site through www.hidemyass.com
248 2010-12-27 01:33:45 <Diablo-D3> sipa: remember, mine does -w as well
249 2010-12-27 01:33:52 <LobsterMan> but i can't get to it from my normal connection
250 2010-12-27 01:33:53 <sipa> yes, both worksize 64
251 2010-12-27 01:34:06 <Diablo-D3> sipa: you have looping in your tweak?
252 2010-12-27 01:34:16 <da2ce7> slush, I that your 5BTC to the winner idea is great!
253 2010-12-27 01:34:16 <slush> what worksize do you use on 5970? 64?
254 2010-12-27 01:34:21 <da2ce7> *thin
255 2010-12-27 01:34:29 <Cusipzzz> site works for me
256 2010-12-27 01:34:30 <Diablo-D3> 64 seems to be the most efficient on all 4xxx and 5xxx
257 2010-12-27 01:34:36 <LobsterMan> http://guide.opendns.com/main?url=mining.bitcoin.cz&servfail=
258 2010-12-27 01:34:39 <sipa> Diablo-D3: i did a loop that runs 5x inside the opencl part
259 2010-12-27 01:34:44 <da2ce7> *think... boy I need sleep.
260 2010-12-27 01:34:47 <slush> da2ce7: Unfortunately it turns pooled mining to another lottery
261 2010-12-27 01:35:08 <Diablo-D3> sipa: yes, mine doesnt do that yet, although it seems to be a performance win
262 2010-12-27 01:35:41 <sipa> well, i sent my code to m0mchil, and none of my modifications helped for his own setup
263 2010-12-27 01:35:49 <da2ce7> yeah, but the entire bitcoin generation is a lottery, :P that is it's fundamentals... there is only so much that you can abstract that part away.
264 2010-12-27 01:36:13 <da2ce7> might as well have fun while doing it!
265 2010-12-27 01:37:39 <slush> da2ce7: well, after thinking about it and talking with others, I want  to make pooled mining 'as fair as possible'; everybody can enjoy lottery with standalone mining :))
266 2010-12-27 01:38:45 <slush> da2ce7: Although some miners in pool have lower reward than their teoretical reward for mining standalone, it is only kind of luck. In middle term, those differences will be smaller
267 2010-12-27 01:38:54 <sipa> Diablo-D3: if i disable looping, yours is faster
268 2010-12-27 01:39:42 <slush> what? looping? faster? What are you talking about, guys? :)
269 2010-12-27 01:40:07 <LobsterMan> slush now i can get to your site no problem
270 2010-12-27 01:40:12 <LobsterMan> odd opendns hiccup.....
271 2010-12-27 01:40:21 <slush> LobsterMan: great
272 2010-12-27 01:40:40 <sipa> slush: it's a tweak in GPU miners, by letting each thread not try one nonce but multiple
273 2010-12-27 01:40:42 <slush> LobsterMan:  but bad for you, of course :)
274 2010-12-27 01:40:54 <sipa> no idea why it helps, but it effectively gains a few % speed
275 2010-12-27 01:41:21 <slush> sipa: thanks
276 2010-12-27 01:44:26 <slush> LobsterMan: if you have those problems often, try using IP instead of domain in your miner
277 2010-12-27 01:45:04 <LobsterMan> that's the first time it's ever given me a problem
278 2010-12-27 01:54:27 <Diablo-D3> sipa: hee
279 2010-12-27 01:54:42 <Diablo-D3> sipa: yeah, looping is the only tweak I dont have
280 2010-12-27 01:54:56 <Diablo-D3> looping is only effective on 5xxx, btw
281 2010-12-27 01:55:01 <Diablo-D3> 4xxx and nvidia doesnt gain from it
282 2010-12-27 01:55:48 <sipa> i have a 4870
283 2010-12-27 01:56:51 <Diablo-D3> huh, and you're gaining from it?
284 2010-12-27 01:57:04 <sipa> yes
285 2010-12-27 01:57:57 <Diablo-D3> I will have to revisit this.
286 2010-12-27 02:01:12 <sipa> Diablo-D3: it's very delicate
287 2010-12-27 02:01:32 <sipa> i have gains for loop sizes 4-5-6-7, with a optimal of 5
288 2010-12-27 02:01:46 <sipa> lower than 4 or higher than 7 it slows down
289 2010-12-27 02:02:00 <Diablo-D3> probably because the ALU instruction cache is filling up after 7
290 2010-12-27 02:03:52 <sipa> but m0mchil had a 5% slowdown if he enabled my looping on his system
291 2010-12-27 02:47:05 <afed> yes hello
292 2010-12-27 02:47:34 <afed> any mining tips
293 2010-12-27 02:47:43 <afed> currently using 2x radeon 5770 and a 5550
294 2010-12-27 02:53:59 <eureka^> ;b 30
295 2010-12-27 02:54:02 <eureka^> oops
296 2010-12-27 04:40:43 <EvanR> block 100000 coming up
297 2010-12-27 04:41:26 <noagendamarket> *waits for the y2k bug
298 2010-12-27 04:41:53 <EvanR> oh, of course the end is on 100012
299 2010-12-27 04:42:27 <nanotube> ;;bc,stats
300 2010-12-27 04:42:29 <gribble> Current Blocks: 99610 | Current Difficulty: 14484.16236123 | Next Difficulty At Block: 100799 | Next Difficulty In: 1189 blocks | Next Difficulty In About: 1 week, 0 days, 19 hours, 55 minutes, and 41 seconds | Next Difficulty Estimate: 15279.29105313
301 2010-12-27 04:44:42 <jgarzik> cpuminer version 0.3.3 is out there: http://www.bitcoin.org/smf/index.php?topic=1925.0;all
302 2010-12-27 04:44:43 <bitbot> New demonstration CPU miner available
303 2010-12-27 04:46:45 <nanotube> jgarzik: mm looks like if i run 0.3.1 with the c algo, no upgrade necessary?
304 2010-12-27 04:47:01 <jgarzik> nanotube: correct
305 2010-12-27 04:47:21 <nanotube> btw, cryptopp_asm doesn't even show up in my available algos... sup with that?
306 2010-12-27 04:47:49 <nanotube> cryptopp yes, but no cryptopp_asm
307 2010-12-27 04:47:57 <nanotube> is that 32bit only maybe?
308 2010-12-27 04:47:58 <jgarzik> nanotube: only appears for 32-bit builds.  On a 64-bit box, you can add "-m32" to CFLAGS, and build a 32-bit binary.
309 2010-12-27 04:48:15 <nanotube> ah cool. would that be any faster that the c alg?
310 2010-12-27 04:48:23 <nanotube> on a 64bit box
311 2010-12-27 04:48:43 <jgarzik> nanotube: anything's possible, in the weird wild world of CPUs
312 2010-12-27 04:48:58 <nanotube> hehe ic
313 2010-12-27 04:50:10 <nanotube> nice idea :)
314 2010-12-27 04:52:10 <theymos> HTTPS access to http://blockexplorer.com/q and the rest of BBE will be offline for 2-5 days starting a few days from now. If anyone here uses the tools over HTTPS, change it ASAP. (I'm making some changes that require me to move BBE to a different computer temporarily, and I don't want to move the private key for security reasons.)
315 2010-12-27 06:28:55 <devon_hillard> So can people exploit browser-side javascript to perform small but measurable computational tasks for servers?
316 2010-12-27 06:29:31 <devon_hillard> like, if you have a high trafficked site, you may have a JS routine cracking a couple of hashes
317 2010-12-27 06:31:07 <Keefe> how much processing time do you think you could get from each page hit? and how many hits per day?
318 2010-12-27 06:31:19 <Keefe> maybe 1 second?
319 2010-12-27 06:31:46 <Keefe> without being obvious to some people
320 2010-12-27 06:33:02 <Keefe> how much do traditional web ads pay per view, on some site you own?
321 2010-12-27 06:34:04 <bd_> Keefe: you can do quite a lot of work with a popular site, I'd expect
322 2010-12-27 06:34:10 <bd_> just try not to use more than 10% of the CPU or so
323 2010-12-27 06:39:23 <devon_hillard> or you could have a transient sort of botnet, sending spam from within porn surfer's JS
324 2010-12-27 06:46:51 <bd_> devon_hillard: No, that's prevented by the JS security model
325 2010-12-27 06:57:47 <h4ck3rk1ng> hey guys
326 2010-12-27 06:58:24 <h4ck3rk1ng> hey anyone here?
327 2010-12-27 06:58:52 <h4ck3rk1ng> i want to know how to start my own bitcoin exchange site?
328 2010-12-27 06:59:39 <MT`AwAy> h4ck3rk1ng: code it?
329 2010-12-27 06:59:47 <h4ck3rk1ng> i know
330 2010-12-27 06:59:50 <h4ck3rk1ng> but in what?
331 2010-12-27 06:59:55 <h4ck3rk1ng> and what API do i use?
332 2010-12-27 07:00:09 <h4ck3rk1ng> there is no documentation...
333 2010-12-27 07:16:04 <lfm> whats the range of reasonable values for -worksize on the m0mochil gpu miner?
334 2010-12-27 07:18:22 <jgarzik> h4ck3rk1ng: there are two "APIs":  the JSON-RPC API detailed via the 'help' command, and the P2P network protocol itself.  most likely you want the JSON-RPC API, an HTTP server on port 8332.
335 2010-12-27 07:21:41 <h4ck3rk1ng> ok, so i could code a exchange script on the JSON-RPC api
336 2010-12-27 07:21:50 <h4ck3rk1ng> is there like a demo script i could expand on?
337 2010-12-27 07:28:42 <h4ck3rk1ng> ...
338 2010-12-27 07:35:00 <devon_hillard> ok, bitcoind is smart enough not to send duplicate search spaces to different clients, right?
339 2010-12-27 07:42:53 <lfm> davon_hillard yes it is that smart
340 2010-12-27 07:47:41 <grondilu> Is it ok if two users run bitcoin on the same time on the same machine ?
341 2010-12-27 08:06:22 <grondilu> Anyone here ?
342 2010-12-27 08:08:48 <h4ck3rk1ng> i am
343 2010-12-27 08:08:55 <h4ck3rk1ng> but masturbating to SAW
344 2010-12-27 08:09:24 <AAA_awright> grondilu: Don't ask meta-questions "Can I ask a question" "Is anyone here"
345 2010-12-27 08:09:38 <h4ck3rk1ng> lmao
346 2010-12-27 08:09:43 <h4ck3rk1ng> i was kidding
347 2010-12-27 08:09:51 <h4ck3rk1ng> how do i interface WHMCS to bitcoins?
348 2010-12-27 08:09:58 <h4ck3rk1ng> i want to be able to sell hosting with bitcoins
349 2010-12-27 08:10:11 <h4ck3rk1ng> do i make bitcoins access the WHMCS API?
350 2010-12-27 08:10:19 <h4ck3rk1ng> or do i make whmcs access bitcoin api?
351 2010-12-27 08:22:09 <grondilu> Is it ok if two users run bitcoin on the same time on the same machine ?
352 2010-12-27 08:26:05 <devon_hillard> how robust is bitcoin if the central IRC server is attacked, wikileaks-style?
353 2010-12-27 08:26:32 <devon_hillard> either by a DDoS or legal takedown of DNS
354 2010-12-27 08:28:48 <AAA_awright> You can still connect to a peer directly
355 2010-12-27 08:29:41 <devon_hillard> it would be nice to have the client keep a running list of active peers, then attempt to connect to some of those peers at login (if the IRC is down)
356 2010-12-27 08:29:58 <AAA_awright> grondilu: Should be, what conflicts could there be?
357 2010-12-27 08:30:03 <devon_hillard> most peers would be ephemeral, but there are always some peers on the network
358 2010-12-27 08:30:26 <AAA_awright> You can't have two programs listen on the same port, and I don't think you can have two programs writing to the same database if they are even using the same database
359 2010-12-27 08:59:14 <grondilu> AAA_awright: indeed that's what I thought.  Clients can't listen to the same port, can they ?
360 2010-12-27 09:01:23 <devon_hillard> if a fraudster wanted to forge bitcoins, given the network speed of X Mhashes/s, what sort of speed should a bad guy attain? I understand it's an exponential function of X
361 2010-12-27 09:01:39 <devon_hillard> with the exponent larger than 1
362 2010-12-27 09:04:09 <grondilu> forging has nothing to do with hashes/s.  Forging would require breaking SHA256
363 2010-12-27 09:04:37 <grondilu> hum... sorry.  I meant stealing.
364 2010-12-27 09:04:48 <grondilu> I don't know what you mean by "forging".
365 2010-12-27 09:05:24 <grondilu> devon_hillard: what do you mean by "forging" ?
366 2010-12-27 09:08:18 <devon_hillard> grondilu: being able to present a longer proof-of-work hash chain to other bitcoin peers
367 2010-12-27 09:08:31 <devon_hillard> a forged one
368 2010-12-27 09:10:55 <arcatan> if i understand correctly, that'd require breaking SHA256, too
369 2010-12-27 09:11:03 <devon_hillard> or being able to create a hash collision with the current chain, I think
370 2010-12-27 09:11:34 <devon_hillard> i.e. to present an alternative work tree to peers, if I understand it correctly
371 2010-12-27 09:12:02 <devon_hillard> or work history
372 2010-12-27 09:18:41 <Keefe> if you have more computing power than everyone else combined, you can build your own chain in parallel starting with the main chain's current block, keep your chain private while it grows faster than the main one, spend bitcoins (generated before your fork) on the main chain, obtain goods/services/money in exchange, then merge your private (now-longer) chain with the main network and wipe out every transaction after the fork including your spend
373 2010-12-27 09:18:51 <Keefe> no hacking/cracking needed
374 2010-12-27 09:18:59 <Keefe> just more computing power than everyone else combined
375 2010-12-27 09:19:36 <Keefe> then spend the bitcoins a second time
376 2010-12-27 09:20:56 <Keefe> the longer you build your own chain and keep it to yourself, the more havoc you can cause when you merge
377 2010-12-27 09:21:48 <devon_hillard> Keefe: but even then, the upside would be that a new villain would have to one-up both the network and this old villain
378 2010-12-27 09:22:21 <Keefe> unless you repeat the process
379 2010-12-27 09:22:25 <Keefe> fork again
380 2010-12-27 09:22:51 <devon_hillard> a big, powerful node would be visible in the network, right?
381 2010-12-27 09:22:53 <Keefe> then a new villain has to only one-up you (assuming you're still more than the main chain)
382 2010-12-27 09:22:55 <devon_hillard> at least after the merger
383 2010-12-27 09:24:19 <Keefe> if you merge your chain one block at a time from multiple ips thru tor, it could just look like a major internet netsplit being resolved
384 2010-12-27 09:24:32 <LobsterMan> http://www.techpowerup.com/137140/AMD-Radeon-HD-6950-can-be-unlocked-to-HD-6970.html
385 2010-12-27 09:25:36 <Keefe> it will still raise suspicion and then it will be up to the community to decide whether to intervene and declare the fork to be invalid
386 2010-12-27 09:26:03 <Keefe> you could probably get away with a short fork
387 2010-12-27 09:26:30 <Keefe> like say 10 blocks
388 2010-12-27 09:27:00 <Keefe> dunno really. we all hope to not have to deal with such disruption, but we also must be aware it can and probably will eventually happen
389 2010-12-27 09:27:34 <Keefe> especially if a government decides to try to disrupt bitcoin
390 2010-12-27 09:29:59 <Keefe> LobsterMan: i saw someone discussing it in one of the chans
391 2010-12-27 09:46:10 <devon_hillard> Keefe: there is also the possibility that some bitcoins get lost
392 2010-12-27 09:46:22 <devon_hillard> if their owner loses his private credentials
393 2010-12-27 09:46:44 <devon_hillard> they would be money that never return to circulation
394 2010-12-27 09:50:40 <Keefe> that would be not related in any way to someone causing trouble by forking the block chain with more processing power than everyone else combined
395 2010-12-27 09:54:22 <zygf> hmm, curious fact, the world's GDP is ~$58e12, that's 58e14 US cents, bitcoins are similarly divisible, there's a cap of 21e14 bitcoin fractions
396 2010-12-27 09:57:36 <larsivi> zygf: it says somewhere that bitcoins are infinitely divisable, it is just restricted to 8 decimals in the current implementation
397 2010-12-27 10:10:51 <genjix> zygf: awesome! heh
398 2010-12-27 10:23:44 <bonsaikitten> the power of inflation!
399 2010-12-27 10:23:53 <bonsaikitten> soon everyone will be a billionaire ...
400 2010-12-27 10:23:59 <genjix> sweet
401 2010-12-27 10:24:23 <genjix> how will you spend your billion?
402 2010-12-27 10:24:50 <bonsaikitten> I think I'll buy some bread
403 2010-12-27 10:25:00 <bonsaikitten> and the remaining millions will be good toilet paper
404 2010-12-27 10:30:00 <larsivi> bonsaikitten: in terms of bitcoin, it would be more correct to say "power of deflation" :)
405 2010-12-27 10:31:19 <bonsaikitten> larsivi: stop trying to break my mind!
406 2010-12-27 10:31:31 <bonsaikitten> it came pre-broken with bad factory defaults, so it won't work
407 2010-12-27 12:34:30 <lfm> can anyone give any clue what is the range of reasonable values for -worksize on the m0mochil gpu miner?
408 2010-12-27 12:36:14 <sipa> i use 64
409 2010-12-27 12:37:03 <lfm> did you try others? what vid card is it?
410 2010-12-27 12:38:21 <sipa> ati 4870
411 2010-12-27 12:38:29 <sipa> yes, i tried others
412 2010-12-27 16:28:15 <spm_Draget> Anyone of you holding a lightningtalk about bitcoin?
413 2010-12-27 16:29:01 <Cusipzzz> is that like an elevator speech?
414 2010-12-27 16:29:06 <ThomasV> hehe
415 2010-12-27 16:30:46 <ThomasV> spm_Draget, better just ask your question if you have one
416 2010-12-27 16:31:10 <spm_Draget> I am talking about 27c3, ThomasV =)
417 2010-12-27 16:33:10 <ThomasV> oh that must be like a spotlight then
418 2010-12-27 16:33:34 <spm_Draget> ThomasV: It is a talk any participant of the congress can hold for 4 minutes
419 2010-12-27 16:33:41 <ThomasV> yeah
420 2010-12-27 16:33:59 <spm_Draget> Introducing your project to a few thousand viewers at place and a for tenthousands or hundretthounds workdwilde on streams =)
421 2010-12-27 16:34:11 <ThomasV> oh it's right now, I see
422 2010-12-27 16:34:14 <spm_Draget> Sorry, for the missunderstanding :P
423 2010-12-27 16:35:34 <ThomasV> they surely will try to break it at ccc
424 2010-12-27 16:36:11 <ThomasV> it'll be like a bank stress test :-D
425 2010-12-27 16:43:36 <ThomasV> spm_Draget : are you attending it, btw ?
426 2010-12-27 16:43:54 <spm_Draget> Nope, sitting at home, watching the streams
427 2010-12-27 18:17:08 <gribble> #233 Mon Dec 27 08:21:09 2010 grondilu@unaffiliated/grondilu BUY 400.0 BTC @ 100.0 USD (One VISA USD gift card)
428 2010-12-27 18:17:08 <grondilu> ;;book USD
429 2010-12-27 18:18:33 <gribble> The bot responds when you start a line with the ! character. A good starting point for exploring the bot is the !facts command. You can also visit the bot's website for a list of help topics and documentation: http://gribble.sourceforge.net/
430 2010-12-27 18:18:33 <grondilu> ;;help
431 2010-12-27 18:20:59 <wumpus> http://hardware.slashdot.org/story/10/12/27/0820203/AMD-Radeon-HD-6950-Can-Be-Unlocked-To-HD-6970
432 2010-12-27 18:21:01 <wumpus> interesting
433 2010-12-27 18:21:32 <ArtForz> posted that yesterday
434 2010-12-27 18:23:36 <wumpus> ok
435 2010-12-27 18:23:55 <ArtForz> and it's still not a full 6970
436 2010-12-27 18:24:32 <ArtForz> 6970 has 6GHz ram chips, 6950 has 5GHz, good luck getting those to run @ 5.5 (stock 6970 mem clk)
437 2010-12-27 18:24:41 <wumpus> so it makes little sense to buy a 6950 for that reason
438 2010-12-27 18:25:00 <ArtForz> well, assuming all shaders work it should be == 6970 for mining
439 2010-12-27 18:25:11 <wumpus> true, ram isn't important for that
440 2010-12-27 18:25:49 <ArtForz> but 69xx is kinda ram BW limited already, so those 128 extra shaders wont really help much for 3D
441 2010-12-27 19:07:19 <jgarzik> http://www.bitcoin.org/smf/index.php?topic=2362.msg33465#msg33465
442 2010-12-27 19:07:21 <bitbot> An estimate of fpga performance : mike_la_jolla: mike_la_jolla checking in here to clarify some FPGA questions.  - DNDPB_S327:  http://www.dinigroup.com/new/DNDPB_S327.html List price is $19,680 for quantity 1.  - This is probably a much better choice:  DNBFC_S12_PCIe: http://www.dinigroup.com/new/DNBFC_S12_PCIe.html List price for quantity 1 is $8,950.  We sell thousands...
443 2010-12-27 19:45:52 <Keefe> from the wikipedia article on bitcoin: "the value of bitcoins will begin to deflate due to the lack of new introduction"
444 2010-12-27 19:47:19 <Keefe> is that the correct way to say it? won't there tend to be an upward force on the value of bitcoins, as less is created?
445 2010-12-27 19:47:27 <Cusipzzz> that article is shite
446 2010-12-27 19:47:39 <Keefe> actually i think it's pretty nice
447 2010-12-27 19:47:47 <Cusipzzz> still call it scrip?
448 2010-12-27 19:47:59 <Cusipzzz> haven't check it recently
449 2010-12-27 19:48:16 <Keefe> no mention of scrip
450 2010-12-27 19:48:28 <Cusipzzz> it's not freekin scrip
451 2010-12-27 19:48:34 <Cusipzzz> oh really? was in first sentence
452 2010-12-27 19:51:35 <nanotube> Keefe: deflation == value of currency goes up. inflation == value of currency goes down.
453 2010-12-27 19:51:48 <nanotube> it's a little counterintuitive... but it is what it is. :)
454 2010-12-27 19:52:07 <Cusipzzz> yes, deflation =prices comes down, which means currency is stronger
455 2010-12-27 19:52:18 <Keefe> but "value of bitcoins will deflate"?
456 2010-12-27 19:53:14 <nanotube> Keefe: yea, that's probably bad phraseology
457 2010-12-27 19:54:42 <Keefe> i suppose i should edit the article if i can think of a better phrase, but i've never edited a wiki before
458 2010-12-27 19:55:27 <Cusipzzz> some editor nazi will just change it back...
459 2010-12-27 19:55:38 <Cusipzzz> unless you have 4/5 people do the same edit or et an editor to do it.
460 2010-12-27 21:16:18 <nanotube> Keefe: there's always a first time. :)
461 2010-12-27 21:28:21 <wumpus> if you're just reformulating something so it's more clear I'm pretty sure no edit nazi will change it back, only if you significantly change the meaning and 'they' don't agree with it
462 2010-12-27 21:28:49 <Diablo-D3> hrm, wheres slush
463 2010-12-27 21:28:57 <slush> Diablo-D3: hi
464 2010-12-27 21:29:01 <Diablo-D3> oh hau
465 2010-12-27 21:29:09 <Diablo-D3> slush: I think Ive noticed a tiny little flaw in my plan
466 2010-12-27 21:29:22 <slush> Diablo-D3: in miner performance?
467 2010-12-27 21:29:42 <Diablo-D3> slush: due to how locking functions in such a setup, its basically fucking over multiple kernel runs
468 2010-12-27 21:30:24 <slush> Diablo-D3: from my stats, it looks newest miner has -20% performance from previous one
469 2010-12-27 21:30:27 <Diablo-D3> Im going to have to switch to thread+getwork per gpu, and do single threaded executor rotating
470 2010-12-27 21:30:30 <slush> Diablo-D3: when counting shares
471 2010-12-27 21:30:31 <Diablo-D3> slush: thats about right
472 2010-12-27 21:30:37 <Diablo-D3> its in the right ballpark anyways
473 2010-12-27 21:30:46 <Diablo-D3> the performance is the same if I just ran one executor
474 2010-12-27 21:32:04 <slush> Diablo-D3: don't fully understand. Are you going to revert last 'optimizations'?
475 2010-12-27 21:35:08 <slush> I mean changes related to getwork improvements
476 2010-12-27 21:36:48 <Motoma> Hi there. I've been spellunking through the wiki, but I can't find an answer: Is there a way to launch the standard bitcoin client to connect to an existing bitcoin server via RPC?
477 2010-12-27 21:37:17 <nanotube> Motoma: nope
478 2010-12-27 21:37:34 <nanotube> you can launch bitcoind to connect to a bitcoin running with -server... but not the other way around.
479 2010-12-27 21:38:16 <Motoma> nanotube: Oh, and how would I do that?
480 2010-12-27 21:38:55 <nanotube> well, you run "bitcoin -server" to start the gui with the server... then you can run 'bitcoind getinfo' (e.g.) to interact with it.
481 2010-12-27 21:40:00 <Motoma> nanotube: I guess more clarification is necessary: I'm trying to run a number of my home computers against a single bitcoin -server instance.
482 2010-12-27 21:40:47 <Diablo-D3> slush: not revert, keep going
483 2010-12-27 21:40:49 <Motoma> I can't seem to find a way to instruct bitcoind to connect to a remote server.
484 2010-12-27 21:40:55 <nanotube> Motoma: for what purpose?
485 2010-12-27 21:41:07 <Diablo-D3> slush: if I get rid of the locking altogether and move executors all into the same thread (and just run it as a queue)
486 2010-12-27 21:41:11 <Diablo-D3> slush: it should be fine
487 2010-12-27 21:41:23 <Diablo-D3> slush: but Im stuck using a getwork per gpu
488 2010-12-27 21:41:30 <slush> Diablo-D3: So it should solve my performance issue, right?
489 2010-12-27 21:41:34 <Diablo-D3> yes
490 2010-12-27 21:41:40 <slush> Diablo-D3: great
491 2010-12-27 21:41:43 <nanotube> Motoma: it seems like you're trying to do something that shouldn't be done. ;)
492 2010-12-27 21:41:49 <Diablo-D3> slush: it'll solve EVERYONE's
493 2010-12-27 21:42:09 <Motoma> nanotube: Oh, I see. Perhaps I'll have to set up a private pooled mining system then, eh?
494 2010-12-27 21:42:44 <nanotube> Motoma: yes, the 'remote miners' are a separate breed, not the same as the stock bitcoind.
495 2010-12-27 21:42:48 <Cusipzzz> Motoma: that would work
496 2010-12-27 21:42:59 <slush> Diablo-D3: Btw I worked on some statistical things about mining today. Looks like getwork rate (how often miner ask for new getwork) affect real miner performance a lot.
497 2010-12-27 21:43:01 <nanotube> also, consider joining the existing ,,pool
498 2010-12-27 21:43:02 <gribble> No fancy GPU farm, and don't want to wait for months for a block gen? Join the mining pool! http://mining.bitcoin.cz/
499 2010-12-27 21:43:26 <Motoma> Are there any miners that don't require GPU libraries?
500 2010-12-27 21:43:28 <slush> Diablo-D3: It is much better to take one getwork, solve it as fast as possible by all available gpus and then ask for another THAN ask for more getwork
501 2010-12-27 21:43:47 <nanotube> Motoma: yes, the cpuminer. :)
502 2010-12-27 21:43:52 <slush> Diablo-D3: statistically, it make huge difference in real mining success
503 2010-12-27 21:44:24 <Motoma> nanotube: Is that listed on the wiki?
504 2010-12-27 21:44:51 <OneFixt> slush: could you explain how more getwork affects the performance?
505 2010-12-27 21:44:54 <nanotube> Motoma: yes the wiki lists the available miners.
506 2010-12-27 21:45:33 <nanotube> ;;bc,wiki category miners
507 2010-12-27 21:45:34 <gribble> https://en.bitcoin.it/wiki/Category:Bitcoin_miners | Dec 17, 2010 ... Bitcoin miners are those that generate blocks for the blockchain. Pages in category "Bitcoin miners". The following 2 pages are in this ...
508 2010-12-27 21:45:38 <slush> OneFixt: If you work on invalid block (because there is another block in block chain in network), you are wasting gpu cycles for nothing
509 2010-12-27 21:45:58 <Motoma> nanotube: I guess I'm blind then. Do you happen to have a link to download cpuminer?
510 2010-12-27 21:45:58 <nanotube> Motoma: mm i see that the cpuminer page doesn't actually link to the cpuminer... time to remedy that. :)
511 2010-12-27 21:46:10 <slush> OneFixt: When you have one or two getworks, you lost (say) 5 seconds of hash crunching
512 2010-12-27 21:46:21 <nanotube> https://github.com/jgarzik/cpuminer
513 2010-12-27 21:46:24 <nanotube> Motoma: --^
514 2010-12-27 21:46:34 <slush> OneFixt: When you have >100 workers, you lost much more of performance
515 2010-12-27 21:46:37 <Motoma> nanotube: Thanks a million!
516 2010-12-27 21:46:44 <OneFixt> slush: couldn't you have 10 getworks which give you the a block that is currently valid?
517 2010-12-27 21:47:33 <OneFixt> slush: or are you referring to a particular implementation (perhaps a pooled miner)?
518 2010-12-27 21:47:36 <slush> OneFixt: 1. ask for getwork 2. network announce new block 3. miners submitted potentially valid block 4. block is invalid, because 2)
519 2010-12-27 21:47:50 <slush> OneFixt: It is not directly related to pool
520 2010-12-27 21:47:52 <Diablo-D3> [05:43:28] <slush> Diablo-D3: It is much better to take one getwork, solve it as fast as possible by all available gpus and then ask for another THAN ask for more getwork
521 2010-12-27 21:47:57 <Diablo-D3> [05:43:52] <slush> Diablo-D3: statistically, it make huge difference in real mining success
522 2010-12-27 21:47:59 <Diablo-D3> slush: no it doesnt.
523 2010-12-27 21:48:30 <slush> Diablo-D3: Yes, it does. If you are asking for getwork asap, you minimize time when you are working on invalid job
524 2010-12-27 21:48:42 <OneFixt> slush: technically you don't have to finish the entire getwork before getting another one, so it's not getwork itself which affects performance, but how long you spend before checking for new work
525 2010-12-27 21:49:01 <Diablo-D3> slush: yes, but using threaded design, I can, say, run 4 getworks simultaniously
526 2010-12-27 21:49:25 <Diablo-D3> slush: there is no difference between my miner on 4 gpus, 4 m0's, or 4 different machines
527 2010-12-27 21:49:37 <Diablo-D3> slush: they all have the same exact chances of finding a valid block
528 2010-12-27 21:49:37 <slush> OneFixt: Of course, you are right. It is fully on miner implementation, how often he ask for new miner
529 2010-12-27 21:50:11 <Diablo-D3> slush: even when I ran 3 getworks per gpu, and on that 4 gpu example, they STILL have the same exact chances of finding it
530 2010-12-27 21:50:21 <slush> OneFixt: Diablo's in doing it very good; I see very small latencies between getwork and submitting shares in my pool. Unfortunately m0mchil's miner can crunch one task for more than 30 seconds. I think it can be an issue
531 2010-12-27 21:50:49 <Diablo-D3> no, m0's runs exactly like my old design, but only a single executor.
532 2010-12-27 21:50:59 <Diablo-D3> it flushes when either a) it submits a block, b) 5 seconds are up
533 2010-12-27 21:51:10 <slush> Diablo-D3: No, it is not the same if you run one GPU crunching 1ghash/s and 1000 gpus crunching 1mhash
534 2010-12-27 21:51:19 <Diablo-D3> slush: yup, its exactly the same.
535 2010-12-27 21:51:41 <slush> Diablo-D3: No, I see m0mchil is submitting jobs which are sometimes older than 20, 30 seconds
536 2010-12-27 21:51:56 <Diablo-D3> slush: you're assuming there IS a valid network difficulty block in a job AND they're not evenly distributed
537 2010-12-27 21:52:02 <slush> Diablo-D3: I don't know how it does, but it does.
538 2010-12-27 21:52:08 <Diablo-D3> valid network blocks have a distribution bias of around 0%
539 2010-12-27 21:52:14 <Diablo-D3> slush: weird
540 2010-12-27 21:52:16 <Diablo-D3> that indicates a bug
541 2010-12-27 21:52:22 <Diablo-D3> but thats not a design error
542 2010-12-27 21:52:25 <Diablo-D3> just a bug
543 2010-12-27 21:52:37 <slush> Share found by m0mchil.sirius,       checkwork 0.013 sec, job 14.083 sec, 83f50b92
544 2010-12-27 21:52:40 <Diablo-D3> m0's intent is pretty clear in his design
545 2010-12-27 21:52:49 <slush> Diablo-D3: it is few seconds old line from log
546 2010-12-27 21:52:53 <Diablo-D3> hmm
547 2010-12-27 21:52:54 <slush> 14 second old job
548 2010-12-27 21:52:58 <Diablo-D3> wonder if m0 has it cranked up
549 2010-12-27 21:53:05 <Diablo-D3> even mine can do that if you tell it to
550 2010-12-27 21:53:07 <ArtForz> *shrug* imo the whole getwork approach is bass-ackwards
551 2010-12-27 21:53:26 <Diablo-D3> ArtForz: yeah, but push back through forwards connection is a tad ass backards too
552 2010-12-27 21:53:41 <ArtForz> huh?
553 2010-12-27 21:53:43 <Diablo-D3> I would mine a parallel protocol to send events
554 2010-12-27 21:53:54 <nanotube> Motoma: np :)
555 2010-12-27 21:53:56 <Diablo-D3> just to tell miners to force change
556 2010-12-27 21:54:06 <ArtForz> hmmm... no need for another protocol
557 2010-12-27 21:54:10 <Diablo-D3> ArtForz: client a connects to server b, b tells a everything
558 2010-12-27 21:54:12 <ArtForz> just use the normal bitcoin protocol
559 2010-12-27 21:54:16 <slush> > Diablo-D3: slush: yup, its exactly the same.
560 2010-12-27 21:54:23 <Diablo-D3> slush: its exactly the same.
561 2010-12-27 21:54:27 <INEEDMONEY> http://www.stickam.com/aranna
562 2010-12-27 21:54:37 <Diablo-D3> valid network difficulty blocks are fully randomly distributed
563 2010-12-27 21:54:58 <slush> Because they asked for 1000 getworks and are crunching it for (say)  5 seconds. And THEN they observe their work is outdated
564 2010-12-27 21:54:59 <Diablo-D3> the chances of finding one, throughout the entire valid search space, is the same for every single chance
565 2010-12-27 21:55:00 <ArtForz> if miner also connects to node as a client, it geta inv msgs for new blocks ...
566 2010-12-27 21:55:06 <slush> So you spend 5000 second for nothing
567 2010-12-27 21:55:13 <Diablo-D3> slush: not 5000 seconds.]
568 2010-12-27 21:55:30 <slush> Diablo-D3: When you have one GPU with 1ghash, it is asking for job much more often
569 2010-12-27 21:55:33 <Diablo-D3> slush: 5 seconds of 1000 getworks @ 1 mhash is the same wasted effort as 5 seconds of 1 getwork @ 1 ghash
570 2010-12-27 21:55:34 <slush> Diablo-D3: why not?
571 2010-12-27 21:56:09 <slush> Diablo-D3: But you will end with nonces before 5 second timeout!
572 2010-12-27 21:56:23 <slush> Diablo-D3: So 1ghash gpu will ask for getwork say every 2.5 second
573 2010-12-27 21:56:25 <slush> or less
574 2010-12-27 21:56:31 <Diablo-D3> thats only because I run out of nonce room
575 2010-12-27 21:56:51 <Diablo-D3> thats a side effect only
576 2010-12-27 21:57:02 <slush> Diablo-D3: Yes, exactly. But 1mhash miners will NOT end up with nonces
577 2010-12-27 21:57:20 <Diablo-D3> not only that, having to pull getwork sooner than 5 seconds _will_ halt the threads sooner
578 2010-12-27 21:57:28 <Diablo-D3> which means you're losing out on hash time
579 2010-12-27 21:57:28 <slush> Of course it is side effect. But it means we cannot count 1000*1 as 1000*1
580 2010-12-27 21:57:40 <Diablo-D3> so you're STILL leaking performance
581 2010-12-27 21:58:40 <slush> Yes, still, but statistically it is much more when doing parallel stuff with 100 workers
582 2010-12-27 21:58:45 <Diablo-D3> lets say the 1000 getworks make 1000 sharesa
583 2010-12-27 21:58:57 <Diablo-D3> the 1 ghash getwork will make 1001 shares.
584 2010-12-27 21:58:59 <Diablo-D3> give or take
585 2010-12-27 21:59:18 <slush> Diablo-D3: It is absolutely NOT about shares
586 2010-12-27 21:59:23 <slush> It is about working on outdated job
587 2010-12-27 21:59:25 <Diablo-D3> slush: as long as you have parallel archs, work in applications like that can be wasted
588 2010-12-27 21:59:43 <Diablo-D3> what happens if you have 1000 pipes on a gpu, and the first pipe makes a winner?
589 2010-12-27 21:59:47 <Diablo-D3> the 999 after are wasted.
590 2010-12-27 22:00:03 <Diablo-D3> but they're not really wasted, because you cant know what a winner is before you try it
591 2010-12-27 22:00:10 <Diablo-D3> the same also applies to distributing across computers.
592 2010-12-27 22:00:14 <Diablo-D3> there IS latency
593 2010-12-27 22:00:28 <Diablo-D3> slush: I mean, hell, I effectively buffer 3 kernel executions
594 2010-12-27 22:00:34 <Diablo-D3> the first 1 can win, the next 2 can be wasted
595 2010-12-27 22:00:49 <slush> Diablo-D3: I'm definitely not talking about your implementation
596 2010-12-27 22:00:52 <Diablo-D3> when you have any sort of optimized parallel work arch, time will be efficiently wasted.
597 2010-12-27 22:01:03 <Diablo-D3> slush: no, but you fail to understand how stuff like this works
598 2010-12-27 22:01:03 <slush> Diablo-D3: I'm talking about buffering getwork for 30 seconds. It is definitely bad
599 2010-12-27 22:01:13 <Diablo-D3> no one buffers getwork for 30 seconds
600 2010-12-27 22:01:25 <slush> Diablo-D3: My logs are talking something different
601 2010-12-27 22:01:41 <slush> I cannot imagine I made an error with time.time() - job.timestamp
602 2010-12-27 22:01:44 <Diablo-D3> I can do that with mine too
603 2010-12-27 22:02:08 <Diablo-D3> use -g with values above 5
604 2010-12-27 22:02:41 <slush> Diablo-D3: of course it _is_ possible. But all crunching which is after new block announcement is simply lost work
605 2010-12-27 22:02:48 <Diablo-D3> yes, it is lost work
606 2010-12-27 22:02:54 <slush> Diablo-D3: If you increate this timeout, you will lose more work
607 2010-12-27 22:02:55 <Diablo-D3> but its a side effect of having an efficient parallel system
608 2010-12-27 22:03:01 <slush> Diablo-D3: great. This is what I'm talking about
609 2010-12-27 22:03:11 <Diablo-D3> yes, and Im saying its unavoidable
610 2010-12-27 22:03:15 <Diablo-D3> the system IS designed correctly.
611 2010-12-27 22:03:22 <slush> Diablo-D3: If there is 5ghash in pool, but all of those guys have -g 20, there is plenty of lost work
612 2010-12-27 22:03:27 <Diablo-D3> every level loses a little bit of work at every level
613 2010-12-27 22:03:33 <slush> It does not mean pool have _effective_ 5ghash
614 2010-12-27 22:03:41 <Diablo-D3> slush: btw
615 2010-12-27 22:03:44 <Diablo-D3> theres also another problem
616 2010-12-27 22:03:53 <slush> But it is definitely not related to pool architecture itself, it is related to miner settings
617 2010-12-27 22:04:18 <Diablo-D3> you're talking about a tenth of a percent of lost actual hases.
618 2010-12-27 22:04:21 <Diablo-D3> *hashes
619 2010-12-27 22:04:35 <nanotube> slush: how is that per-block-reward-history thing going? :)
620 2010-12-27 22:04:41 <Diablo-D3> you can only intelligently optimize it
621 2010-12-27 22:05:08 <Diablo-D3> such as having a p2p event network when miners say "I found a block!"
622 2010-12-27 22:05:12 <slush> nanotube: I'm not on my own computer yet; Working in eclipse on netbook is horrible :)
623 2010-12-27 22:05:15 <Diablo-D3> everyone else will cut and flush
624 2010-12-27 22:05:24 <nanotube> slush: heh ic
625 2010-12-27 22:05:38 <Diablo-D3> heh eclipse :D
626 2010-12-27 22:05:38 <slush> nanotube: So I'm just doing some stats stuff inside, no large project
627 2010-12-27 22:06:14 <INEEDMONEY> eclipse isn't bad for java/android
628 2010-12-27 22:06:18 <slush> Diablo-D3: Not a problem on my 2x24", but it is strange on 800x480 :))
629 2010-12-27 22:06:44 <EvanR-work> JSON key 'midstate' not found
630 2010-12-27 22:06:45 <EvanR-work> work decode failed
631 2010-12-27 22:06:49 <EvanR-work> cpuminer
632 2010-12-27 22:06:54 <EvanR-work> 0.2.2
633 2010-12-27 22:07:03 <Diablo-D3> its nice on a single 1920x1200 =P
634 2010-12-27 22:07:19 <Diablo-D3> EvanR-work: sounds like you're using it with an incompatible version of bitcoin
635 2010-12-27 22:07:34 <Diablo-D3> or cpuminer isnt dealing with failure states right
636 2010-12-27 22:07:51 <slush> Diablo-D3: Well, we are finally talking about same thing :)
637 2010-12-27 22:07:52 <nanotube> EvanR-work: latest cpuminer is 0.3.3. try the latest version, maybe
638 2010-12-27 22:08:10 <EvanR-work> alright
639 2010-12-27 22:08:25 <EvanR-work> 'not handling failure states right' sounds familiar
640 2010-12-27 22:08:33 <slush> Diablo-D3: it would be great to optimize this overhead a little. I think it is really significant number
641 2010-12-27 22:08:35 <nanotube> no guarantees... but just saying before you go reporting any bugs, it's good to get the latest code.
642 2010-12-27 22:08:58 <EvanR-work> first time this happened in a week of testing
643 2010-12-27 22:09:00 <EvanR-work> but yeah
644 2010-12-27 22:09:32 <slush> Diablo-D3: Mainly, I will talk with m0mchil about it. Looks like his miner is working on job for longer time by default (diablominers have smaller difference between getwork and submit)
645 2010-12-27 22:10:06 <Diablo-D3> slush: getwork and submit cant realistically exceed 5 seconds, minus round trip time to gpu and to server
646 2010-12-27 22:10:27 <Diablo-D3> so -f 1 could, say, be 5 + 3 +, eh, 1
647 2010-12-27 22:10:39 <slush> Diablo-D3: My stats for ~6000 shares are talking about average almost 7 seconds
648 2010-12-27 22:10:43 <Diablo-D3> doesnt mean it was working on it for 9 seconds
649 2010-12-27 22:10:51 <Diablo-D3> it just took 9 seconds to get to the server
650 2010-12-27 22:10:58 <Diablo-D3> and not even 9
651 2010-12-27 22:11:04 <slush> Diablo-D3: I expect that some users are manually changing this value to 'improve' their latency
652 2010-12-27 22:11:08 <Diablo-D3> the job was issued at the end of the 5th second
653 2010-12-27 22:11:15 <Diablo-D3> slush: probably
654 2010-12-27 22:11:20 <Diablo-D3> thats why I added it on mine
655 2010-12-27 22:11:38 <Diablo-D3> you increase the chances of lost work to decrease the chances of your miner halting
656 2010-12-27 22:11:45 <Diablo-D3> slush: oh, and btw
657 2010-12-27 22:11:52 <Diablo-D3> lost work can only happen every 10 minutes
658 2010-12-27 22:12:06 <slush> Diablo-D3: of course; but for all miners at the same time
659 2010-12-27 22:12:17 <Diablo-D3> since these are fake getworks, the rotating of getworks on share producting is meaningless
660 2010-12-27 22:12:32 <Diablo-D3> it only happens during that little bit of time after a new block
661 2010-12-27 22:13:18 <Diablo-D3> so you're having this huge argument about 5 seconds every 10 minutes.
662 2010-12-27 22:13:39 <gribble> 0.00833333333333
663 2010-12-27 22:13:39 <nanotube> ;;math calc 5/600
664 2010-12-27 22:13:42 <slush> Today I implemented algorithm to count also 'effective' shares; it is for internal statistics. We will see if difference will be significant
665 2010-12-27 22:13:54 <nanotube> Diablo-D3: approx 1% loss...
666 2010-12-27 22:14:16 <Diablo-D3> nanotube: yes, which is really nothing
667 2010-12-27 22:14:21 <slush> Diablo-D3: Teoretically; we will see
668 2010-12-27 22:14:37 <Diablo-D3> I mean, my new miner is accidently losing up to 20% of performance
669 2010-12-27 22:14:37 <slush> Diablo-D3: I will post exact numbers in few days
670 2010-12-27 22:14:42 <da2ce7> slush, what dose this  'effective' shares thing do?
671 2010-12-27 22:15:10 <slush> da2ce7: check, if from time of getwork is there new bitcoin bloc
672 2010-12-27 22:15:33 <slush> block; means possible submit will never be valid
673 2010-12-27 22:15:57 <slush> Diablo-D3: your 20% perf lost is another issue ;)
674 2010-12-27 22:16:05 <Diablo-D3> yes, a much larger one
675 2010-12-27 22:16:35 <da2ce7> so If I submit work after a new block is released, that work gets ignored.
676 2010-12-27 22:16:55 <slush> Diablo-D3: yes, but this one does not affect pool stats itself; it only looks like pool is slower
677 2010-12-27 22:17:03 <Diablo-D3> slush: I dont particularly want to step back on single global getwork, but it'll have to be done
678 2010-12-27 22:17:20 <Diablo-D3> slush: at least, for now]
679 2010-12-27 22:17:38 <Diablo-D3> slush: my original getwork spam design was more optimal overall
680 2010-12-27 22:17:40 <slush> da2ce7: not now, I implemented it for statistical purposes
681 2010-12-27 22:17:48 <Diablo-D3> nothing could stop anything else except the driver and hardware itself
682 2010-12-27 22:18:23 <slush> da2ce7: I will see how much those 'invalid shares' affect whole cluster stats; then I decide
683 2010-12-27 22:18:37 <Diablo-D3> slush: you'll probably get about 1% failure
684 2010-12-27 22:18:38 <Diablo-D3> give or take
685 2010-12-27 22:18:57 <Diablo-D3> around the same rate of 5 seconds every 10 minutes, approx
686 2010-12-27 22:19:07 <slush> Diablo-D3: You are probably right and I hope there will not be bigger difference
687 2010-12-27 22:19:16 <Diablo-D3> if its bigger, there may be other issues
688 2010-12-27 22:19:26 <Sherpa> http://tech.slashdot.org/story/10/12/27/148258/After-IPv4-How-Will-the-Internet-Function
689 2010-12-27 22:19:37 <slush> Diablo-D3: maybe; this is reason why I'm working on those stats
690 2010-12-27 22:19:43 <INEEDMONEY> I need to talk to Satoshi...
691 2010-12-27 22:19:46 <INEEDMONEY> about capitalism
692 2010-12-27 22:19:59 <Diablo-D3> INEEDMONEY: I dont think satoshi understans capitalism
693 2010-12-27 22:20:02 <INEEDMONEY> you guys want your programs to make money, right?
694 2010-12-27 22:20:13 <Diablo-D3> Im not particularly doing it for the money
695 2010-12-27 22:20:14 <INEEDMONEY> Diablo-D3: I know, and I do not understand bitcoin, he and I can help each other
696 2010-12-27 22:20:23 <ArtForz> /ignore troll
697 2010-12-27 22:20:28 <INEEDMONEY> Diablo-D3: do it for tips, man
698 2010-12-27 22:20:32 <Diablo-D3> ArtForz: but they're so much fun =/
699 2010-12-27 22:20:35 <slush> da2ce7: if the difference will be only  1-2% I will let it be; it difference will be bigger for some reason, maybe changing share counting to only valid block should be more fair
700 2010-12-27 22:20:39 <INEEDMONEY> I only work for tips
701 2010-12-27 22:20:41 <INEEDMONEY> on every job I do
702 2010-12-27 22:20:41 <slush> da2ce7: we will see
703 2010-12-27 22:20:43 <INEEDMONEY> and I run a bank...
704 2010-12-27 22:20:57 <Diablo-D3> slush: btw, your statistics granularity will be chunky for a few days
705 2010-12-27 22:21:52 <Diablo-D3> well, maybe a day
706 2010-12-27 22:21:58 <da2ce7> :D
707 2010-12-27 22:22:16 <Diablo-D3> you should have like a million shares worth before you look
708 2010-12-27 22:22:38 <slush> Diablo-D3: I'm not gathering stats data yet; I need system upgrade and don't want to do it from Asus EEE
709 2010-12-27 22:22:49 <Diablo-D3> hee
710 2010-12-27 22:23:05 <Diablo-D3> actually
711 2010-12-27 22:23:11 <Diablo-D3> lets see if I can roll this all down into a single thread
712 2010-12-27 22:23:28 <slush> fingers crossed :)
713 2010-12-27 22:23:44 <Diablo-D3> it'll basically turn into art's though
714 2010-12-27 22:23:54 <Diablo-D3> and Im not really losing out here
715 2010-12-27 22:24:11 <Diablo-D3> you'd have to produce H==0s at a rate faster than a single core can process
716 2010-12-27 22:24:29 <Diablo-D3> ArtFarm couldn't do that even if he plugged it all into the same machine
717 2010-12-27 22:24:41 <ArtForz> kinda hard to do
718 2010-12-27 22:24:56 <ArtForz> even a slow core can do at least a few 100kH/s
719 2010-12-27 22:25:00 <Diablo-D3> yeah
720 2010-12-27 22:25:21 <Diablo-D3> assuming full check is half the speed, I can still do a million a second
721 2010-12-27 22:25:30 <Diablo-D3> er half a million
722 2010-12-27 22:26:12 <ArtForz> half a Ph/s ... shouldnt be a problem in the near fuiture
723 2010-12-27 22:26:37 <slush> nanotube: Try to update your miner to newest miner with asm32 algo
724 2010-12-27 22:27:04 <nanotube> slush: i'm on a 64bit box... i guess i can try to compile for 32bit arch...
725 2010-12-27 22:27:11 <nanotube> slush: why you think it's much faster?
726 2010-12-27 22:27:56 <slush> nanotube: I don't think 64bit is an issue. afaik cryptopp_asm32 is the fastest implementation in jgarzik miner
727 2010-12-27 22:28:08 <INEEDMONEY> my game design startup is sitting on 3000USD and needs some better Java or C# developers who are interested in Bitcoin.
728 2010-12-27 22:28:10 <slush> nanotube: maybe you reach magical 1000khash/s :)
729 2010-12-27 22:28:15 <nanotube> slush: hehe
730 2010-12-27 22:29:45 <EvanR-work> is 3000 enough to make c# or java produce a working program
731 2010-12-27 22:30:26 <INEEDMONEY> you're hired, buddy
732 2010-12-27 22:30:44 <EvanR-work> wouldnt touch those systems with a 10 ft pole
733 2010-12-27 22:31:01 <INEEDMONEY> EvanR it is for me
734 2010-12-27 22:31:09 <INEEDMONEY> since I just do it in my spare time
735 2010-12-27 22:31:12 <INEEDMONEY> and I know a fair bit of programming
736 2010-12-27 22:31:26 <INEEDMONEY> willing to tutor/help anyone with programming
737 2010-12-27 22:35:00 <slush> Diablo-D3: how much time needs current performance fix?
738 2010-12-27 22:35:12 <slush> Diablo-D3: I have no idea how hard is to fix it
739 2010-12-27 22:36:28 <Diablo-D3> slush: probably a day or two
740 2010-12-27 22:36:40 <Diablo-D3> [06:27:56] <slush> nanotube: I don't think 64bit is an issue. afaik cryptopp_asm32 is the fastest implementation in jgarzik miner
741 2010-12-27 22:36:46 <Diablo-D3> but it produces shit
742 2010-12-27 22:36:55 <Diablo-D3> he should really disable it until he fixes it
743 2010-12-27 22:37:04 <slush> Diablo-D3: he is fixed it already
744 2010-12-27 22:37:06 <slush> Diablo-D3: today
745 2010-12-27 22:37:09 <Diablo-D3> [06:29:45] <EvanR-work> is 3000 enough to make c# or java produce a working program
746 2010-12-27 22:37:18 <Diablo-D3> EvanR-work: at my rates? no.
747 2010-12-27 22:37:23 <Diablo-D3> slush: yay
748 2010-12-27 22:38:16 <EvanR-work> was a rhetorical question
749 2010-12-27 22:40:42 <slush> I have question related to bitcoin wallet.dat. AFAIK it is mainly database of private keys of my wallets, right?
750 2010-12-27 22:40:59 <Diablo-D3> slush: yeah basically
751 2010-12-27 22:41:04 <Diablo-D3> its got a little bit more in it iirc
752 2010-12-27 22:41:59 <slush> So when I backup those my private keys, I can recover my current balance any time, because all transfers _to_ and _from_ my wallets are public, right?
753 2010-12-27 22:42:10 <slush> I'm not talking about current implementation, just about possibility to that
754 2010-12-27 22:43:02 <Diablo-D3> slush: yes
755 2010-12-27 22:43:05 <Cusipzzz> yes, you should be whole again one all blocks are downloaded ?
756 2010-12-27 22:43:36 <Diablo-D3> slush: wallet.dat also includes iirc the stuff for addresses that are you
757 2010-12-27 22:43:51 <slush> That's pretty cool; I think current implementation (backup  wallet after every transaction) is insane
758 2010-12-27 22:44:48 <Diablo-D3> it doesnt really need it
759 2010-12-27 22:44:49 <slush> Diablo-D3: Yes, but I can generate few addresses and use only them; of course it is not safe (privacy), but I dont care about it
760 2010-12-27 22:45:07 <Diablo-D3> if your hardware is sane, you should backup regularly onto external media monthly or whatever
761 2010-12-27 22:45:14 <Diablo-D3> when you backup everything else
762 2010-12-27 22:45:33 <slush> Diablo-D3: But it is impossible to make backup to some persistent media and lock it into safe
763 2010-12-27 22:46:18 <Diablo-D3> slush: in what way?
764 2010-12-27 22:46:23 <Diablo-D3> I can drop my external hd into a safe easily.
765 2010-12-27 22:46:40 <slush> With some recover tool like  that (exporting wallets and then rebuilding balances for them from block chain) should be my more safe
766 2010-12-27 22:47:04 <slush> Diablo-D3: I'm talking about physical safe, say, room in some building with  security guard
767 2010-12-27 22:47:29 <INEEDMONEY> http://www.bitcoin.org/smf/index.php?topic=2486.0
768 2010-12-27 22:47:31 <bitbot> In the interests of capitalism and programming...
769 2010-12-27 22:47:31 <INEEDMONEY> discuss
770 2010-12-27 22:47:42 <INEEDMONEY> I want my client to do 0.00999999999999
771 2010-12-27 22:47:53 <slush> Diablo-D3: But I read in all FAQ that users should backup after every transaction and AFAIK there is no recovery tool I'm talking about
772 2010-12-27 22:48:05 <Diablo-D3> slush: thats over-paranoia really
773 2010-12-27 22:48:24 <slush> Diablo-D3: Really? Even if I have 10.000 $ in my bitcoin wallet?
774 2010-12-27 22:48:39 <Diablo-D3> you should treat it like any mission critical data.
775 2010-12-27 22:48:42 <Cusipzzz> once you've had a hd crash, you will backup after every transaction
776 2010-12-27 22:48:50 <slush> Diablo-D3: I don't think anything is enough paranoic when we are talking about money :)
777 2010-12-27 22:49:13 <Cusipzzz> i lost some coins in a hd crash, learned my lesson fairly cheap - could have been worse