If the ref cards fan doesn't die within the first month or so (= manufacturing defect), it'll probably last near forever. The sapphire extreme 5830/5850... not so much, those have really cheap sleeve bearing fans. I'd be surprised if it lasted much over 6 months at 70%.
|
|
|
If they aren't overvolted... I'd say "yes, but". The AX850 is a pretty good PSU, just look at JGs review OCed 5870 at stock V is 200-220W, like 180 with mem @ 300 OCed 5970 at stock V is ~300W, about 270 with mem @ 300 Stock wattage is ~ the same as OCed with underclocked mem. So that's < 700W load on a PSU rated for 840W continous on 12V. now, theres the "but" part... simple, I wouldn't want to run a PSU at > 80% rated 24/7 for months.
|
|
|
Wait, you tried installing a accelero 5970 with NO INSTRUCTIONS?
Sounds like you killed your card improperly installing it then. The installation instructions that come with it (and are also available on ACs website) state VERY clearly to leave the backplate on and to *NOT* fully tighten the 4 x-plate screws around each GPU or the screws mounting the main cooler to the board (it's ok to fully tighten the 2 screws holding the VRM heatsink).
Overtighten the x-plates and you kill your DVI outputs. Overtighten the standoffs (especially with no backplate) and you bend the board so badly it's usually a paperweight.
|
|
|
5970s at stock V are < 300W. at 850/300 and stock V they're also < 300W. 930/1000 at 1150mV ~400W each So that PSU should be fine as long as you don't overvolt em.
|
|
|
Yep, does it power on with just green/black shorted? Also, it's generally a VERY bad idea to only load old/cheap 300/350W PSUs on +12. Those are pretty much all group-regulated designs taking feedback from +5V or +3.3V -> load +12 significantly with +5/+3.3 unloaded and +12 ends up *way* low, I've personally seen < 9.5V with 60% rated load on +12 and nothing on +5/+3.3.
|
|
|
Something like this should do the trick (disable the broken one-week crap, switch to solidcoin algo after block 14639). diff -Nur I0coinClientv1.0-linux-stable-orig/src/main.cpp I0coinClientv1.0-linux-stable-dev/src/main.cpp --- I0coinClientv1.0-linux-stable-orig/src/main.cpp 2011-08-18 23:15:50.000000000 +0200 +++ I0coinClientv1.0-linux-stable-dev/src/main.cpp 2011-08-25 08:23:52.143725519 +0200 @@ -644,7 +644,7 @@ return nSubsidy + nFees; } -unsigned int static GetNextWorkRequired(const CBlockIndex* pindexLast) +unsigned int static GetNextWorkRequired_OLD(const CBlockIndex* pindexLast) { const int64 nTargetTimespan = 7 * 24 * 60 * 60; // two weeks const int64 nTargetSpacing = 5 * 60; @@ -659,6 +659,7 @@ // Only change once per interval if ( nRemaining != 0) { +/* HORRIBLY BROKEN, *NEVER* use time() in here const CBlockIndex* pindexFirst = pindexLast; for (int i = 0; pindexFirst && i < nRemaining-1; i++) pindexFirst = pindexFirst->pprev; @@ -666,6 +667,7 @@ int64 rema = GetAdjustedTime() - pindexFirst->GetBlockTime(); if(rema < nTargetTimespan) +*/ return pindexLast->nBits; } @@ -691,6 +693,71 @@ if (bnNew > bnProofOfWorkLimit) bnNew = bnProofOfWorkLimit; + + /// debug print + printf("GetNextWorkRequired RETARGET\n"); + printf("nTargetTimespan = %"PRI64d" nActualTimespan = %"PRI64d"\n", nTargetTimespan, nActualTimespan); + printf("Before: %08x %s\n", pindexLast->nBits, CBigNum().SetCompact(pindexLast->nBits).getuint256().ToString().c_str()); + printf("After: %08x %s\n", bnNew.GetCompact(), bnNew.getuint256().ToString().c_str()); + + return bnNew.GetCompact(); +} + +//blatantly stolen from SolidCoin +unsigned int static GetNextWorkRequired(const CBlockIndex* pindexLast) +{ + const int64 nTargetTimespan = 12 * 60 * 60; // 12 hours + const int64 nTargetSpacing = 3 * 60; //3 minute blocks + const int64 nInterval = nTargetTimespan / nTargetSpacing; + + // Genesis block + if (pindexLast == NULL) + return bnProofOfWorkLimit.GetCompact(); + +//okay, maybe not this line + if ((pindexLast->nHeight+1) < 14640) + return GetNextWorkRequired_OLD(pindexLast); + + // Only change once per interval + if ((pindexLast->nHeight+1) % nInterval != 0) + return pindexLast->nBits; + + // Go back by what we want to be 14 days worth of blocks + const CBlockIndex* pindexFirst = pindexLast; + for (int i = 0; pindexFirst && i < nInterval-1; i++) + pindexFirst = pindexFirst->pprev; + assert(pindexFirst); + + // Limit adjustment step + int64 nActualTimespan = pindexLast->GetBlockTime() - pindexFirst->GetBlockTime(); + int64 nTwoPercent = nTargetTimespan/50; + //printf(" nActualTimespan = %"PRI64d" before bounds\n", nActualTimespan); + + if (nActualTimespan < nTargetTimespan) //is time taken for a block less than 3minutes? + { + //limit increase to a much lower amount than dictates to get past the pump-n-dump mining phase + //due to retargets being done more often it also needs to be lowered significantly from the 4x increase + if(nActualTimespan<(nTwoPercent*16)) //less than a minute? + nActualTimespan=(nTwoPercent*45); //pretend it was only 10% faster than desired + else if(nActualTimespan<(nTwoPercent*32)) //less than 2 minutes? + nActualTimespan=(nTwoPercent*47); //pretend it was only 6% faster than desired + else + nActualTimespan=(nTwoPercent*49); //pretend it was only 2% faster than desired + + //int64 nTime=nTargetTimespan-nActualTimespan; + //nActualTimespan = nTargetTimespan/2; + } + else if (nActualTimespan > nTargetTimespan*4) nActualTimespan = nTargetTimespan*4; + + // Retarget + CBigNum bnNew; + bnNew.SetCompact(pindexLast->nBits); + bnNew *= nActualTimespan; + bnNew /= nTargetTimespan; + + + if (bnNew > bnProofOfWorkLimit) + bnNew = bnProofOfWorkLimit; /// debug print printf("GetNextWorkRequired RETARGET\n");
|
|
|
The one week part of "one week or 2016 blocks" retargeting is b0rked. Downloading the chain with a fresh datadir makes it try to retarget at block 1. Then it tries to go back 2016 blocks. from block 1. Boom, NULL pointer. With a client that already has the blockchain up to some point in the past, it'll start disagreeing on what the target should be (again, mistrigger of the after-7-days stuff)... around now. Easy Fix: pull a Thomas and blatantly rip off solidcoins retargeting algo.
|
|
|
If they're anything like the fans on these: http://www.newegg.com/Product/Product.aspx?Item=N82E16814102873Once they start "growling" the single sleeve bearing (!) is totally shot. Takes about 3-5 months at 100%. And there's no way to lubricate them. Whoever decided to put a cheap-ass sleeve bearing fan right on top of the hottest part of the heatsink was a true genius.
|
|
|
That was a loooong time ago, nowadays it's < 1%.
|
|
|
50 != bulk. Hell, that's not even one tray.
|
|
|
Clients don't spend 0/unconfirmed coins they get from somewhere else. They will spend their own 0/unconfirmed change if left no other option. Original reason for that, "viral bitdust": A transaction that split 0.01 btc into 100s of tiny outputs and had no fee, so it staid at 0/unconf "forever". Clients that received one of the outputs sooner or later used them as inputs in their own transactions... and as a tx can only go into a block if all its inputs are in the same or older blocks... Yeah.
|
|
|
Simple, anyone mining back then and thinking like that most likely sold at $1.
|
|
|
Is this evaluation ok? Does it have no correlation with about 32SC ~ 3min and 50BTC ~ 10min ?
Target time/block doesn't enter this calculation (it only affects what difficulty a network will end up at at a given hashrate). Just block reward (thats where the 32 and 50 come from), difficulty and exchange rate. another way to put per-block reward, difficulty and exchange rate together: 1 share on BTC gets you about 50/1806098 BTC = 0.000027684 btc/share 1 share on SC at a exchage rate of 0.0017btc/sc gets you ... 32/2703*0.0017 BTC = 0.000020126 btc/share 0.000020126 / 0.000027684 = 0.726990319 -> you're making about 27.3% less mining SC and selling em than mining BTC. now... what exchange rate do we need to come out even? 50/1806098 = 32/2703*X or... X = (50 / 1806098) / (32 / 2703) X = 0.002338432 you can also swap things around to suit you X = (2703 / 32) / (1806098 / 50) X = 2703 / 1806098 * 50 / 32 or replace the * 50 / 32 with * 1.5625 X = 1.5625 * 2703 / 1806098 ... and it's still X = 0.002338432 Wonders of arithmetic!
|
|
|
Oh i see. Excellent explanation. Thanks again So it seems with my 2 GH/s i make slightly more mining solidcoins if the price is 0.0017 BTC/SC. I guess I'll wait 12 hours like you suggested and check it out then. Err... No. At the time of your post, SC difficulty was about 2700 (2.7k/32)/(1807k/50) = 0.00233467 that's what btc/sc would have to be to end up +-0. at 0.0017btc/sc, you're making roughly 30% less mining SC.
|
|
|
24pin ATX has 2 +12V pins and 8 GND pins, rated for 6A each. So thats a max current of 12A on the 12V side, 48A on the return. That's the short version of why splitting out GND isn't needed.
|
|
|
You realize a bitcoinhash is *2* sha256 blocks operations, right? Well, not exactly 2 thanks to some optimizations possible you can drop the last 3 rounds completely (they don't change H), and lose part of the previous round (you only need the E output of the 4th-to-last) Initial rounds can be optimized as well, as the last DWORD of hMerkleRoot and nTime/nBits don't change between loops, so you can drop the equivalent of ~3 rounds there as well. Same thing goes for optimizing/precalculating parts of the W mangling, as we're feeding in quite a bit of constants. Register access is basically free on GPUs (they mask reg r/w by pipelining 4 "threads" on the shader pipeline). Ch() can be done in 1 cycle, and Maj() in 2. Also, what LUT accesses? just hardcode the K constants in the instruction stream.
So while you came up with a somewhat reasonable result, you did so by pure chance using invalid assumptions and numbers.
|
|
|
I'd pile onto the "OC too high" heap, but... Look at the gpu-z log, that card is running stock 725/1000 clocks at stock voltage.
Edit: core and vddc temps look ok, maybe a tad high for my taste. But 30% fanspeed? what do the VDDC temps on the slave GPU look like?
|
|
|
Well, or just make sure to always leave at least a 0.01 balance or empty your wallet completely.
|
|
|
There you go, now the faucet is giving away 500 again
|
|
|
|