 Show Posts
|
|
Pages: « 1 2 3 [4] 5 »
|
Hey folks - we should respond to these questions regarding tsiv's bounty: https://bitcointalk.org/index.php?topic=656841.msg7536834#msg7536834I would like to comment briefly on awarding the bounty since I am responsible for the largest pledge (I think).
From what I have seen, tsiv should receive a bounty payment. I'm not sure he should receive the full amount quite yet. (Though I am also not saying he should not.)
I would like to hear public comments on the following points before reaching a conclusion.
My criteria for awarding the bounty are:
1. Quality. Is the miner 100% reliable, or does it crash or hang?
 If quality was important, they would never have used cryptonote (most crappy wallet/network usage disk usage, user unfriendliness, etc...) very picky for such a coin... may-be they should first try to improve the quality of their wallet before judging on the quality of what other do I agree! Holy crap u totally nailed them on that one! XD The bigger issues are, what is the main purpose of currency? If you wanted to save a nest egg..it would be better for you to get a 401-K, but these crypto-note coins have a major major flaw: The Blockchain being downloaded is going over a GB, and the wallet uses a ton of data to uphold the chain. With that being said the flaw is how do you get to mobile? Answer..you can't unless you are willing to spending $$$ in mobile data plans..crypto-note is doomed in my opinion..unless they are able to fix this. |
|
|
|
So my thought is: With the white paper describing the FFT process as opposed to S-Box lookup, couldn't we use cuFFT (as a part of the Cuda Toolbox) to elminate the need for any of the S-boxes in the encryption methods, and then change the formula's given for the each hash and run them all through cuFFT. Maybe I am missing something in the math..I probably am, but I think if we could implement this, we could see huge advantages... Yes? No?
I think there is an algo which is already solved through fft (though my understanding of those algo is pretty scarce at the moment) I checked the code and I am not coming up with any calls for cuFFT. In the SPH simd.c it does use FFT to solve it. I wonder why we don't use cudaSIMD as well as cuFFT. So..I think maybe their are hidden improvements that could be made? |
|
|
|
|
So my thought is: With the white paper describing the FFT process as opposed to S-Box lookup, couldn't we use cuFFT (as a part of the Cuda Toolbox) to elminate the need for any of the S-boxes in the encryption methods, and then change the formula's given for the each hash and run them all through cuFFT. Maybe I am missing something in the math..I probably am, but I think if we could implement this, we could see huge advantages... Yes? No?
|
|
|
|
|
I hold Darkcoin, but I mine and hold Gun coin. I think Gun coin is really a good niche market coin. I fully support it, and hope that it does well! I will continue to mine and hold!
|
|
|
|
I run a H81 BTC Pro with 6 EVGA 750Ti's (without the 6-pin PCI-E power), I have 5 cards on 16x-1x powered risers, and the 1 card in the x16 slot on a 16x-16x unpowered riser. Rig has been running flawelessly for nearly 4 months now!
can you overclock them? im still reading massage #16128 trying to catch up to you guys.  That's cool..I would suggest holding questions until you get caught up to the last couple of days (maybe a week or so)...some people come, and some people go, and they may not be around to answer your questions. |
|
|
|
I tried the nvidia test drive. I got DC maybe 15 seconds after runing tsiv CCminer and now if I try to reconnect its telling me that my minutes are all exhausted.  It probably crashed the driver like it does on our computers, especially since it is Keppler. I would suggest making sure you use a low threadsxblock, although if it seems that as soon as you crash the driver it kicks you out... |
|
|
|
|
As a matter of fact..I wonder if they are using Fast Fourier Transform chips in the mining ASIC's to obtain higher hash/rates...
|
|
|
|
I think using some tricks with the implentation of AES would speed the hash rate up probably into the Kh/s range....there is a cuda application that can pass through the AES using an integration attack on the implentation. (ie: The Integral attack on AES)
Any opinions?
cant you let the cpu do teh aes stuff? You could, but that won't speed anything up. It will get bottlenecked at the CPU. Now you may get 100-300h/s more...but if we used F.. we could get rid of the S-box alltogether and do the rounds a lot more efficiently and in parallel. |
|
|
|
|
I think using some tricks with the implentation of AES would speed the hash rate up probably into the Kh/s range....there is a cuda application that can pass through the AES using an integration attack on the implentation. (ie: The Integral attack on AES)
Any opinions?
|
|
|
|
Hey guys... just a heads up to anyone who HASN'T taken advantage of it... http://www.nvidia.com/object/trygrid.htmlfree 24hrs of K520 grid mining (usually 4x K520 cards) add tsiv's ccminer = profit *note* These do have a keep-alive time... so you must have the window active, and doing something in order to keep connection alive. It says for North America only... but I have 1 word... PROXY. My testing on amazon gave around 90-100 KH/s, so 4 of them will give around 400kh/s not sure if that is anywhere profitable though. at 400KH/s is about 0.2XMR/hour or 0.001008 BTC is about $0.57US/hr My mistake I meant 100h/s My mistake too...at 400 h/s it is significantly less...at 200h/s I got about .03XMR over 6 hours running 1 Nvidia K40 with tsiv code. |
|
|
|
Hey guys... just a heads up to anyone who HASN'T taken advantage of it... http://www.nvidia.com/object/trygrid.htmlfree 24hrs of K520 grid mining (usually 4x K520 cards) add tsiv's ccminer = profit *note* These do have a keep-alive time... so you must have the window active, and doing something in order to keep connection alive. It says for North America only... but I have 1 word... PROXY. My testing on amazon gave around 90-100 KH/s, so 4 of them will give around 400kh/s not sure if that is anywhere profitable though. at 400KH/s is about 0.2XMR/hour or 0.001008 BTC is about $0.57US/hr |
|
|
|
|
Any news coming out on Zetacoin soon?
|
|
|
|
|
G2.2xlarge is .65cents/hour ../instance
but depending on the speed of the instance you really need a profitable coin to make money on this.
|
|
|
|
The systems would have nvidia grid but super locked down. I am sure Amazon would have realised the flaw if they didnt  I am still mystified by the gpu instance with " 32 cores and 244GB memory". What would that one be? The Nvidia Grid GPU K2 has 8GB of GDDR. 8 * 32 GridK2's = 266GB of total memory..maybe it takes 22GB just as reserved for the system... |
|
|
|
CCMiner w/Cryptonight: I think the error with the windows drivers might be that one of the pointer addresses is either undeclared or unallocated. It definetly crashes because of a memory access error in the Nvidia driver dll's.
How did you find this out? Which operating system were you running? What nVidia driver version were you running? It went into DEBUG/DASSY and took a look. I am running Win 7, 64Bit, with drivers: GTX: 3288, and Quadro/Tesla: 3311 The K40 is running ok...not fast..at about 200H/s with -l 6x250 here is the image for the crash/debug: http://s18.postimg.org/nwv7wytah/CCMiner_Crypto_Night_Busted.jpg |
|
|
|
|
CCMiner w/Cryptonight: I think the error with the windows drivers might be that one of the pointer addresses is either undeclared or unallocated. It definetly crashes because of a memory access error in the Nvidia driver dll's.
|
|
|
|
The mining has turned into a kill be killed all out war, and you have 3 options these days. Option 1: sit with the herd mining what you can and slowly gain coins, and more and more equipment making the equipment designer/sellers rich..but not yourself unless you get lucky. And hope someone comes out with some software that will help your equipment along.
Option2: trade it and try your hand at speculation on the multitude of exchanges..I mean what fuck happened with Fractal coin..it exploded, but without its basis on its own name it is fucked!.(FRACTALs..which by the way using fractal cryptography with a new protocol not based on satoshi would be awesome!)
Option 3: this is the only real option to make real money: program your own code and use your own ideas on how to use the IMMENSE tools out there to make your own mining as fast and efficient as possible. I have many ideas and have spent the last month learning cuda C myself..since I seriously invested in the hardware, but I am not gonna share it because then you become part of option 1...so unless you are willing to pay the Devs what they are worth and that is in the Millions of dollars.. Then general public mining will die...
That would be okay but that would just kill cryptos ....99% of the population of the world could not code to save face. All your doing is encouraging more of "people with get more and people with not get nothing"....not the way you want to start going. On the other hand, I do think people need to be more open and donating to devs for the work. I just started to do that myself because I understand many things come with a cost, and the devs take time to make such a good item for all of us. That is exactly the point. The software is worth cold hard cash. @10 bitcoins every 3 days that 100bitcoins a month from what this guy is doing, no Dev is going to release code for this because the second you do your profits drop to maybe 1btc a day. So really the software for a new coin algo (like Monero or Boolberry) is worth at least 300-500BTC because that is what they are able to mine on their own.. |
|
|
|
|
