You do realise your trying to run a gpu miner on a cpu yeh?
may-be it is an nvidia cpu... (funny though as it worked) 
Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 306. (Read 3426976 times)
June 25, 2014, 08:23:51 AM
may-be it is an nvidia cpu... (funny though as it worked) 
June 25, 2014, 08:22:06 AM
IIRC, the GPU instances on AWS are kepler based, which performs poorly with tsiv's code (thus keeping Christian happy 
)
Would be super if we could at least get some better performance for Maxwell (the 750ti's), since that would not directly interfere with cloud mining operations... Any thoughts Christian?
Cheers,
~ Myagui
)Would be super if we could at least get some better performance for Maxwell (the 750ti's), since that would not directly interfere with cloud mining operations... Any thoughts Christian?
Cheers,
~ Myagui
June 25, 2014, 08:06:33 AM
You do realise your trying to run a gpu miner on a cpu yeh?
g2.2xlarge has nvidia grid and I am not entirely sure how fast or slow it is (yeah noob in that way 
) June 25, 2014, 07:58:14 AM
You do realise your trying to run a gpu miner on a cpu yeh?
June 25, 2014, 07:54:09 AM
one word, LOL
this is just the final hash and is barely noticeable in the total time spent. the real work is in the long cryptonight algorithm on the GPU.
I don't think tsiv is doing the cryptonight quite optimally yet.
But then I am not going the help with a) efficiency and b) deployment on EC2 because
that would be in direct competition to our lucrative operation here.
Well i guess at $2.80/hour for 32 cores and 244GB memory its not bad
What sort of hashrates do they give per core christian?
June 25, 2014, 06:28:19 AM
wonder if the same trick could be applied
...get it on djm
are those code lines all the trick to get cryptonight working on a gpu?
(actually, it is a lot more cryptic than jpc)
June 25, 2014, 06:23:19 AM
one word, LOL
this is just the final hash and is barely noticeable in the total time spent. the real work is in the long cryptonight algorithm on the GPU.
I don't think tsiv is doing the cryptonight quite optimally yet.
But then I am not going the help with a) efficiency and b) deployment on EC2 because
that would be in direct competition to our lucrative operation here.
Well i guess at $2.80/hour for 32 cores and 244GB memory its not bad
What sort of hashrates do they give per core christian?
June 25, 2014, 06:22:18 AM
wonder if the same trick could be applied
...get it on djm
are those code lines all the trick to get cryptonight working on a gpu?
He's conveniently ignoring the three massive for-loops that run 131072, 262144 and 131072 iterations of AES-encryption each
The final SINGLE keccak and SINGLE jh/groestl/blake/skein are literally nothing compared to the shit that comes before them
June 25, 2014, 06:18:55 AM
one word, LOL
this is just the final hash and is barely noticeable in the total time spent. the real work is in the long cryptonight algorithm on the GPU.
I don't think tsiv is doing the cryptonight quite optimally yet.
But then I am not going the help with a) efficiency and b) deployment on EC2 because
that would be in direct competition to our lucrative mining operation here.
June 25, 2014, 06:17:12 AM
Now, the whole cryptonote system is not geeky just crappy.
It even kills my network bandwidth when it runs.
It even kills my network bandwidth when it runs.
Get a test version of NetLimiter 4 ( http://www.netlimiter.com/nl4testing.php expires in September) and cap the upload speed of the daemon like I have...
The download speed of the blockchain will be unaffacted and still comes down as fast as it can, depending on other peers and availability.
Tick the UL Limit box for the bitmonerod.exe entry and it's capped. You can specify a different rate to cap it to by double clicking the number next to the tick box.
June 25, 2014, 06:09:19 AM
wonder if the same trick could be applied
...get it on djm
are those code lines all the trick to get cryptonight working on a gpu?
June 25, 2014, 05:40:17 AM
wonder if the same trick could be applied
Now that would be amazing
June 25, 2014, 05:38:13 AM
wonder if the same trick could be applied
June 25, 2014, 05:27:50 AM
I was looking into cryptonight algo (through tsiv code)... so basically it is just jackpotcoin first version (the buggy one) ?seriously?
cryptonight_keccakf((uint64_t*)&state->hs, 24);
switch( state->hs.b[0] & 3 ) {
case 0:
cn_blake((const uint8_t *)state, 200, (uint8_t *)hash);
break;
case 1:
cn_groestl((const BitSequence *)state, 200, (BitSequence *)hash);
break;
case 2:
cn_jh((const BitSequence *)state, 200, (BitSequence *)hash);
break;
case 3:
cn_skein((const BitSequence *)state, 200, (BitSequence *)hash);
break;
default:
break;
}
one word, LOL
June 25, 2014, 05:25:57 AM
I was looking into cryptonight algo (through tsiv code)... so basically it is just jackpotcoin first version (the buggy one) ?seriously?
cryptonight_keccakf((uint64_t*)&state->hs, 24);
switch( state->hs.b[0] & 3 ) {
case 0:
cn_blake((const uint8_t *)state, 200, (uint8_t *)hash);
break;
case 1:
cn_groestl((const BitSequence *)state, 200, (BitSequence *)hash);
break;
case 2:
cn_jh((const BitSequence *)state, 200, (BitSequence *)hash);
break;
case 3:
cn_skein((const BitSequence *)state, 200, (BitSequence *)hash);
break;
default:
break;
}
and jackpotcoin (ok there is may-be more round...)
KECCAK512_80(HASH.U8);
uint rounds = HASH.U4[0x00] & 0x00000007U;
for (uint i = 0; i < 8; i++) {
if (i < rounds) {
uint method = HASH.U4[0x00] & 0x00000003U;
if (method == 0) { BLAKE512(HASH.U8); }
else if (method == 1) { GROESTL512(HASH.U8, LT0, LT1, LT2, LT3, LT4, LT5, LT6, LT7); }
else if (method == 2) { JH512(HASH.U8); }
else if (method == 3) { SKEIN512(HASH.U8); }
}
}
June 25, 2014, 05:24:34 AM
I was looking into cryptonight algo (through tsiv code)... so basically it is just jackpotcoin first version (the buggy one) ?seriously?
June 25, 2014, 05:22:19 AM
I was looking into cryptonight algo (through tsiv code)... so basically it is just jackpotcoin first version (the buggy one) ?Strangely I am even not really surprised
June 25, 2014, 04:51:37 AM
Crytonote coin are too user unfriendly...
I think this will change in the next weeks.
CN coins are quite young, so maybe there is potential in future when the pools are stable and the UI is equal userfriendly as other coins.
Also the really huge blockcain is quite a big barrier at the moment.
June 25, 2014, 04:47:41 AM
Do we have to run monero daemon and simple wallet together? Or, can the wallet be open when the daemon isn't?
I think it can, however it won't be up to date... as it is the daemon job to watch the blockchainNow, the whole cryptonote system is not geeky just crappy.
It even kills my network bandwidth when it runs.
Thinking it has been around since btc, and it is still a piece of crap... I don't understand why anybody want to make cryptonotes...
June 25, 2014, 04:45:14 AM
Do we have to run monero daemon and simple wallet together? Or, can the wallet be open when the daemon isn't?
Jump to: