[ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 827.

69charger

full member

Activity: 173

Merit: 100

Vertcoin is currently very profitable.

cbuchner1

hero member

Activity: 756

Merit: 502

okay, this looks like it might overflow very quickly

Code:

uint32_t end_nonce = 0xffffffffU / opt_n_threads * (thr_id + 1) - 0x20;

assume only 1 thread mining, then end_nonce will be 0xffffffdfU

now the issue is that when we compute in chunks of about 30000 hashes at once
we will overflow work.data[19] so that the following condition is never met!

Code:

if (work.data[19] >= end_nonce)
stratum_gen_work(&stratum, &g_work);

TODO: replace the 0x20 with a value large enough so that overflow will
not occur. Alternatively make sure we never update work.data[19]
with a value smaller than the previous value. Instead write 0xffffffffU.

kalenen86

newbie

Activity: 15

Merit: 0

I am getting a "Stratum connection failed" message, is this because of the program or am i looking in the wrong direction for help/understanding of this error

Code:

[2014-02-09 18:21:10] Starting Stratum on stratum+tcp://eu.middlecoin.com:3333
[2014-02-09 18:21:10] 1 miner threads started, using 'scrypt' algorithm.
[2014-02-09 18:21:32] Stratum connection failed: Failed connect to eu.middlecoin
.com:3333; No error
[2014-02-09 18:21:32] ...retry after 15 seconds
[2014-02-09 18:22:08] Stratum connection failed: Failed connect to eu.middlecoin
.com:3333; No error
[2014-02-09 18:22:08] ...retry after 15 seconds
[2014-02-09 18:22:45] Stratum connection failed: Failed connect to eu.middlecoin
.com:3333; No error
[2014-02-09 18:22:45] ...retry after 15 seconds
[2014-02-09 18:23:21] Stratum connection failed: Failed connect to eu.middlecoin
.com:3333; No error
[2014-02-09 18:23:21] ...retry after 15 seconds
[2014-02-09 18:23:58] Stratum connection failed: Failed connect to eu.middlecoin
.com:3333; No error
[2014-02-09 18:23:58] ...retry after 15 seconds
[2014-02-09 18:24:35] Stratum connection failed: Failed connect to eu.middlecoin

Stanr010

member

Activity: 70

Merit: 10

Quote from: ivanlabrie on February 09, 2014, 12:25:38 PM

You guys never rest... Grin

Will keep an eye on your work, as usual, wish I could be of help. Undecided

I'm sure you can. His donation address is just over there >.>

ManIkWeet

full member

Activity: 182

Merit: 100

Quote from: bigjme on February 09, 2014, 12:26:24 PM

Quote from: ivanlabrie on February 09, 2014, 12:25:38 PM

You guys never rest... Grin

Will keep an eye on your work, as usual, wish I could be of help. Undecided

rest is for the lazy, hell im lazy and i still dont rest!

There's a trade-off between being lazy and making money as lazy as possible

bigjme

sr. member

Activity: 350

Merit: 250

Quote from: ivanlabrie on February 09, 2014, 12:25:38 PM

You guys never rest... Grin

Will keep an eye on your work, as usual, wish I could be of help. Undecided

rest is for the lazy, hell im lazy and i still dont rest!

ManIkWeet

full member

Activity: 182

Merit: 100

New github compile with the nonce print:

Code:

[2014-02-09 18:22:40] 1 miner threads started, using 'keccak' algorithm.
[2014-02-09 18:22:42] Stratum detected new block
[2014-02-09 18:22:43] GPU #0: GeForce GTX 780 with compute capability 3.5
[2014-02-09 18:22:43] GPU #0: interactive: 1, tex-cache: 0 , single-alloc: 0
[2014-02-09 18:22:43] GPU #0: 32 hashes / 0.0 MB per warp.
[2014-02-09 18:22:43] GPU #0: using launch configuration K2304x32
[2014-02-09 18:22:43] GPU #0: GeForce GTX 780, 107236 khash/s
[2014-02-09 18:23:04] GPU #0: GeForce GTX 780 submitting nonce $c8953943 to serv
er...
[2014-02-09 18:23:04] GPU #0: GeForce GTX 780, 163643 khash/s
[2014-02-09 18:23:04] accepted: 1/1 (100.00%), 163643 khash/s (yay!!!)
[2014-02-09 18:23:30] GPU #0: GeForce GTX 780 submitting nonce $c8953943 to serv
er...
[2014-02-09 18:23:30] GPU #0: GeForce GTX 780, 49.83 khash/s
[2014-02-09 18:23:30] GPU #0: GeForce GTX 780, 131065 khash/s
[2014-02-09 18:23:30] accepted: 1/2 (50.00%), 131065 khash/s (booooo)
[2014-02-09 18:23:36] Stratum detected new block
[2014-02-09 18:23:36] GPU #0: GeForce GTX 780, 163622 khash/s
[2014-02-09 18:23:41] Stratum detected new block
[2014-02-09 18:23:41] GPU #0: GeForce GTX 780, 163562 khash/s
[2014-02-09 18:23:46] Stratum detected new block
[2014-02-09 18:23:46] GPU #0: GeForce GTX 780, 163041 khash/s
[2014-02-09 18:23:47] GPU #0: GeForce GTX 780 submitting nonce $055d7c75 to serv
er...
[2014-02-09 18:23:47] GPU #0: GeForce GTX 780, 158360 khash/s
[2014-02-09 18:23:47] accepted: 2/3 (66.67%), 158360 khash/s (yay!!!)
[2014-02-09 18:24:13] GPU #0: GeForce GTX 780 submitting nonce $055d7c75 to serv
er...
[2014-02-09 18:24:13] GPU #0: GeForce GTX 780, 50.15 khash/s
[2014-02-09 18:24:13] GPU #0: GeForce GTX 780, 134809 khash/s
[2014-02-09 18:24:13] accepted: 2/4 (50.00%), 134809 khash/s (booooo)
[2014-02-09 18:24:36] Stratum detected new block
[2014-02-09 18:24:36] GPU #0: GeForce GTX 780, 164243 khash/s
[2014-02-09 18:24:37] GPU #0: GeForce GTX 780 submitting nonce $01175648 to serv
er...
[2014-02-09 18:24:37] GPU #0: GeForce GTX 780, 154699 khash/s
[2014-02-09 18:24:37] accepted: 3/5 (60.00%), 154699 khash/s (yay!!!)
[2014-02-09 18:24:48] GPU #0: GeForce GTX 780 submitting nonce $6f6a2a84 to serv
er...
[2014-02-09 18:24:48] GPU #0: GeForce GTX 780, 162951 khash/s
[2014-02-09 18:24:48] accepted: 4/6 (66.67%), 162951 khash/s (yay!!!)
[2014-02-09 18:24:56] GPU #0: GeForce GTX 780 submitting nonce $bff55d1f to serv
er...
[2014-02-09 18:24:56] GPU #0: GeForce GTX 780, 163280 khash/s
[2014-02-09 18:24:56] accepted: 5/7 (71.43%), 163280 khash/s (yay!!!)
[2014-02-09 18:24:59] GPU #0: GeForce GTX 780 submitting nonce $db27a3bb to serv
er...
[2014-02-09 18:24:59] GPU #0: GeForce GTX 780, 164042 khash/s
[2014-02-09 18:24:59] accepted: 6/8 (75.00%), 164042 khash/s (yay!!!)
[2014-02-09 18:25:03] GPU #0: GeForce GTX 780 submitting nonce $01175648 to serv
er...
[2014-02-09 18:25:03] GPU #0: GeForce GTX 780, 163452 khash/s
[2014-02-09 18:25:03] accepted: 6/9 (66.67%), 163452 khash/s (booooo)
[2014-02-09 18:25:14] GPU #0: GeForce GTX 780 submitting nonce $6f6a2a84 to serv
er...
[2014-02-09 18:25:14] GPU #0: GeForce GTX 780, 163527 khash/s
[2014-02-09 18:25:14] accepted: 6/10 (60.00%), 163527 khash/s (booooo)

No improvements.

cbuchner1

hero member

Activity: 756

Merit: 502

http://bitcoin.stackexchange.com/questions/5048/what-is-the-extranonce

maybe we could increment the timestamp field or something in case of wraparound? clearly this pooler cpuminer wasn't designed with hundred MHash/s speeds in mind...

EDIT: I find that the ntime field is submitted by stratum together with the nonce and it is stored in work->data[17]
so it's safe to increment then, I guess?

Code:

le32enc(&ntime, work->data[17]);
le32enc(&nonce, work->data[19]);
ntimestr = bin2hex((const unsigned char *)(&ntime), 4);
noncestr = bin2hex((const unsigned char *)(&nonce), 4);
xnonce2str = bin2hex(work->xnonce2, work->xnonce2_len);
sprintf(s,
"{\"method\": \"mining.submit\", \"params\": [\"%s\", \"%s\", \"%s\", \"%s\", \"%s\"], \"id\":4}",
rpc_user, work->job_id, xnonce2str, ntimestr, noncestr);
free(ntimestr);
free(noncestr);

ivanlabrie

hero member

Activity: 812

Merit: 1000

You guys never rest... Grin

Will keep an eye on your work, as usual, wish I could be of help. Undecided

bigjme

sr. member

Activity: 350

Merit: 250

there is one called extranonce2 but its always 0

cbuchner1

hero member

Activity: 756

Merit: 502

Could it be that we are seeing a wraparound of the nonce search space? As our cards have hundred megahash/s search power, it may be that the nonce word (32 bit) just wraps around before a new block is found. It would also explain why faster cards report MORE duplicates. The word wraps around quicker.

I think there is a field called "extranonce" which may have be be incremented in this case.

Christian

bigjme

sr. member

Activity: 350

Merit: 250

new githib is much worse christian
9/21

duplicates came in this rate;

4
6
1
1

another saying job not found

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: orrett3 on February 09, 2014, 12:15:47 PM

It might be that you are a wizard... BTW check your inbox in a minute.

well I was too quick. It's at 96.0% now. About 1.5 % better than before

ManIkWeet

full member

Activity: 182

Merit: 100

In case it's still of interest:

Code:

[2014-02-09 18:14:44] GPU #0: GeForce GTX 780 submitting nonce $26b2cdef to serv
er...
[2014-02-09 18:14:44] GPU #0: GeForce GTX 780, 163641 khash/s
[2014-02-09 18:14:44] accepted: 6/7 (85.71%), 163641 khash/s (yay!!!)
[2014-02-09 18:15:10] GPU #0: GeForce GTX 780 submitting nonce $26b2cdef to serv
er...
[2014-02-09 18:15:10] GPU #0: GeForce GTX 780, 50.10 khash/s
[2014-02-09 18:15:10] GPU #0: GeForce GTX 780, 134809 khash/s
[2014-02-09 18:15:10] accepted: 6/8 (75.00%), 134809 khash/s (booooo)
[2014-02-09 18:15:37] GPU #0: GeForce GTX 780 submitting nonce $26b2cdef to serv
er...
[2014-02-09 18:15:37] GPU #0: GeForce GTX 780, 164198 khash/s
[2014-02-09 18:15:37] accepted: 6/9 (66.67%), 164198 khash/s (booooo)
[2014-02-09 18:15:39] Stratum detected new block
[2014-02-09 18:15:39] GPU #0: GeForce GTX 780, 163114 khash/s

This is before Github fix.

orrett3

newbie

Activity: 33

Merit: 0

Quote

On a different note the fix I posted above has brought my accept rate to 100%.
Explain this. ;-)

It might be that you are a wizard... BTW check your inbox in a minute.

bigjme

sr. member

Activity: 350

Merit: 250

compiling now, will update this post as i get news

Edit 1
"job b14 not found" gave me a reject
4 duplicate shares after each other

Edit 2
6 duplicate shares after each other
now at 8/18

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: bigjme on February 09, 2014, 12:07:26 PM

right then, just noticed something, its submitting codes it did much earlier

like i just had 4 rejects in a row, reject 1 and 4 had the same nonce
and i have a feeling 2 and 3 many have had the same nonce as earlier submissions

Okay try the fix that is now in github. I think it will make you happy.

bigjme

sr. member

Activity: 350

Merit: 250

Quote from: cbuchner1 on February 09, 2014, 12:05:40 PM

Quote from: bigjme on February 09, 2014, 12:03:28 PM

every nonce value is different on all the rejects....
i should know, 6 one after another :-(

Yeah WTF then are the pool operators cheating?

right then, just noticed something, its submitting codes it did much earlier

like i just had 4 rejects in a row, reject 1 and 4 had the same nonce
and i have a feeling 2 and 3 many have had the same nonce as earlier submissions

i wonder if there is a way to get cudaminer to save all the nonce values and if they were a yay or boo in a file to look up whats being sent?

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: Stanr010 on February 09, 2014, 12:05:50 PM

Selling for under 0.002 feels like just throwing it away but I'd hate to hold onto coins if it's still go down and never come back up.

if you believe in a spike then by all means buy with some of the previous earnings.

Stanr010

member

Activity: 70

Merit: 10

I'm beginning to wonder if we should start holding onto MAX for the price spike when it hits cryptsy

Selling for under 0.002 feels like just throwing it away but I'd hate to hold onto coins if it's still go down and never come back up.

Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 827. (Read 3426976 times)