[ANN][GAP] Gapcoin - Prime Gap Search - New Math Algo - CPU / GPU - Zero Premine - page 96.

antonio8

legendary

Activity: 1400

Merit: 1000

I'll admit that I am not always the brightest and this has been asked a ton.

I created 2 different folders one 280x (device 0) and one 270x (device 1). I have in 280x folder --use-gpu 0. I have in 270x folder --use-gpu 1.

The 280x runs fine. I start the bat in 270x (gpu 1) and the bat runs gpu 0, the 280x.

Can someone enlighten me please.

qwep

legendary

Activity: 1008

Merit: 1000

Quote from: dcct on November 07, 2014, 08:40:02 PM

Quote from: Palmdetroit on November 07, 2014, 08:16:37 PM

Tested GPU miner 7950s, worked great

How many pps?

240k

nonce-pool

full member

Activity: 149

Merit: 100

Until stratum works for windows miners we have opened another longpoll port on 4205 for gpu miners.

It has a higher difficulty for better pps and less stales.

Palmdetroit

legendary

Activity: 910

Merit: 1000

PHS 50% PoS - Stop mining start minting

Quote from: dcct on November 07, 2014, 08:40:02 PM

Quote from: Palmdetroit on November 07, 2014, 08:16:37 PM

Tested GPU miner 7950s, worked great

How many pps?

~150k-200k each, without OC , Windows, I would imagine that can be tweaked higher with current miners and improvements, also on a relative slow CPU - 8150

Miner seems CPU heavy

Didn't try to tweak it much.

dcct

sr. member

Activity: 280

Merit: 250

Quote from: Palmdetroit on November 07, 2014, 08:16:37 PM

Tested GPU miner 7950s, worked great

How many pps?

Palmdetroit

legendary

Activity: 910

Merit: 1000

PHS 50% PoS - Stop mining start minting

Good job on the gpuminer

Tested GPU miner 7950s, worked great

tried old ubuntu box with a 6870 and:

Code:

[2014-11-07 20:14:01] pps: -2147483648 / -2147483648 10g/h -2147483648 / -21474 83648 15g/h -2147483648 / -2147483648
[2014-11-07 20:14:11] pps: 0 / 0 10g/h 0 / 0 15g/h 0 / 0

Seemed to always be 0 and never submit share, issue with legacy gpu?

Thanks will test some other cards this weekend if able.

q327K091

legendary

Activity: 1792

Merit: 1010

Quote from: djm34 on November 07, 2014, 12:55:38 PM

Quote from: q327K091 on November 07, 2014, 07:07:35 AM

ok got passed OpenCL compiler error, now this

( replaced

// uint4 q[2] = {0, 0};
uint4 q[2] = {0, 0, 0, 0};
)

yep that's experimental Grin

initializing uint4 as uint2 is indeed experimental
(however it isn't obvious that opencl and cuda use the same definition for these structures...)

so NVIDIA users are shut out from this software for the time being ? Cool

drumingspz

sr. member

Activity: 392

Merit: 250

Quote from: shifty252 on November 07, 2014, 04:27:09 PM

Quote from: j0nn9 on November 07, 2014, 04:12:20 PM

Quote from: shifty252 on November 07, 2014, 04:06:16 PM

hi guys, can you please post a sample gapcoin.conf please?

also, has anyone with a radeon 270x tried the gpu miner? if yes, care to share your results?

Code:

rpcuser=user
rpcpassword=password
server=1
addnode=212.227.143.155
addnode=104.131.69.82
addnode=95.215.44.9
addnode=155.254.36.179
addnode=199.127.226.192

thanks!

i get the following when running gpu miner with stratum(both 32&64):

Code:

E:\gap\windows\64\gpu>gapminer-gpu.exe -o mine3.gap.nonce-pool.com -p 3385 -u user.1 -x password -t 1 --stratum --use-gpu
[2014-11-07 23:24:48] Found platform[0] name = AMD Accelerated Parallel Processi
ng
[2014-11-07 23:24:48] Found 1 device(s)
[2014-11-07 23:24:48] Loaded kernel binary size = 872620 bytes
[2014-11-07 23:24:56] Using GPU 0 [Pitcairn]: which has 20 CUs
[2014-11-07 23:24:56] Requesting work failedsend: The file handle supplied is no
t valid
[2014-11-07 23:24:56] Error receiving message form server: read_until: The file
handle supplied is not valid

getwork mines ok at ~141000 pps

I see a similar issues - Win7x64 64 bit miner (stratum connection)

Code:

D:\Downloads\Mining\GAPCoin\GPU Miner\windows\64\gpu>gapminer-gpu -o mine3.gap.n
once-pool.com -p 3385 -u username -x password -t 1 -g --stratum
[2014-11-07 15:53:02] Found platform[0] name = AMD Accelerated Parallel Processing
[2014-11-07 15:53:02] Found 1 device(s)
[2014-11-07 15:53:02] Loaded kernel binary size = 872620 bytes
[2014-11-07 15:53:02] Using GPU 0 [Tahiti]: which has 32 CUs
[2014-11-07 15:53:02] Requesting work failedsend: The file handle supplied is not valid
[2014-11-07 15:53:02] Error receiving message form server: read_until: The file handle supplied is not valid

Same system - longpoll connection. Miner looks like its beginning to work, then dies...

Code:

D:\Downloads\Mining\GAPCoin\GPU Miner\windows\64\gpu>gapminer-gpu -o mine3.gap.n
once-pool.com -p 4200 -u username -x password -t 1 -g
[2014-11-07 15:54:41] Found platform[0] name = AMD Accelerated Parallel Processing
[2014-11-07 15:54:41] Found 1 device(s)
[2014-11-07 15:54:41] Loaded kernel binary size = 872620 bytes
[2014-11-07 15:54:41] Using GPU 0 [Tahiti]: which has 32 CUs
[2014-11-07 15:54:42] Server supports longpoll
[2014-11-07 15:54:42] Got new target: 13.0000000 @ 22.4626094

I've tried increasing my CPU threads, as I only have 1 GPU in this machine, but that hasn't helped. The system has 12GB of RAM, so I don't think that's the issue either.

qwep

legendary

Activity: 1008

Merit: 1000

Dev Is it possible to reduce the load on the processor Angry

bsunau7

member

Activity: 114

Merit: 10

Quote from: Supercomputing on November 07, 2014, 05:44:48 PM

@bsunau7

Thanks for the info and I will implement a small sieve for validation and use a single exponentiation test at each end point; it should be at least 100x faster and less than 1/(2¹²⁸) probability of it being a larger gap.

There was talk on GMP mailing lists about speeding up gmp_nextprime in a very similar manner, sieve the start/end gap before running expensive MR tests. As the gap has been pre-sieved you might just be able to MR test the non-composite in the mining sieve and get the same result.

I personally didn't bother with the PoW validation code as for larger gaps it just isn't called enough to matter.

Also you can tweak the pprocessor->process() to terminate the sieve early which will cure the stales and get you to the next "block" a few seconds faster at the risk of having a slightly wrong difficulty (in effect a non issue).

PS. Just pushed testnet to >8 might slow down your stale shares some (I am only getting only 75-90% stale at this difficulty).

Regards,

--
bsunau7

Supercomputing

sr. member

Activity: 278

Merit: 250

Quote from: bsunau7 on November 07, 2014, 04:56:24 PM

This seems to be normal for an optimized miner.

The miner does not stop mining the sieve until the original one is verified by a slow mpz_nextprime() call. When your miner is significantly faster than the PoW function you tend to find lots of shares before the PoW can stop the sieve from mining.

Regards,

--
bsunau7

@bsunau7

Thanks for the info and I will implement a small sieve for validation and use a single exponentiation test at each end point; it should be at least 100x faster and less than 1/(2¹²⁸) probability of it being a larger gap.

qwep

legendary

Activity: 1008

Merit: 1000

Quote from: PorcoRosso on November 07, 2014, 04:21:41 PM

CPU load 100% in GPU miner =(

yep

ZXSpectrum

newbie

Activity: 29

Merit: 0

Quote from: ZXSpectrum on November 07, 2014, 04:54:28 PM

Quote from: j0nn9 on November 07, 2014, 03:06:59 PM

Hi there,

II know I said next week, but i managed to build some experimental gpu miner binaries Grin

.
I created a new release which includes the stratum changes from nonce-pool,
and some bug fixes regarding the device selection.

But still, the binaries and source code are very experimental!

Works, pointed at the nonce-pool for now - thanks. I am getting some stale shares however - how can I troubleshoot that?

Sorry for another post, card 0 is a 280x and only using around 80-90% of gpu... card 1 is a 290 and sits at 100% all the time but sometimes goes to 0 then back to 100, I guess this is a code thing... is there anything I Can provide to help improve?

go6ooo1212

legendary

Activity: 1512

Merit: 1000

quarkchain.io

Quote from: alibabacool on November 07, 2014, 04:54:02 PM

Quote from: go6ooo1212 on November 07, 2014, 04:52:33 PM

Quote from: alibabacool on November 07, 2014, 04:43:58 PM

Quote from: go6ooo1212 on November 07, 2014, 04:39:53 PM

Cant actually involve more than 1 card a rig , even with this option :
--gpu-dev 0,1,2,3,4

One miner instance for each gpu

Yes Ive tried so ,and suppose to work but the speeds are different for each card ...

Which cpu do you have on a farm?

Im with celeron, sadly...

bsunau7

member

Activity: 114

Merit: 10

Quote from: Supercomputing on November 07, 2014, 04:44:20 PM

@j0nn9

One less item on my to-do list for this weekend now that you've released a GPU based miner. Grin

Sorry if I missed something, but how do you configure the miner to run on testnet? I received too many stale submissions followed by the miner hanging. Please see the log file below:

CPU: AMD FX-8350
GPU: AMD R9 280X

./gapminer -o 10.0.0.2 -p 9876 -u u_rpcuser_u -x p_rpcpass_p -t 8 -f 23 -i 800000 -s 8388608 -g -d 0
[2014-11-07 15:25:05] Found platform[0] name = NVIDIA CUDA
[2014-11-07 15:25:05] Found platform[1] name = AMD Accelerated Parallel Processing
[2014-11-07 15:25:05] Found 1 device(s)
[2014-11-07 15:25:05] Compiling ...
[2014-11-07 15:25:05] Source: 205055 bytes
[2014-11-07 15:25:08] Compiled kernel binary size = 903840 bytes
[2014-11-07 15:25:08] Loaded kernel binary size = 903840 bytes
[2014-11-07 15:25:08] Using GPU 0 [Tahiti]: which has 32 CUs
[2014-11-07 15:25:09] Got new target: 1.0000000 @ 1.0000000
[2014-11-07 15:25:15] Found Share: 2.4725781 => accepted
[2014-11-07 15:25:15] Found Share: 1.3788212 => stale!
[2014-11-07 15:25:15] Found Share: 1.4127911 => stale!
[2014-11-07 15:25:15] Found Share: 1.2721381 => stale!
[2014-11-07 15:25:16] Found Share: 1.4317729 => stale!
[2014-11-07 15:25:16] Found Share: 3.6779116 => stale!
[2014-11-07 15:25:16] Found Share: 1.4460171 => stale!
[2014-11-07 15:25:17] Found Share: 1.2714187 => stale!
[2014-11-07 15:25:17] Found Share: 1.4480290 => stale!
[2014-11-07 15:25:17] Found Share: 2.7009248 => stale!
[2014-11-07 15:25:17] Found Share: 2.0624349 => stale!
[2014-11-07 15:25:18] Found Share: 7.0916937 => stale!
[2014-11-07 15:25:18] Found Share: 2.2394908 => stale!
[2014-11-07 15:25:18] Found Share: 1.8181628 => stale!
[2014-11-07 15:25:19] Found Share: 2.4725407 => stale!
[2014-11-07 15:25:19] Found Share: 1.2781374 => stale!
[2014-11-07 15:25:19] Found Share: 1.5778022 => stale!
[2014-11-07 15:25:19] Got new target: 1.0000000 @ 1.0000000
[2014-11-07 15:26:13] shutdown..
[2014-11-07 15:26:19] shutdown..
[2014-11-07 15:26:20] I'm on it, just wait! (press again to kill)

This seems to be normal for an optimized miner.

The miner does not stop mining the sieve until the original one is verified by a slow mpz_nextprime() call. When your miner is significantly faster than the PoW function you tend to find lots of shares before the PoW can stop the sieve from mining.

Regards,

--
bsunau7

ZXSpectrum

newbie

Activity: 29

Merit: 0

Quote from: j0nn9 on November 07, 2014, 03:06:59 PM

Hi there,

II know I said next week, but i managed to build some experimental gpu miner binaries Grin

.
I created a new release which includes the stratum changes from nonce-pool,
and some bug fixes regarding the device selection.

But still, the binaries and source code are very experimental!

Works, pointed at the nonce-pool for now - thanks. I am getting some stale shares however - how can I troubleshoot that?

alibabacool

full member

Activity: 139

Merit: 100

Quote from: go6ooo1212 on November 07, 2014, 04:52:33 PM

Quote from: alibabacool on November 07, 2014, 04:43:58 PM

Quote from: go6ooo1212 on November 07, 2014, 04:39:53 PM

Cant actually involve more than 1 card a rig , even with this option :
--gpu-dev 0,1,2,3,4

One miner instance for each gpu

Yes Ive tried so ,and suppose to work but the speeds are different for each card ...

Which cpu do you have on a farm?

go6ooo1212

legendary

Activity: 1512

Merit: 1000

quarkchain.io

Quote from: alibabacool on November 07, 2014, 04:43:58 PM

Quote from: go6ooo1212 on November 07, 2014, 04:39:53 PM

Cant actually involve more than 1 card a rig , even with this option :
--gpu-dev 0,1,2,3,4

One miner instance for each gpu

Yes Ive tried so ,and suppose to work but the speeds are different for each card ...

Supercomputing

sr. member

Activity: 278

Merit: 250

@j0nn9

One less item on my to-do list for this weekend now that you've released a GPU based miner. Grin

Sorry if I missed something, but how do you configure the miner to run on testnet? I received too many stale submissions followed by the miner hanging. Please see the log file below:

CPU: AMD FX-8350
GPU: AMD R9 280X

./gapminer -o 10.0.0.2 -p 9876 -u u_rpcuser_u -x p_rpcpass_p -t 8 -f 23 -i 800000 -s 8388608 -g -d 0
[2014-11-07 15:25:05] Found platform[0] name = NVIDIA CUDA
[2014-11-07 15:25:05] Found platform[1] name = AMD Accelerated Parallel Processing
[2014-11-07 15:25:05] Found 1 device(s)
[2014-11-07 15:25:05] Compiling ...
[2014-11-07 15:25:05] Source: 205055 bytes
[2014-11-07 15:25:08] Compiled kernel binary size = 903840 bytes
[2014-11-07 15:25:08] Loaded kernel binary size = 903840 bytes
[2014-11-07 15:25:08] Using GPU 0 [Tahiti]: which has 32 CUs
[2014-11-07 15:25:09] Got new target: 1.0000000 @ 1.0000000
[2014-11-07 15:25:15] Found Share: 2.4725781 => accepted
[2014-11-07 15:25:15] Found Share: 1.3788212 => stale!
[2014-11-07 15:25:15] Found Share: 1.4127911 => stale!
[2014-11-07 15:25:15] Found Share: 1.2721381 => stale!
[2014-11-07 15:25:16] Found Share: 1.4317729 => stale!
[2014-11-07 15:25:16] Found Share: 3.6779116 => stale!
[2014-11-07 15:25:16] Found Share: 1.4460171 => stale!
[2014-11-07 15:25:17] Found Share: 1.2714187 => stale!
[2014-11-07 15:25:17] Found Share: 1.4480290 => stale!
[2014-11-07 15:25:17] Found Share: 2.7009248 => stale!
[2014-11-07 15:25:17] Found Share: 2.0624349 => stale!
[2014-11-07 15:25:18] Found Share: 7.0916937 => stale!
[2014-11-07 15:25:18] Found Share: 2.2394908 => stale!
[2014-11-07 15:25:18] Found Share: 1.8181628 => stale!
[2014-11-07 15:25:19] Found Share: 2.4725407 => stale!
[2014-11-07 15:25:19] Found Share: 1.2781374 => stale!
[2014-11-07 15:25:19] Found Share: 1.5778022 => stale!
[2014-11-07 15:25:19] Got new target: 1.0000000 @ 1.0000000
[2014-11-07 15:26:13] shutdown..
[2014-11-07 15:26:19] shutdown..
[2014-11-07 15:26:20] I'm on it, just wait! (press again to kill)

alibabacool

full member

Activity: 139

Merit: 100

Quote from: go6ooo1212 on November 07, 2014, 04:39:53 PM

Cant actually involve more than 1 card a rig , even with this option :
--gpu-dev 0,1,2,3,4

One miner instance for each gpu

Topic: [ANN][GAP] Gapcoin - Prime Gap Search - New Math Algo - CPU / GPU - Zero Premine - page 96. (Read 287695 times)