[ANN] ccminer 2.3 - opensource - GPL (tpruvot) - page 164.

tbearhere

legendary

Activity: 3164

Merit: 1003

Quote from: royalpro on December 18, 2014, 12:34:01 PM

So you have to set intensity for each card.

You can and should if your cards are different, but if you just set it once it should use that for all cards.
[/quote]
ok thanks. Did you try sending 3 cards doing one coin and the other 3 doing another? I get one instance looking for api saying api not found will try again in 30sec.

royalpro

newbie

Activity: 9

Merit: 0

Quote from: tbearhere on December 18, 2014, 12:31:59 PM

So you have to set intensity for each card.

You can and should if your cards are different, but if you just set it once it should use that for all cards.

tbearhere

legendary

Activity: 3164

Merit: 1003

Quote from: scryptr on December 18, 2014, 12:25:39 PM

CCMINER INTENSITY-

I have an Ubuntu 14.04.1 build, x64, and have had success with x13 and an intensity of 20.5. I have since tried various intensity settings with x11 and Lyra2RE algorithms. Algo x11 maxes out with "-i 20.5"' but the improvement is very small compared to the default intensity of 19, only about +25kh/s for the complete rig (6 x 750ti). Lyra2RE does not improve with an increased intensity, or a lowered intensity. With Lyra2RE, an intensity setting of "-i 21" or greater resulted in HUGE kh/s values and no accepted shares. Algos x11 and x13 generated "out of memory" errors at settings of "-i 20.7".

I have used the setting of "./ccminer -a x13 -d 0,1,2,3,4,5 -i 20.5,20.5,20.5,20.5,20.5,20.5 -o stratum+tcp://URL" in my launch string.

So, overall, x13 mining with "-i 20.5" gives +125kh/s on my 6-card rig (14,300kh/s), x11 mining with "-i 20.5" improves by 25kh/s on the same rig (17,900kh/s). For Lyra2RE, the intensity settings had little or no effect on accepted hash rate.
--scryptr

So you have to set intensity for each card.

scryptr

legendary

Activity: 1797

Merit: 1028

CCMINER INTENSITY-

I have an Ubuntu 14.04.1 build, x64, and have had success with x13 and an intensity of 20.5. I have since tried various intensity settings with x11 and Lyra2RE algorithms. Algo x11 maxes out with "-i 20.5"' but the improvement is very small compared to the default intensity of 19, only about +25kh/s for the complete rig (6 x 750ti). Lyra2RE does not improve with an increased intensity, or a lowered intensity. With Lyra2RE, an intensity setting of "-i 21" or greater resulted in HUGE kh/s values and no accepted shares. Algos x11 and x13 generated "out of memory" errors at settings of "-i 20.7".

I have used the setting of "./ccminer -a x13 -d 0,1,2,3,4,5 -i 20.5,20.5,20.5,20.5,20.5,20.5 -o stratum+tcp://URL" in my launch string.

So, overall, x13 mining with "-i 20.5" gives +125kh/s on my 6-card rig (14,300kh/s), x11 mining with "-i 20.5" improves by 25kh/s on the same rig (17,900kh/s). For Lyra2RE, the intensity settings had little or no effect on accepted hash rate.
--scryptr

tbearhere

legendary

Activity: 3164

Merit: 1003

Quote from: Epsylon3 on December 18, 2014, 09:21:28 AM

yes its for all algos... default is not the max to let a margin of work and reduce ui slow down effect

I tried the intensity 20.5 and cmd window showed way higher then normal hash then closed, error . Then I set it to 19 and showed very low hash. I had to reboot to get back to normal without the intensity. Not on x11 algo. I compiled it with x64, it didn't have x86 available.

Epsylon3

legendary

Activity: 1484

Merit: 1082

ccminer/cpuminer developer

yes its for all algos... default is not the max to let a margin of work and reduce ui slow down effect

scryptr

legendary

Activity: 1797

Merit: 1028

CCMINER INTENSITY -

I made an experiment with the new intensity "-i" command-line switch today. I am using version 1.5.1 final, mining x13, and a 6-card 750ti rig. After first testing "-i 19.5, 19.5,..." for the six cards, there was improvement. At program start, there was a notice of the amount of cuda threads being added to the default value.

I then moved to "-i 20,20,...", and am currently running at 14,275kh/s for the entire 6 card rig. This is an improvement of 100kh/s for the whole rig, which ran at 14,175kh/s at the default setting. The individual cards now run from 2365kh/s to 2410kh/s, and both the individual cards and the overall total kh/s values are very consistent. The error rate hovers at 1%. And, at program start, there was a notice of the total amount of cuda threads being processed.

My suspicion is that I am at or near the limit for the setting, at least for x13. Does the intensity setting work for all ccminer algorithms, or is it effective only on a particular few? Thank you for your hard work. --scryptr

EDIT: Further testing revealed that a setting of 20.7 red-lined with an "out of memory" error, and that a setting of 20.5 hit right on the 14,300kh/s mark for the 6-card rig. Again, thanks! --scryptr

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: Epsylon3 on December 15, 2014, 02:12:33 AM

djm34: did you read my commit message :

https://github.com/tpruvot/ccminer/commit/c5b349e079876be6f4f24c82f713cddd36b736aa

I might have read the title only, but still you should have waited a bit...

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: skunk on December 14, 2014, 01:40:43 PM

Quote from: djm34 on December 14, 2014, 08:44:02 AM

Quote from: skunk on December 14, 2014, 06:56:27 AM

hi again,
fyi, lyra2 doesn't seem to work on sm30 (at least not on nicehash and yaamp), the miner says it's hashing at 471khs but no shares are submitted and pool's PING api's output stays at 0.

no it doesn't... epsylon3 released a preliminary version (without even asking... Angry

and forgetting somehow to site the author... who is still a bit pissed about it Grin

yeah well...) which didn't have yet the 3.0 kernel as it was for a preliminary test.

Support for the kernel 3.0 can be found here: https://github.com/djm34/ccminer
(the current version has adjustable throughput, you need --throughput 4 to get to metallicjames release, the interest is that you can set it a bit higher for the big guns...)
If you don't want to go through the trouble of setting it up: metallicjames release should also have these new kernel.

cheers,
The pissed off author. Grin

thank you djm34, unfortunately even your miner doesn't work as expected (just one share was sent in about half hour, tried with and without "-F 4" switch")...
as metallicjames's miner, after fixing many compilation errors, i finally got stuck on some "undefined reference" errors i'm not able to fix...
i'll forget about lyra2 for now...

you have to use "--throughput 4" and "--diff 128"
(the difficulty adjustment is required for the moment, but I was told that pool will update, that's why I haven't put it directly into the code yet...)

Epsylon3

legendary

Activity: 1484

Merit: 1082

ccminer/cpuminer developer

djm34: did you read my commit message :

https://github.com/tpruvot/ccminer/commit/c5b349e079876be6f4f24c82f713cddd36b736aa

skunk

sr. member

Activity: 329

Merit: 250

Quote from: djm34 on December 14, 2014, 08:44:02 AM

Quote from: skunk on December 14, 2014, 06:56:27 AM

hi again,
fyi, lyra2 doesn't seem to work on sm30 (at least not on nicehash and yaamp), the miner says it's hashing at 471khs but no shares are submitted and pool's PING api's output stays at 0.

no it doesn't... epsylon3 released a preliminary version (without even asking... Angry

and forgetting somehow to site the author... who is still a bit pissed about it Grin

yeah well...) which didn't have yet the 3.0 kernel as it was for a preliminary test.

Support for the kernel 3.0 can be found here: https://github.com/djm34/ccminer
(the current version has adjustable throughput, you need --throughput 4 to get to metallicjames release, the interest is that you can set it a bit higher for the big guns...)
If you don't want to go through the trouble of setting it up: metallicjames release should also have these new kernel.

cheers,
The pissed off author. Grin

thank you djm34, unfortunately even your miner doesn't work as expected (just one share was sent in about half hour, tried with and without "-F 4" switch")...
as metallicjames's miner, after fixing many compilation errors, i finally got stuck on some "undefined reference" errors i'm not able to fix...
i'll forget about lyra2 for now...

royalpro

newbie

Activity: 9

Merit: 0

Quote from: djm34 on December 14, 2014, 08:44:02 AM

Quote from: skunk on December 14, 2014, 06:56:27 AM

hi again,
fyi, lyra2 doesn't seem to work on sm30 (at least not on nicehash and yaamp), the miner says it's hashing at 471khs but no shares are submitted and pool's PING api's output stays at 0.

no it doesn't... epsylon3 released a preliminary version (without even asking... Angry

and forgetting somehow to site the author... who is still a bit pissed about it Grin

yeah well...) which didn't have yet the 3.0 kernel as it was for a preliminary test.

Support for the kernel 3.0 can be found here: https://github.com/djm34/ccminer
(the current version has adjustable throughput, you need --throughput 4 to get to metallicjames release, the interest is that you can set it a bit higher for the big guns...)
If you don't want to go through the trouble of setting it up: metallicjames release should also have these new kernel.

cheers,
The pissed off author. Grin

Excuse me pissed off author, but when I try to compile yours on ubuntu 14.04 and cuda 6.5 and get this error

Code:

nvcc -g -O2 -I . -Xptxas "-abi=no -v" -gencode=arch=compute_30,code=\"sm_30,compute_30\" -gencode=arch=compute_35,code=\"sm_35,compute_35\" -gencode=arch=compute_50,code=\"sm_50,compute_50\" -gencode=arch=compute_52,code=\"sm_52,compute_52\" --maxrregcount=80 --ptxas-options=-v  -o heavy/heavy.o -c heavy/heavy.cu
nvcc fatal   : Unsupported gpu architecture 'compute_52'
make[2]: *** [heavy/heavy.o] Error 1
make[2]: Leaving directory `/home/miner/mining/cclyra2'
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory `/home/miner/mining/cclyra2'
make: *** [all] Error 2

Would you be kind enough to help me figure it out?

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: skunk on December 14, 2014, 06:56:27 AM

hi again,
fyi, lyra2 doesn't seem to work on sm30 (at least not on nicehash and yaamp), the miner says it's hashing at 471khs but no shares are submitted and pool's PING api's output stays at 0.

no it doesn't... epsylon3 released a preliminary version (without even asking... Angry

and forgetting somehow to site the author... who is still a bit pissed about it Grin

yeah well...) which didn't have yet the 3.0 kernel as it was for a preliminary test.

Support for the kernel 3.0 can be found here: https://github.com/djm34/ccminer
(the current version has adjustable throughput, you need --throughput 4 to get to metallicjames release, the interest is that you can set it a bit higher for the big guns...)
If you don't want to go through the trouble of setting it up: metallicjames release should also have these new kernel.

cheers,
The pissed off author. Grin

skunk

sr. member

Activity: 329

Merit: 250

hi again,
fyi, lyra2 doesn't seem to work on sm30 (at least not on nicehash and yaamp), the miner says it's hashing at 471khs but no shares are submitted and pool's PING api's output stays at 0.

Epsylon3

legendary

Activity: 1484

Merit: 1082

ccminer/cpuminer developer

Quote from: sp_ on December 07, 2014, 10:48:44 AM

Latest git version needs more than 4GB mem to compile in visual studio.

Visual studio is a pain Wink

If i could find a way to compile cuda stuff in mingw...

Epsylon3

legendary

Activity: 1484

Merit: 1082

ccminer/cpuminer developer

The 1.5.1 should be more stable, i found some issues in solo mining

sp_

legendary

Activity: 2926

Merit: 1087

Team Black developer

Latest git version needs more than 4GB mem to compile in visual studio.

xfelinho

member

Activity: 68

Merit: 10

i really like this ccminer but it always starts crashing after some hours...i use 6x 750 ti with qubit algo and without overclock
i never had this problems with djms ccminer

jjjordan

sr. member

Activity: 271

Merit: 251

Quote from: sp_ on December 03, 2014, 05:50:55 AM

Quote from: Epsylon3 on December 03, 2014, 05:10:37 AM

yep, im not so confident about one pool graph... But i need to compare also...
But its not related to the extranonce feature... which is not even used by most pools Wink

its only one request on start in this case.

I have thested both the 1.5.0 and my modded version and they seem to report the correct hash on other pools like wafflepool, yaamp, and nicehash, so this seems to be a pool issue, but it might be a bug in ccminer as well.

I second that - all multicoin/profit switching pools I've tried work fine, but most the single coin ones report much lower hash speed than what's reported by ccminer. At least that's the pattern I've noticed. version8 is the last one that works fine I think... All the newer report higher hashrates but pools don't "like" them.

Bombadil

hero member

Activity: 644

Merit: 500

Quote from: djm34 on December 03, 2014, 07:39:30 AM

what is the current status of the api+monitoring, thinking to use it but I don't want to break anything in my own release ?

API seems fully implemented, all that I'll need, and probably everyone else too, is in there. But my monitoring app is going slowly Tongue

(It's just noobish me)
But I work on it every day, so I expect a basic tool for next week (don't take my word for it Cheesy

)

Topic: [ANN] ccminer 2.3 - opensource - GPL (tpruvot) - page 164. (Read 500200 times)