[ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX]

[cbuchner1]

--algo=scrypt-jane:MRC  !!!

I do not think he is using cudaminer...

ok, so I guess he is in the wrong thread here...

This thread discusses the use of the cudaminer software using the nVidia CUDA API.

I have no experience with OpenCL based mining software - and frankly: such software sucks on nVidia anyway.

[1713913308]

1713913308

[1713913308]

1713913308

[1713913308]

1713913308

[elvitto]

Use -i 1, that will still show 100%, but Desktop is much more responsible. There is not really an "official" way to run it at 60%. Maybe you could lower the found KXXxXX number, that would also reduce the speed, but might lower the GPU usage.

patoberli hello, I Realize the change that you suggested and now it's going very well!

Many thanks for your help!

[bla2501]

I tried to include the --algo=scrypt-jane:MRC in my .bat file, but then the miner immediately closes, my configurations looks like this:
cudaminer.exe -H 1 -i 0 -l auto -C 1 -o stratum+tcp://pool url.port -O username.worker:pass --a scrypt-jane:MRC
Can you please tell me exactly what I have to do, or can you please paste me the what to include in .bat file and how should my .bat file look like.

Thank you for your help

P.S. I have started mining a couple days ago and I am still such a noob, so I would like to apologize for my questions that are probably extremely stupid. i am also sorry for the confusion i made for mentioning the OpenCL based mining software.

[irv]

Could someone offer some advice please?

I've got my hands on a ex-company Nvidia GRID K1, but I'm only managing to get 17khash/s per GPU and this thing has 4x Keplers in it. cudaminer picks the card up as a Grid K1/Kepler so it is identifying it on autotune.

No matter what I do, it won't go any faster.  The command line I'm using is:-

-d 0 -i 1 -l auto -C 1 -m 1

And then running it 4x times (changing the d switch) so I have 1 instance running per GPU.  I've tried running multiple instances per GPU but that just makes it slower.

I'd expect a card like this to pull over 200khash/per GPU or am I being way too optimistic?

Thanks

[cbuchner1]

not having any luck on this MRC pool

Code:

./cudaminer -H 2 --algo=scrypt-jane:MRC -C 1 -l Y14x16 -b 256 -i 1 -m 1 -o stratum+tcp://mrc.easy-mining.net:3361 -O ...secret...

[2014-01-27 16:04:47] Given scrypt-jane parameters: MRC
[2014-01-27 16:04:47] Nfactor is 7 (N=256)!
[2014-01-27 16:04:47] GPU #0: GeForce GTX 660 Ti with compute capability 3.0
[2014-01-27 16:04:47] GPU #0: interactive: 1, tex-cache: 1D, single-alloc: 1
[2014-01-27 16:04:47] GPU #0: 32 hashes / 1.0 MB per warp.
[2014-01-27 16:04:47] GPU #0: using launch configuration Y14x16
[2014-01-27 16:04:47] GPU #0: GeForce GTX 660 Ti, 410.84 khash/s
[2014-01-27 16:04:48] GPU #0: GeForce GTX 660 Ti, 864.91 khash/s
[2014-01-27 16:04:48] accepted: 0/1 (0.00%), 864.91 khash/s (booooo)
[2014-01-27 16:04:49] GPU #0: GeForce GTX 660 Ti, 921.06 khash/s
[2014-01-27 16:04:49] accepted: 0/2 (0.00%), 921.06 khash/s (booooo)
[2014-01-27 16:04:53] GPU #0: GeForce GTX 660 Ti, 924.85 khash/s
[2014-01-27 16:04:53] accepted: 0/3 (0.00%), 924.85 khash/s (booooo)
[2014-01-27 16:04:54] GPU #0: GeForce GTX 660 Ti, 918.26 khash/s
[2014-01-27 16:04:54] accepted: 0/4 (0.00%), 918.26 khash/s (booooo)
[2014-01-27 16:04:55] GPU #0: GeForce GTX 660 Ti, 919.02 khash/s
[2014-01-27 16:04:55] accepted: 0/5 (0.00%), 919.02 khash/s (booooo)
[2014-01-27 16:04:57] GPU #0: GeForce GTX 660 Ti, 918.61 khash/s
[2014-01-27 16:04:57] accepted: 0/6 (0.00%), 918.61 khash/s (booooo)

I am getting boo'ed at all the time. Is the pool broken, or is my code broken?

EDIT: time to solo-mine! According to postings in the MRC thread, most pools are not accepting shares right now.

[cbuchner1]

I tried to include the --algo=scrypt-jane:MRC in my .bat file, but then the miner immediately closes, my configurations looks like this:
cudaminer.exe -H 1 -i 0 -l auto -C 1 -o stratum+tcp://pool url.port -O username.worker:pass --a scrypt-jane:MRC
Can you please tell me exactly what I have to do, or can you please paste me the what to include in .bat file and how should my .bat file look like.

Thank you for your help

either --algo or -a  I think, but not --a

add a pause statement to the BAT file so it doesn't close immediately.

Also you need some very recent code for MRC mining. Like...compiled within the last 4 hours.

Christian

[cbuchner1]

Could someone offer some advice please?

I've got my hands on a ex-company Nvidia GRID K1, but I'm only managing to get 17khash/s per GPU and this thing has 4x Keplers in it. cudaminer picks the card up as a Grid K1/Kepler so it is identifying it on autotune.

Really make sure you've got the latest code from github.

If ECC can be disabled with nvidia-smi, please do so.

try passing -l Y  instead of -l auto. Also use -i 0 and -H 2

Always run GPU single instances per cudaminer, e.g. with flag -d N   (0<N<4) (multi-GPU support currently seems very much broken). Seems you're already doing so.

Monitor your card's utilization and clock speeds with nvidia-smi. Is it clocking up to full speed while mining?

About the Grid K1:
768 CUDA cores          <--- 4 SMX
4x entry Kepler GPUs  <--- "entry".. hmmm
Passive Cooling           <--- this

Expect about the performance of a GTX 650Ti from one of these cores. i.e. limit your expectations

And watch your temperatures... this passive cooling might be problematic.

Christian

[Never13]

I have a nvidia geforce 710m is currently 30 kh / s, can not increase more?

"-H 2 -i 0 -C 1 -m 0" in Bat File
Try 32 and 64 Cudaminer
Try to use MSI Afterburner with CoreClock Max (I have +162) and MemoryClock +270 or lower (in total i have 775/1170MHz)
And result is 38.16 kh\s . F24x8.
Carefuly with Afterburner)
Good Luck!)

[muliukov]

I tried to include the --algo=scrypt-jane:MRC in my .bat file, but then the miner immediately closes, my configurations looks like this:
cudaminer.exe -H 1 -i 0 -l auto -C 1 -o stratum+tcp://pool url.port -O username.worker:pass --a scrypt-jane:MRC
Can you please tell me exactly what I have to do, or can you please paste me the what to include in .bat file and how should my .bat file look like.

Thank you for your help

either --algo or -a  I think, but not --a

add a pause statement to the BAT file so it doesn't close immediately.

Also you need some very recent code for MRC mining. Like...compiled within the last 4 hours.

Christian

Where can I get this code?

[bigjme]

The cudaminer github

[muliukov]

The cudaminer github

I don't understand anything there Can you help me? Maybe telling what to do by steps.

[cbuchner1]

The cudaminer github

I don't understand anything there Can you help me? Maybe telling what to do by steps.

the next official release is due before the end of the month anyways...

[incorrect]

the next official release is due before the end of the month anyways...

Good to hear. I cannot wait even though I'm a Fermi user I hope to see some improvement.

Any 550ti users out there? Wondering if my settings are optimal or not.

Code:

-H 1 -i 1 -d 0 -C 1 -l F8x16 -m 1 

Any feedback would be appreciated.

[muliukov]

the next official release is due before the end of the month anyways...

Good to hear. I cannot wait even though I'm a Fermi user I hope to see some improvement.

Any 550ti users out there? Wondering if my settings are optimal or not.

Code:

-H 1 -i 1 -d 0 -C 1 -l F8x16 -m 1 

http://gyazo.com/7da56867f27ca6430d78bc975aa9c972.png

Any feedback would be appreciated.

I'm 550ti user, but I don't have cudaminer for MCR yet

[Mapin]

Hey there, i ran into a little linux-newb problem when i tried installing CUDA5.5. environment on Linux...

Since the thread is terribly big already and I dont wanna inflate it even more with my semi-related problem I started a topic of my very own
But you guys should be the ones who might help me to fix it fastly. Please take a short look:

https://bitcointalk.org/index.php?topic=435181.0

Thanks to everybody in advance.

[Magister1]

I have a nvidia geforce 710m is currently 30 kh / s, can not increase more?

performace-wise, GT 710 is at the low end...

The current github cudaminer version with the "Y" Kernel will give you the most performance, I guess. That is assuming it's a Kepler part... (not sure)

Christian

http://en.wikipedia.org/wiki/GeForce_700_Series

710M is Fermi -F or X kernel

[Devasive]

Mining vertcoin with cudaminer is giving me a lot of trouble. I'm dealing with a lot of crashing and freezing of my computer with just about any launch configuration I try. My kh/s are also about 1/4 of what I normally get mining scrypt coins (100-120 vs 390-420), but I guess that is normal for this new type of coin? It seems to mine just fine until I start using my computer while it is running, then either immediately or after a little while of use my computer freezes and sometimes crashes if I cannot close cudaminer quick enough. I have tried changing many settings but this seems to inevitably always happen.

Will the newest version of cudaminer on Github ease my troubles with mining vertcoin? I would love to give it a try but I have no idea how to utilize the code on Github.

Edit: After getting auto-tune to finally work correctly my system is much more stable. My hashrate is only slightly improved but I will take what I can get for now.

[bla2501]

The cudaminer github

I don't understand anything there Can you help me? Maybe telling what to do by steps.

the next official release is due before the end of the month anyways...

i would rather wait for the official release. thanks for help anyway

[Silverwolf_Ru]

If anyone is interested in the latest binaries, which I compiled today here they are. http://rghost.ru/51986308 Use at your own risk.

You will need to add according .dll files yourself. I compiled this for x64 architecture, sm 21 and Windows Getting 500-700 khashes on MCR with my GTX 560Ti Play around with different X and F kernels in -D --benchmark mode with --algo=scrypt:MCR

I find that the MCR algorithm makes Cudaminer eat 25% of my 4 core CPU. Not quite sure why so far, that might be the algorithm doing what it needs to do.

After further tries on Microcoin the result doesn't validate on my CPU, then it shows "share above target" I'm going to try a Velocity coin pool. Which gives the same results that the result doesn't validate on the CPU. What should the start parameters be to mine them for a Fermi card?

[cbuchner1]

After further tries on Microcoin the result doesn't validate on my CPU, then it shows "share above target" I'm going to try a Velocity coin pool. Which gives the same results that the result doesn't validate on the CPU. What should the start parameters be to mine them for a Fermi card?

share above target is a problem with nearly all MicroCoin pools. I can currently solo-mine though.

validation errors with Fermi kernels have been seen with sm_21 compilations and the CUDA 5.0 toolkit. The CUDA toolkit 5.5 worked fine.

I went back to compute_10 because apparently sm_21 brings no speed gain, but makes the cards run hotter. This and the validation problem.

Have you tried both the F and the X kernels in MRC mode? Both fail to validate?