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

[manofcolombia]

Yeah, except you're all chasing an AMD guy running YACMiner (cgminer based)  (and I have 1 more card waiting to go in, just need to short the PCIe slot detection pins)

....cheater... :p
I'm ordering two 280xs once amazon gets the MSI/ASUS ones instock

Average round time going down with a few new miners...but so are my estimated payouts...good/bad :p

[1714164723]

1714164723

[1714164723]

1714164723

[1714164723]

1714164723

[1714164723]

1714164723

[Beave162]

Will someone please help me find this 14-01-11 github version from the spreadsheet? I have a 640m.

Ok I fixed this issue. I installed the CUDA 5.5 update, but I had to use the notebook drivers (I have an all-in-one 640M Windows 8 64-bit). I also installed the free version of Visual Studio through the link given to me when installing CUDA 5.5. I am using patoberli's build.

I am only getting 0.70 khash/s with these settings...

-a scrypt-jane -H 0 -m 1 -i 0 -C 2 -l K7x1 -o stratum+tcp://coinmine.pl:9088 -O user:pass

[manofcolombia]

Will someone please help me find this 14-01-11 github version from the spreadsheet? I have a 640m.

Use -C 0 and should help a good deal

Ok I fixed this issue. I installed the CUDA 5.5 update, but I had to use the notebook drivers (I have an all-in-one 640M Windows 8 64-bit). I also installed the free version of Visual Studio through the link given to me when installing CUDA 5.5. I am using patoberli's build.

I am only getting 0.70 khash/s with these settings...

-a scrypt-jane -H 0 -m 1 -i 0 -C 2 -l K7x1 -o stratum+tcp://coinmine.pl:9088 -O user:pass

Use -C 0 and that should help a good deal

[Beave162]

-a scrypt-jane -H 2 -m 0 -i 0 -C 0 -l K14x1 -o stratum+tcp://coinmine.pl:9088 -O user:pass

1.2 khash/s

Thank you manofcolumbia... is it worth playing with the -b batchsize setting?

[manofcolombia]

-a scrypt-jane -H 2 -m 0 -i 0 -C 0 -l K14x1 -o stratum+tcp://coinmine.pl:9088 -O user:pass

1.2 khash/s

Thank you manofcolumbia... is it worth playing with the -b batchsize setting?

Good question...not for me to answer since I just used what Christian suggested for a 660/ti

[juggs]

Christian,

Would it possible to add functionality to be able to signal a running cudaminer instance to toggle between interactive and non-interactive mode?

I'm thinking in a Linux environment, maybe trap SIGUSR1 (signal 10) or SIGUSR2 (signal 12) as a toggle, so if the process receives a signal it switches interactiveness to whichever state it was not previously in.

This is probably pure greedy self-interest on my own part, but I do also see in this thread a few people mentioning they switch modes when going to work or to sleep. Restarting the process causes some inactive minutes as the process does whatever it does as it starts up, so it would be nice to be able to just signal the process to gobble or ungobble GPU resource.

Clearly this is in the "nice to have" bracket on the list, but it would be nice to have

[manofcolombia]

Christian,

Would it possible to add functionality to be able to signal a running cudaminer instance to toggle between interactive and non-interactive mode?

I'm thinking in a Linux environment, maybe trap SIGUSR1 (signal 10) or SIGUSR2 (signal 12) as a toggle, so if the process receives a signal it switches interactiveness to whichever state it was not previously in.

This is probably pure greedy self-interest on my own part, but I do also see in this thread a few people mentioning they switch modes when going to work or to sleep. Restarting the process causes some inactive minutes as the process does whatever it does as it starts up, so it would be nice to be able to just signal the process to gobble or ungobble GPU resource.

Clearly this is in the "nice to have" bracket on the list, but it would be nice to have

Read my mind, just had to restart the instance in interactive in order to watch an episode of HIMYM. But like you said its a thing that's just for convience.

[Prima Primat]

Restarting the process causes some inactive minutes as the process does whatever it does as it starts up,

That's just the auto-tuning. Specify your ideal launch configuration via -l and the process will start mining as soon as it's connected. That way, and with different shortcuts right on your desktop, switching really isn't a hassle.

[ManIkWeet]

Read my mind, just had to restart the instance in interactive in order to watch an episode of HIMYM. But like you said its a thing that's just for convience.

Problem would be that is has to check if it should go (non)interactive constantly, slowing it down marginally.

[Flo354]

My feedback for solo mining at 3.35 KH/s
23h -> 11h : no block found !

I think I will return mining on my pool...

[patoberli]

If I didn't make a computation error, it means for solo hashing that with around 1 kh/s you should get on average every 30 hours a successful block. That number is based on yac.coinmine.pl statistics where it currently takes around 1 minute per block with those 1850 kh/s. This number clearly is not scientific correct, I just took the last 20 blocks and made a guess
But it should help to get a rough feeling for the speed.

[cbuchner1]

I have had my fair share of dry spells even with >10kHash/s

but then I just found another two consecutive blocks early this morning...

Solo mining is a lottery.

[Flo354]

I have had my fair share of dry spells even with >10kHash/s

but then I just found another two consecutive blocks early this morning...

Solo mining is a lottery.

You are lucky 
But with all the time you spent for us (and you obviously), it's the least you can do !

[Tweek]

Any chances of an update that includes back-up pools any time soon?

currently busy with this:

I am currently trying to port Dave Anderson's code design to Fermi. I will have to replace the Kepler-specific __shfl() intrinsics with shared memory.

I gladly take code submissions for entire features   In one case I have even sent a working GTX 560 graphics
card as a "thank you, now would you please take my electronic waste" gift to greece.

I'll check it out one of these days, but I'm not sure wether I am able to program that. Anyways good to know you are still working on this because the updates seemed to have slow down. Still a great project and massive improvements in kh/s in the last few versions!

[Sukki]

Awesome!!!!
I get 520 kh/s with low intensity @ GTX 780 TI , Autotune tells me it can 620 kh/s non overclocked.
Im try overclocking later ;-).  BTW its a EVGA 780 TI ACX SC @ 74 °C and 95%.
Great job!

[Lionel]

Guys let's try playing games while mining with cuda!
Try adding this options and let me know if the game lags

-i 1 -l S2x1

If the game lags try with

-i 1 -l S1x1

Lemme know

[Tweek]

Guys let's try playing games while mining with cuda!
Try adding this options and let me know if the game lags

-i 1 -l S2x1

If the game lags try with

-i 1 -l S1x1

Lemme know

Lol that hashrate... 3.8kh/s while my 670 does 240kh/s on my own settings

[Sukki]

Awesome!!!!
I get 520 kh/s with low intensity @ GTX 780 TI , Autotune tells me it can 620 kh/s non overclocked.
Im try overclocking later ;-).  BTW its a EVGA 780 TI ACX SC @ 74 °C and 95%.
Great job!

Update: I played a little with coreclock ect.. and finally i got 601 kh/s stable out of it  
With this Nvidia isnt that far away from AMD  


[2014-01-14 14:34:03] accepted: 118/118 (100.00%), 573.58 khash/s (yay!!!)
[2014-01-14 14:34:12] GPU #0: GeForce GTX 780 Ti, 5690880 hashes, 604.09 khash/s
[2014-01-14 14:34:13] accepted: 119/119 (100.00%), 604.09 khash/s (yay!!!)
[2014-01-14 14:34:19] GPU #0: GeForce GTX 780 Ti, 3893760 hashes, 603.37 khash/s
[2014-01-14 14:34:19] accepted: 120/120 (100.00%), 603.37 khash/s (yay!!!)
[2014-01-14 14:34:24] GPU #0: GeForce GTX 780 Ti, 3306240 hashes, 603.29 khash/s
[2014-01-14 14:34:25] accepted: 121/121 (100.00%), 603.29 khash/s (yay!!!)
[2014-01-14 14:34:28] GPU #0: GeForce GTX 780 Ti, 1912320 hashes, 601.14 khash/s
[2014-01-14 14:34:28] accepted: 122/122 (100.00%), 601.14 khash/s (yay!!!)
[2014-01-14 14:34:31] GPU #0: GeForce GTX 780 Ti, 2016000 hashes, 600.50 khash/s
[2014-01-14 14:34:31] accepted: 123/123 (100.00%), 600.50 khash/s (yay!!!)
[2014-01-14 14:34:36] GPU #0: GeForce GTX 780 Ti, 2972160 hashes, 602.84 khash/s
[2014-01-14 14:34:36] accepted: 124/124 (100.00%), 602.84 khash/s (yay!!!)
[2014-01-14 14:34:39] GPU #0: GeForce GTX 780 Ti, 1820160 hashes, 600.88 khash/s
[2014-01-14 14:34:39] accepted: 125/125 (100.00%), 600.88 khash/s (yay!!!)
[2014-01-14 14:34:49] GPU #0: GeForce GTX 780 Ti, 6347520 hashes, 604.78 khash/s
[2014-01-14 14:34:50] accepted: 126/126 (100.00%), 604.78 khash/s (yay!!!)
[2014-01-14 14:34:53] GPU #0: GeForce GTX 780 Ti, 1889280 hashes, 601.26 khash/s
[2014-01-14 14:34:53] accepted: 127/127 (100.00%), 601.26 khash/s (yay!!!)
[2014-01-14 14:34:58] Stratum detected new block
[2014-01-14 14:34:58] GPU #0: GeForce GTX 780 Ti, 3444480 hashes, 603.52 khash/s
[2014-01-14 14:35:01] GPU #0: GeForce GTX 780 Ti, 1681920 hashes, 600.22 khash/s
[2014-01-14 14:35:01] accepted: 128/128 (100.00%), 600.22 khash/s (yay!!!)
[2014-01-14 14:35:13] GPU #0: GeForce GTX 780 Ti, 6992640 hashes, 604.92 khash/s
[2014-01-14 14:35:13] accepted: 129/129 (100.00%), 604.92 khash/s (yay!!!)
[2014-01-14 14:35:17] GPU #0: GeForce GTX 780 Ti, 2453760 hashes, 602.41 khash/s
[2014-01-14 14:35:17] accepted: 130/130 (100.00%), 602.41 khash/s (yay!!!)
[2014-01-14 14:35:27] GPU #0: GeForce GTX 780 Ti, 6048000 hashes, 604.70 khash/s

[cbuchner1]

You guys may want to run the X kernel from github on your Fermi devices, both for scrypt and scrypt-jane. Please report results...

Note: run autogen.sh and configure again when on Linux. On Windows change compilation options to sm_20 for test_kernel.cu with 32 registers max.

[patoberli]

Big thanks!
First of all, interactive mode works smooooth
Speed scrypt-jane went from 0.46 kh/s (non-interactive) to 0.59 kh/s (interactive) or 0.63 (non-interactive) on my Quadro 4000 with this start configuration:

Code:

D:\bitcoin\yacoin\test>cudaminer.exe -a scrypt-jane -l X12x1 -o stratum+tcp://yac.coinmine.pl:9088 -O username.2:password

Scrypt went from 84 kh/s (interactive) down to 63 kh/s (interactive), including some HW errors:

Code:

D:\bitcoin\yacoin\test>cudaminer.exe -i 1 -H 2 -C 1 -l X8x16 -o stratum+tcp://gld.cryptcoins.net:5533 -O username.2:password
           *** CudaMiner for nVidia GPUs by Christian Buchner ***
                     This is version 2013-12-18 (beta)
        based on pooler-cpuminer 2.3.2 (c) 2010 Jeff Garzik, 2012 pooler
               Cuda additions Copyright 2013 Christian Buchner
           My donation address: LKS1WDKGED647msBQfLBHV3Ls8sveGncnm

[2014-01-14 15:45:22] 1 miner threads started, using 'scrypt' algorithm.
[2014-01-14 15:45:22] Starting Stratum on stratum+tcp://gld.cryptcoins.net:5533
[2014-01-14 15:45:22] Stratum detected new block
[2014-01-14 15:45:23] GPU #0: Quadro 4000 with compute capability 2.0
[2014-01-14 15:45:23] GPU #0: interactive: 1, tex-cache: 1D, single-alloc: 1
[2014-01-14 15:45:23] GPU #0: using launch configuration X8x16
[2014-01-14 15:45:23] GPU #0: Quadro 4000, 22.76 khash/s
[2014-01-14 15:45:40] GPU #0: Quadro 4000 result does not validate on CPU (i=203
5, s=0)!
[2014-01-14 15:45:42] GPU #0: Quadro 4000, 63.17 khash/s
[2014-01-14 15:45:42] accepted: 1/1 (100.00%), 63.17 khash/s (yay!!!)
[2014-01-14 15:45:55] GPU #0: Quadro 4000, 63.35 khash/s
[2014-01-14 15:45:56] accepted: 2/2 (100.00%), 63.35 khash/s (yay!!!)
Ctrl-C
^C

With scrypt and -C 0 speed is down to 57 kh/s.

Please note that I also had to change the compiler option compute_30 to _20 and not just sm_20.

The GPU load with scrypt-jane is still a bit lower (93-99% in GPU-Z) than with scrypt (constant 98-99%), but I appreciate that, because the GPU fan is running much slower and the temp of the GPU is also much lower


[Edit]
Just discovered that my CPU load went from ~1% to 11% with this new version with scrypt-jane.