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

[Morgahl]

I cant remember what version i was using because i overwrote it with this version, i downloaded it about a month ago, which would have been the latest version at the time..
I didnt even use any launch configs with that version and i got 45 KH/s
Im thinking of going back to it if i can find a link to the earlier versions.
I only downloaded this version to mine vertcoin, which is really no point with such a low powered card anyway, the hashrates im referring to are for any scrypt based coin such as dogecoin and flappycoin etc.

You know, now that I think about it, I had the pleasure of working with a GTX 275 earlier this week and saw fairly similar numbers.

One thing to try, look at rolling back to the 332.21 Drivers. Some have noted a drop in hash rate with the newest drivers especially on older models. In addition it is not a bad idea to go back to a previous version of CudaMiner for a card that old as legacy cards are just not supported anymore (though the F and f kernels are your best bet in the recent versions).

[1714544330]

1714544330

[1714544330]

1714544330

[cvax]

found some basic information about the cgminer API at last: https://github.com/ckolivas/cgminer/blob/master/API-README

cudaminer would probably have to implement only the monitoring APIs at first. The part where you're allowed to switch parameters could be added later (and if so, only the subset that matters to us)

Christian

Adding Monitoring as well as Fail Over will be a huge step toward moving Nvidia mining beyond Hobby-scale (<10MH/s).

On a side note, I've noticed that the topic of Hash per Watt has only ever come up seriously for the Maxwell cards due to how improved it is over the comparable AMD setups. I realized that we almost all look at how high we can tune the hash-rate on a given card, but I don't think anyone has seriously delved into tuning a card for the best Hash/Watt. We've made comparisons between each algo and associated tuning but the main driver has been hash-rate > all. With these new Maxwell Cards it might be a good idea to start seriously exploring Hash/Watt since we are on the cusp of legitimate wide scale use of NVidia

Remote API
Fail Over
Hash/Watt increase

I'm excited to say the least  though patient as you will get to it when you get to it 

I have done a lot of spreadsheet math this past week analyzing how to make Maxwell profitable nowadays. All of the online calculators I saw are 100% useless and do not even come close to giving you a clear picture on profitability due to the structuring of these various coins. Obviously hash/watt is a critical component that can make or break the investment, but through all the variables I've tracked and differing values I've tried it isn't as important as other factors. The biggest most important one being coin selection (and maybe BTC/USD rates).

[bacon_wrangler]

It may be hard to get an idea if just that 1 component died..but if thats the only one that let loose, and you are familliar with SMT work, it may be repairable..would need a desoldering station (preferabally air) and some solder paste and a replacement component. Thats a bummer tho.. I feel for ya

It's on the warranty so I'll RMA, it's just a hassle.
Scared the sh*t outta me, it was loud.  

I bet THAT was an attention-grabber. 

[j0achim]

Anyone who runs multiple nVidia cards on risers?

I am looking into running 5-6 cards per rig, but i think someone earlier in this thread mentioned that cudaMiner still has some trouble with high traffic so that risers could be a problem. (1x to 16x risers, 5 or 6 per rig)

[ivanlabrie]

Anyone who runs multiple nVidia cards on risers?

I am looking into running 5-6 cards per rig, but i think someone earlier in this thread mentioned that cudaMiner still has some trouble with high traffic so that risers could be a problem. (1x to 16x risers, 5 or 6 per rig)

Get a quad sli capable board...use 4 per rig.

[Bearclaw]

Anyone who runs multiple nVidia cards on risers?

I am looking into running 5-6 cards per rig, but i think someone earlier in this thread mentioned that cudaMiner still has some trouble with high traffic so that risers could be a problem. (1x to 16x risers, 5 or 6 per rig)

I notice a severe slowdown from 340 khash to ~280 khash when going to a 1 to 16 riser, I was able to recover some speed at the suggestion of Christian to use the -H 2 switch which got me back up to the 320 khash level, but nothing I can do gets me back to 340.  I will be getting 16 to 16 risers in tomorrow for testing this weekend. Not sure if anyone has done any serious testing of the 1 to 16 risers with NVidia cards and cudaminer, I will try with the cards I have (750 Ti, 680, 780 Ti) but it will have to wait till this weekend.  I will post what I find.  If anyone has any special requests, just post 'em here.  I also have a motherboard with 4 x1 and 3 x16 slots coming in tomorrow.

[Morgahl]

Adding Monitoring as well as Fail Over will be a huge step toward moving Nvidia mining beyond Hobby-scale (<10MH/s).

On a side note, I've noticed that the topic of Hash per Watt has only ever come up seriously for the Maxwell cards due to how improved it is over the comparable AMD setups. I realized that we almost all look at how high we can tune the hash-rate on a given card, but I don't think anyone has seriously delved into tuning a card for the best Hash/Watt. We've made comparisons between each algo and associated tuning but the main driver has been hash-rate > all. With these new Maxwell Cards it might be a good idea to start seriously exploring Hash/Watt since we are on the cusp of legitimate wide scale use of NVidia

Remote API
Fail Over
Hash/Watt increase

I'm excited to say the least  though patient as you will get to it when you get to it 

I have done a lot of spreadsheet math this past week analyzing how to make Maxwell profitable nowadays. All of the online calculators I saw are 100% useless and do not even come close to giving you a clear picture on profitability due to the structuring of these various coins. Obviously hash/watt is a critical component that can make or break the investment, but through all the variables I've tracked and differing values I've tried it isn't as important as other factors. The biggest most important one being coin selection (and maybe BTC/USD rates).
[/quote]

Profitability, such as the craziness surrounding the MAX release where you just wanted to throw as much hash at it as possible and damn the power draw, can easily trump a Max/Watt. I'm not saying that Max/Watt should be pursued to the ends of time, but it DOES need to be a consideration for any build as well as tuning.

There is a point when Hash/Watt is a bad thing and while Maxwell does very well in this it, does mean that older legacy cards will not do nearly as well and really anyone intending to seriously mine with an older card needs to take into account Hash/Watt for each Algo and really examine if this is a hobby or something they intend to make money on. Scyrpt-Jane at lower -L can really sit back and sip at wattage wher a full blown -L 6 or Scrypt mining can make you really examine your power bill

TL;DR - Hash/Watt is a necessary consideration for everyone

[cbuchner1]

TL;DR - Hash/Watt is a necessary consideration for everyone

I am in the process of replacing power guzzling cards like 560Ti 448 core edition, 660Ti with the new Maxwell GTX
750Ti line of cards.

I won't be replacing the expensive beasts like GTX 780Ti, GTX 780, Asus MARS anytime soon because they haven't
really done enough work yet,  so throwing them out at this point is not justified

Christian

[rahrahrah]

Anyone who runs multiple nVidia cards on risers?

I am looking into running 5-6 cards per rig, but i think someone earlier in this thread mentioned that cudaMiner still has some trouble with high traffic so that risers could be a problem. (1x to 16x risers, 5 or 6 per rig)

Get a quad sli capable board...use 4 per rig.

New 750 Ti are not SLI capable / compatible from what I've read.

[rahrahrah]

TL;DR - Hash/Watt is a necessary consideration for everyone

I am in the process of replacing power guzzling cards like 560Ti 448 core edition, 660Ti with the new Maxwell GTX
750Ti line of cards.

I won't be replacing the expensive beasts like GTX 780Ti, GTX 780, Asus MARS anytime soon because they haven't
really done enough work yet,  so throwing them out at this point is not justified

Christian

Great news as I've just purchased 5 of them! Also keeping my 7990s for a while... but looking forward to Cudaminer running even better on the 750 Ti cards

[KoSoVaR]

I got 6 x MSI 750 TI working in Linux (only 5 are mining with cudaminer).  With the 5 x GPUs I'm pulling 1.1MH/s and 340w at the wall right now.  

Shit I'm having issues with now...

• getting 6th card to mine, it shows up in lspci | grep VGA
• getting more hash power.. I am going to flash from 35mv to 65mv tomorrow.  tried in windows but failed, it's late now...

If you're interested in some quick directions, here you are.  I had a really tough time getting this going, especially with nomodeset.  Xubuntu will freeze on install and Debian will also give you tons of errors about nouveau.  The 750TI doesn't like it.  

Code:

Install Xubuntu 13.10 - you must boot with nomodeset http://ubuntuforums.org/showthread.php?t=1613132

Download Nvidia 331.49
http://www.nvidia.com/object/unix.html

Download Cuda toolkit (12.10 seemed to work)
https://developer.nvidia.com/cuda-downloads

sudo vi /etc/modprobe.d/blacklist.conf 

Add the following:
blacklist amd76x_edac
blacklist vga16fb
blacklist nouveau
blacklist rivafb
blacklist nvidiafb
blacklist rivatv

sudo apt-get remove --purge nvidia*

Ctrl+Alt+F1
sudo service lightdm stop

sudo sh NVIDI*.run

Should be able to boot back into x now.

sudo service lightdm start

sudo apt-get install freeglut3 freeglut3-dev build-essential libx11-dev libxmu-dev libxi-dev libgl1-mesa-glx libglu1-mesa libglu1-mesa-dev gcc g++ gcc-4.6 g++-4.6 linux-headers-generic linux-source libcurl3 libcurl3-dev build-essential autoconf libtool screen vim

sudo ln -s /usr/lib/x86_64-linux-gnu/libglut.so.3 /usr/lib/libglut.so

sudo update-alternatives --remove-all gcc
sudo update-alternatives --config gcc
sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-4.6 1
sudo update-alternatives --install /usr/bin/gcc gcc /usr/bin/gcc-4.8 2

sudo update-alternatives --config gcc # select 4.6 (just need this for cuda toolkit compile)
sudo chmod +x cuda*.run
sudo ./cuda*.run

You'll need to add a few things to $PATH and $LD_LIBRARY_PATH

vim ~/.bashrc

add this at the bottom:

export LD_LIBRARY_PATH=$LD_LIBRARY_PATH:/usr/local/cuda-5.5/lib64
export PATH=$PATH:/usr/local/cuda-5.5/bin

# Should now be able to build cudaminer

 git clone https://github.com/cbuchner1/CudaMiner.git
./autogen.sh
./configure --prefix=$HOME
make
make install

#I then created a script called startCudaMiner.sh

vim startCudaMiner.sh
#insert following line in to file
./cudaminer -o stratum+tcp://URL -u USERNAME.WORKER -p PASSWORD -i 0 -l T5x24 -C 1 -H 2
#save and quit text editor
chmod a+x startCudaMiner.sh

#and drumroll...

./startCudaMiner.sh

If this helped consider donating some beer 14D6t8hEV6PHzpg4gJ9hSqPH9g1BAdapJa (BTC)

Here are my cards...


Here they are hashing...

[djm34]

what kind of hashing power does get the gtx750ti on MRC ?

[cbuchner1]

I got 6 x MSI 750 TI working in Linux (only 5 are mining with cudaminer).  With the 5 x GPUs I'm pulling 1.1MH/s and 340w at the wall right now.  

Your GPU #4 only shows 96 kHash/s. This is odd... Does this also happen when you pass the option -d 4 to mine ONLY on this card? What's your CPU utilization while mining?

overclocking tools for Linux will be coming, I am just not sure when exactly (this is not under my control). Be careful not to brick your cards doing BIOS mods.

Christian

[liomojo1]

what kind of hashing power does get the gtx750ti on MRC ?

152-155khs,  overclocked to 1360mhz

[j0achim]

Anyone who runs multiple nVidia cards on risers?

I am looking into running 5-6 cards per rig, but i think someone earlier in this thread mentioned that cudaMiner still has some trouble with high traffic so that risers could be a problem. (1x to 16x risers, 5 or 6 per rig)

I notice a severe slowdown from 340 khash to ~280 khash when going to a 1 to 16 riser, I was able to recover some speed at the suggestion of Christian to use the -H 2 switch which got me back up to the 320 khash level, but nothing I can do gets me back to 340.  I will be getting 16 to 16 risers in tomorrow for testing this weekend. Not sure if anyone has done any serious testing of the 1 to 16 risers with NVidia cards and cudaminer, I will try with the cards I have (750 Ti, 680, 780 Ti) but it will have to wait till this weekend.  I will post what I find.  If anyone has any special requests, just post 'em here.  I also have a motherboard with 4 x1 and 3 x16 slots coming in tomorrow.

Thanks! Guess i hold off buying that considerably large shipment just a bit longer


Im also very interested to hear what results your getting on that 7x motherboard.

[djm34]

I received today my 1 to 16 riser and I am testing it. Apparently it seems to work well.
I was rather unsure as I have to plug the monitor into the 290x to get the temp readout and oc, but it seems to work fine
(it becomes just a bit laggy when I need to modify a setting in afterburner, even though it is for the other card) now the temps are fine for both card. However the motherboard tends to get hot (45°C, but it is less than before).
In terms of hashrate, I don't see any difference (but it is for a radeon) actually it hashes faster as I can increase a bit the TDP and keeping the card relatively cool.

edit: while I was writing that message, I had some artefacts on screen and had to power on-off... well... (really need to find a way so I don't need to plug the card into

[ivanlabrie]

I received today my 1 to 16 riser and I am testing it. Apparently it seems to work well.
I was rather unsure as I have to plug the monitor into the 290x to get the temp readout and oc, but it seems to work fine
(it becomes just a bit laggy when I need to modify a setting in afterburner, even though it is for the other card) now the temps are fine for both card. However the motherboard tends to get hot (45°C, but it is less than before).
In terms of hashrate, I don't see any difference (but it is for a radeon) actually it hashes faster as I can increase a bit the TDP and keeping the card relatively cool.

And the 750s on risers? :p

We don't care bout radeon...not much, already well documented xD

[djm34]

sorry but it's an all new experience for me ;-)

[hater]

I got 6 x MSI 750 TI working in Linux (only 5 are mining with cudaminer).  With the 5 x GPUs I'm pulling 1.1MH/s and 340w at the wall right now.  

Your GPU #4 only shows 96 kHash/s. This is odd... Does this also happen when you pass the option -d 4 to mine ONLY on this card? What's your CPU utilization while mining?

overclocking tools for Linux will be coming, I am just not sure when exactly (this is not under my control). Be careful not to brick your cards doing BIOS mods.

Christian

Hello
My os: WIN8 64X
MEMONY：8 G
CARD :GTX 760 X2

When I run the application, application error, tip: application error, address : 0xc000007b. Press the OK button to close the application.

http://www.icolorlife.cn/img/e1.jpg

[KoSoVaR]

what kind of hashing power does get the gtx750ti on MRC ?

152-155khs,  overclocked to 1360mhz

Yes it does happen when I use -d 4.  Just confirmed.

If you'd like remote access to this box I'd be glad to give it to you - you can mine wherever you want with it, develop whatever you want.  It's basically my cudaminer playground right now, and I assume it would be nice for you too.  Shoot me a PM and we can gtalk/IRC/whatever you prefer if you're on the Telegram wagon.

When I use -d 4 I'm getting 96w idle with 6 cards spinning fans, but the 6th card never spins.  In terms of hardware I am using the same USB risers that I was using with 6 x 280X on the same motherboard (I had them undervolted and 2 PSUs).

For people asking questions about risers... I'm not sure whether the 5th/6th GPU aren't getting enough power or if they're having issues communicating because of the risers, but the other cards seem to be doing just fine.  Here's a pic of the rig