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

[Amph]

it's strange, the program start only with bitcoin, but not with litecoin...
with litecoin it give me an error

This is meant only to work with litecoin not bitcoin

yeah i know, this is why it's strange, maybe can be that i connect to stratum, using p2pool?

it's strange, the program start only with bitcoin, but not with litecoin...
with litecoin it give me an error

What command are you trying to execute? are drivers up to date?

i'm using p2pool,
i run litecoin client, then p2pool with "--net litecoin", and then cudaminer, but it say fail to execute or something, and then restart in  15 sec

[1714323692]

1714323692

[1714323692]

1714323692

[CumpsD]

I've added a lot of results for a EVGA GTX 680 to the Google Docs.

Big difference in kernel selection between interactive and non-interactive (S-kernel)

Thanks for this miner!

[cbuchner1]

Big difference in kernel selection between interactive and non-interactive (S-kernel)

interactive picks smaller kernels that complete within 1/30th of a second. And then the CPU pauses for a millisecond (actually twice per hash computation, once inbetween the A and B kernels and once after the B kernel), giving you about 60 frames per second for doing other stuff, like watching movies.

[CumpsD]

That explains the difference

Using the interactive auto-tuned setting with not-interactive (so without letting it re-tune) had a nice effect of increased rates, without much lag.

I guess it choose a small kernel which completes fast, but then doesn't pause the CPU.

[cbuchner1]

When using the autotuned setting from non-interactive mode in interactive mode, you get higher hash rates. The CPU will still pause inbetween, but the computation takes longer. So the ratio of computation vs. pause is higher -> more hashes/sec at increased lag.

[CumpsD]

When using the autotuned setting from non-interactive mode in interactive mode, you get higher hash rates. The CPU will still pause inbetween, but the computation takes longer. So the ratio of computation vs. pause is higher -> more hashes/sec at increased lag.

Makes sense. What I tried was the other way around actually. Using the autotuned setting from interactive in non-interactive. (Just for testing, now I've moved on to the more correct settings)

[cbuchner1]

I am currently a bit concerned because large block configurations like 56x8 or 64x4 cause results to not validate on both of my 560Ti cards. Even with texture cache turned off. I wonder what I broke and when that happened. UPDATE: okay, this must be the memory requirement - my 560 cards aren't blessed with 3Gigs of VRAM.

[narayan]

Anyone had any luck with the GTX 260, I had an old one lying around which I chucked in a test machine but can't get more than 20kh out of it.

[Stoneysilence]

Anyone had any luck with the GTX 260, I had an old one lying around which I chucked in a test machine but can't get more than 20kh out of it.

Looking at the stats for other cards I am guessing 20kh is probably about right.  My 460M in my laptop is about 50kh, in the google doc from page 1 there is a 210 and it gets about 6kh.

[cbuchner1]

Anyone had any luck with the GTX 260, I had an old one lying around which I chucked in a test machine but can't get more than 20kh out of it.

I get 44 kHash with Windows 7, cudaminer Version older than 2013-04-30 using

-i0 -l S27x3

It should work a bit faster on Windows XP or Linux

Christian

[statdude]

Help, how can I slow this down on my GTX  570 - it's running over 90 degrees...

My command line:

cudaminer -o http://ltc.crypdough.com:9332/ -u user -p pass -i 0

[cbuchner1]

Help, how can I slow this down on my GTX  570 - it's running over 90 degrees...

For one, -i 1 helps a little (you lose some 25% of hashing speed on fast cards)

Or use a tool called nvidia-inspector

-Set the fan to Manual and higher Speeds
-clock down the card's core speed and possibly undervolt a bit.

[TiWu]

Help, how can I slow this down on my GTX  570 - it's running over 90 degrees...

My command line:

cudaminer -o http://ltc.crypdough.com:9332/ -u user -p pass -i 0

I use MSI afterburner on my (non-MSI) GTX 570. It allows you to redraw the automatic slope for fan speed.
You can find it here: http://event.msi.com/vga/afterburner/

Trough settings-> fan you can see the graph and adapt it.
I've set 4 point for my card:

• 30% at 30°
• 55% at 60°
• 85% at 80°
• 100% at 100° (but the Club3D GTX570 I have only goes up to 85%)

I set the fan update speed to 5000ms and a temp hysteresis of 1° to avoid constant 'flapping' of the fan speed.

This is running 24/7 at around 70° at 70% fan, with these options in cudaminer: -i 0 -C 2 -l 30x8, giving me around 226 kh/s.
Hope this helps!

edit: BTW, I run this GPU dedicated for mining, my desktop is using the internal graphics of my i5. If you are using your desktop on the card, I suggest using the -i 1 option.

[mhps]

So my question is disabling longpoll a recommended practice where it is an option?

From here it does seem longpolling is an extra protocol that doesn't bring any benefit for using local stratum proxy: https://mining.bitcoin.cz/stratum-mining

[420]

What speed should I get with a GTX 460M

or a GTX 680

check here
https://docs.google.com/spreadsheet/ccc?key=0AjMqJzI7_dCvdG9fZFN1Vjd0WkFOZmtlejltd0JXbmc#gid=1


that should give you an idea

yay 185khash with 680 not overclocked.

now how do I get the miner to connect to COINOTRON?

0.02BTC reward!

[tacotime]

What speed should I get with a GTX 460M

or a GTX 680

check here
https://docs.google.com/spreadsheet/ccc?key=0AjMqJzI7_dCvdG9fZFN1Vjd0WkFOZmtlejltd0JXbmc#gid=1


that should give you an idea

yay 185khash with 680 not overclocked.

now how do I get the miner to connect to COINOTRON?

0.02BTC reward!

get mining_proxy.exe from ltcmine.ru
mining_proxy.exe -pa scrypt -o coinotron.com -p 3334
connect cudaminer to localhost port 8332, use normal miner password

[420]

yay 185khash with 680 not overclocked.

now how do I get the miner to connect to COINOTRON?

0.02BTC reward!

get mining_proxy.exe from ltcmine.ru
mining_proxy.exe -pa scrypt -o coinotron.com -p 3334
connect cudaminer to localhost port 8332, use normal miner password

HTTP request failed: Failed connect to localhost:8332
json_rpc_call failed

my file:
cudaminer.exe -o localhost:8322 -O user:pass

PROXY:
mining_proxy.exe -pa scrypt -o coinotron.com -p 3334

[dell00]

cannot download the software.  Anyone can give an aternative downloading site?

[mhps]

What speed should I get with a GTX 460M

or a GTX 680

check here
https://docs.google.com/spreadsheet/ccc?key=0AjMqJzI7_dCvdG9fZFN1Vjd0WkFOZmtlejltd0JXbmc#gid=1


that should give you an idea

yay 185khash with 680 not overclocked.

now how do I get the miner to connect to COINOTRON?

0.02BTC reward!

Hi this is how I do -- just tried and it works

download  http://www.wemineltc.com/files/stratum_proxy.exe 
Put it in your cudaminer directory.
Note that there are different versions on the net.

Run this to start proxy

Code:

stratum_proxy.exe -o coinotron.com -p 3334 -pa scrypt -nm

Run this to mine

Code:

cudaminer -o http://127.0.0.1:8332 -O username:passwd -d 1 -l 112x8   -i 0 -C 2 --no-longpoll

You should change the "-d 1 -l 112x8   -i 0 -C 2 " to suit your card. Use -D to find out the best -l parameters.

If it works 1HxvAEC4nj37hyenz9DwsyrMMnkwzpbzno

[verzon]

What speed should I get with a GTX 460M

or a GTX 680

check here
https://docs.google.com/spreadsheet/ccc?key=0AjMqJzI7_dCvdG9fZFN1Vjd0WkFOZmtlejltd0JXbmc#gid=1


that should give you an idea

yay 185khash with 680 not overclocked.

now how do I get the miner to connect to COINOTRON?

0.02BTC reward!

Hi this is how I do -- just tried and it works

download  http://www.wemineltc.com/files/stratum_proxy.exe  
Put it in your cudaminer directory.
Note that there are different versions on the net.

Run this to start proxy

Code:

stratum_proxy.exe -o coinotron.com -p 3334 -pa scrypt -nm

Run this to mine

Code:

cudaminer -o http://127.0.0.1:8332 -O username:passwd -d 1 -l 112x8   -i 0 -C 2 --no-longpoll

You should change the "-d 1 -l 112x8   -i 0 -C 2 " to suit your card. Use -D to find out the best -l parameters.

If it works 1HxvAEC4nj37hyenz9DwsyrMMnkwzpbzno

Hey :-)

i got a "result does not validate on CPU!" after starting cudaminer (after try (start) stratum_proxy) .. (coionotron)

what does that message mean?