[ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 1121.

gchil0

newbie

Activity: 59

Merit: 0

Quote from: dentldir on April 16, 2013, 02:32:34 AM

When compiling 04-14 in Linux (Ubuntu 12.04), I'm getting the following message not seen in 04-09:

/usr/local/cuda-5.0/bin/nvcc -g -O2 -arch=compute_10 --maxrregcount=124 --ptxas-options=-v -I./compat/jansson -o salsa_kernel.o -c salsa_kernel.cu
salsa_kernel.cu(269): error: no instance of overloaded function "tex2D" matches the argument list
argument types are: (texture, unsigned int, unsigned int)

Found this when playing with an Amazon EC2 GPU cluster instance on a pair of M2050's. Same error on known working Linux box.

04-09 gave about 200kH/s on the pair. That with the 12 hyperthreads on the CPU nets about 250kH/s. Still ~60x too expensive to be worth it, but fun to try anyway.

Cheers.

The newest code no longer supports a 64-bit compile. You need to compile a 32-bit binary.

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: grosminer on April 15, 2013, 10:09:52 PM

I got a 660ti and the last 4 configs were:
42x7
98x3
166x2
84x4

and which one performed best? take notes, and keep using that config. Wink

dentldir

sr. member

Activity: 333

Merit: 250

When compiling 04-14 in Linux (Ubuntu 12.04), I'm getting the following message not seen in 04-09:

/usr/local/cuda-5.0/bin/nvcc -g -O2 -arch=compute_10 --maxrregcount=124 --ptxas-options=-v -I./compat/jansson -o salsa_kernel.o -c salsa_kernel.cu
salsa_kernel.cu(269): error: no instance of overloaded function "tex2D" matches the argument list
argument types are: (texture, unsigned int, unsigned int)

Found this when playing with an Amazon EC2 GPU cluster instance on a pair of M2050's. Same error on known working Linux box.

04-09 gave about 200kH/s on the pair. That with the 12 hyperthreads on the CPU nets about 250kH/s. Still ~60x too expensive to be worth it, but fun to try anyway.

Cheers.

SubNoize

newbie

Activity: 47

Merit: 0

Finally found a thread from 2011 with instructions on how to fix the gigabyte 580gtx SOC heating/temperature issues. Cards come with 1111 volts and you're able to knock them back to 1025 which knocks like 20-30 degrees off the cards.

I know this doesn't have much to do with Cudaminer but it's now running amazingly smooth and not causing my cards to heat up. If anyone else has problems with gigabyte soc try the above.

wyodude

newbie

Activity: 35

Merit: 0

Thank you for you work. Performs great. This is on GTX 480. Contribution is forthcoming once I have some mined.

grosminer

hero member

Activity: 718

Merit: 500

Is it normal that auto-tune sets a new config everytime i start cudaminer..
I got a 660ti and the last 4 configs were:
42x7
98x3
166x2
84x4

wndrbr3d

hero member

Activity: 914

Merit: 500

Quote from: ?? on ??

Nope, not working on my droidx either, could u add a 2nd alternative download source, please?

Christian, could you just post the source to git and host the binaries there?

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: ?? on ??

Anyone else having downloading problems from mega? I've tried everything, mega just loads a blank page. I've tried on several PC's.

try disabling any flash and popup blockers temporarily.

Christian

haderach

newbie

Activity: 12

Merit: 0

I was hoping it was just a 32 vs. 64bit issue with cudaminer. But that driver issue is unfortunate. I'll have to look into TCC or running linux.

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: haderach on April 15, 2013, 10:14:42 AM

Ah, so if you're using the full scratchpad space right now do you have an idea why I can't use any launch configurations that go over 2.8GB? The program says there isn't enough memory, but according to the calculation there should be. I'm wondering if there is an integer overflow comparison somewhere?

Windows Driver Model, or WDDM doesn't allow for really huge single memory allocations. For this to work you need to run the card either on Windows XP, Linux or Windows with the TCC Driver (which is officially only for Tesla cards). Vista, 7 and 8 use WDDM.

ALSO: On 32 Bit systems or using 32 bit binaries (like cudaminer.exe) one can only address 4 GB of memory on the card at most.

Christian

Lacan82

sr. member

Activity: 247

Merit: 250

Quote from: Wole on April 15, 2013, 01:15:56 PM

Oh, If the pool isnt responding, like this;

Quote

[2013-04-15 20:16:39] HTTP request failed: Failed connect to eu-pool.give-me-ltc.com:8080;
No error
[2013-04-15 20:16:39] submit_upstream_work json_rpc_call failed
[2013-04-15 20:16:39] ...retry after 15 seconds

, ctrl-c isnt working.

I have this issue on my 650M card, but not on my GTX 570

If I hit control+C is just takes forever to end. I can force it to reset by end tasking cudaminer

Wole

newbie

Activity: 22

Merit: 0

Oh, If the pool isnt responding, like this;

Quote

[2013-04-15 20:16:39] HTTP request failed: Failed connect to eu-pool.give-me-ltc.com:8080;
No error
[2013-04-15 20:16:39] submit_upstream_work json_rpc_call failed
[2013-04-15 20:16:39] ...retry after 15 seconds

, ctrl-c isnt working.

Lacan82

sr. member

Activity: 247

Merit: 250

Is that a typo on the google docs? Someone saying they are receiving 5000 KH solo mining?

Lacan82

sr. member

Activity: 247

Merit: 250

Quote from: crunchy on April 15, 2013, 11:53:11 AM

Hmmmn,

tried cudaminner on another clean win8 pc , and it wont run complaining about some msvcp100.dll which is missing. visual c++ 2010 runtimes are installed.

the file is where it should be windows\system32

When i copy the msvcp100.dll to the directory of cudaminer: it still wont start giving me windows error code 0xc000007b.

on a clean pc that is VERY strange in the least

Huh

?

from command line try:
regsvr32 /u msvcp100.dll
regsvr32 /msvcp100.dll

This program runs fine on my windows 8 laptop.

crunchy

newbie

Activity: 19

Merit: 0

Hmmmn,

tried cudaminner on another clean win8 pc , and it wont run complaining about some msvcp100.dll which is missing. visual c++ 2010 runtimes are installed.

the file is where it should be windows\system32

When i copy the msvcp100.dll to the directory of cudaminer: it still wont start giving me windows error code 0xc000007b.

on a clean pc that is VERY strange in the least

Huh

?

haderach

newbie

Activity: 12

Merit: 0

Quote from: cbuchner1 on April 14, 2013, 11:32:55 PM

Quote from: haderach on April 14, 2013, 05:15:07 PM

Following the computation on total memory used, what is the reason that not even half of the memory available on the Titan is being used? Up until the last release today (4/14) autotune has always picked a configuration of 300-307x2 (now 263x2) which comes out to about 2.5GB of used memory at most. Since scrypt is a space/computation tradeoff algorithm isn't there potential for almost a doubling in hash performance if the full 6GB could be used?

At the moment there is no trade-off. I use the full scratchpad size per thread. So required memory size scales with the number of threads.

A trade-off happens when you try to reduce the scratchpad size at the cost of increased computation. Trying this is on my TODO list.

Christian

Ah, so if you're using the full scratchpad space right now do you have an idea why I can't use any launch configurations that go over 2.8GB? The program says there isn't enough memory, but according to the calculation there should be. I'm wondering if there is an integer overflow comparison somewhere?

cbuchner1

hero member

Activity: 756

Merit: 502

Quote from: buysolar on April 15, 2013, 09:13:11 AM

huge improvement in interactive mode. Not sure it is the new update or that I used -C 0 this time. But interactive mode autotuned me to 8x8 and that has my khash at ~46 now on my gtx460m.

The x8 kernel launch configurations were added with the very latest update. And interactive mode performance on Windows was also greatly increased. The only remaining problem is the increased CPU use in interactive mode.

nst6563

sr. member

Activity: 252

Merit: 254

Quote from: cairpre on April 14, 2013, 11:36:02 PM

I created a google doc to track card hash performance with cudaMiner, might be easier than sifting through the thread.

https://docs.google.com/spreadsheet/lv?key=0AjMqJzI7_dCvdG9fZFN1Vjd0WkFOZmtlejltd0JXbmc

Added my gtx560se and gt430 oc to the list.

buysolar

member

Activity: 125

Merit: 12

huge improvement in interactive mode. Not sure it is the new update or that I used -C 0 this time. But interactive mode autotuned me to 8x8 and that has my khash at ~46 now on my gtx460m. Which is better than my previous non interactive hash rate by 1khash. And it's better than my previous interactive hash rate by 10khash about a 30% improvement. Was nervous at first because it gave me 3 boos in a row but then the yays started coming. I'll try it in non interactive mode tonight and see if the huge bump in performance carries over to that as well.

SubNoize

newbie

Activity: 47

Merit: 0

Quote from: nst6563 on April 15, 2013, 06:41:08 AM

Quote from: SubNoize on April 15, 2013, 06:37:45 AM

Is there any way to limit the temperature the cards reach? The first version had my 580gtx's sit at around 80degrees but they're now rising way past that.

Go to techpowerup.com and grab nvidiainspector. It will let you adjust the fan speeds on the cards manually (I just set mine at 100). It will also let you set clockspeeds and voltages through command line (I made a batch file to do all of the above on my 3 nvidia cards).

I user afterburner, i think i'm having issues with my cards. Damn gigabyte put the worst coolers on them.

Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 1121. (Read 3426947 times)