Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 781. (Read 3426922 times)

Here's a good video for help choosing your 750ti:  https://www.youtube.com/watch?v=vpVEA6zBrcc

I'm getting a boatload at 173...

Nice!!

Going to mine motorboat these suckers till a) they take off and are worth something or b) coin collapses and is no longer being supported by devs.

I believe in Titties

After I upgraded my three pieces of software, I did do a reboot of the pc  ( win 7 64 , 8gb ram on an OLD ASUS M2N68LA mobo ) I tend to oc PRETTY hard and OV aswell to max allowable settings, its the only way I can get the max hash out of this lil 650Ti, I keep her in the mid 70's for temps. I tend to think its a driver crash, as I run it just on the brink @ all times, as Im 90% in front of the PC and catch a failure withn seconds usually. Mabey when Im asleep, I will back her down a tad.. its probabally the best way. FWIW this ASUS GTX 650Ti (1 GB version )is running between 96-98 kh/s on N-Scrypt. With a K4x24 setting. My bat is as follows:

cudaminer.exe  --algo=scrypt:2048 -H 1 -i 0 -C 0 -m 1 -R 1 -l K4x24 -o  ect ect ect

Plus their method of Overclocking for alt coin is seriously in need of a change, I've weighed in on the thread

After a graphics driver update, any GPU. Reboot. Regardless. Can't hurt. 

I saw a similar phenomenon while updating to the new Nvidia driver with Cudaminer running. It did not recur before or after.

BTW, my hash rate on Maxcoin increased by around 2% after updating the driver (old 580 card).

the "unknown error" after several minutes of successful mining might be a sign of a hardware related fault. Maybe something is unstable on the PCI express bus.

a better error recovery in CUDA would be nice to have, but it would not avoid running into these errors in the first place.

Try if the BIOS allows you to reduce PCI express bus speed (e.g. switching from PCI express 3.0 to 2.0) or to change some bus related options.

I've recently had a github support ticket open with the same error appearing with some Tesla boards. I've closed it somewhat disgruntled when the guy stated that I was about as helpful as a merry-go-round.  Well, I cannot solve your hardware ailments, guys. I might work on implementing better error recovery though.

BTW: CUDA 6.0 shows much better error messages now (not just unknown error, or unspecified launch failure). But cudaminer has to be built with the CUDA 6.0 toolkit for this to work - and there's performance related problems with this.

Christian

to recover from random errors (cause by aggressive overclock etc...) mine with --time-limit=900 in a .bat file that has

:start
cudaminer.exe --time-limit=900 and other arguments
goto start

I really hate that my first post in this forum has to be "cry for help" so please go easy on me!

I have a GTX 770 that was working fine in one rig (avg 330kh/s) but after moving it to another rig cudaminer is now throwing "cudaerror 30" errors after a few mins of successful mining.  I've reinstalled Windows twice in this new rig, going from x64 to x32 then back to x64. I've tried every version of cudaminer since release 12-18, both 32- and 64-bit versions and they all throw some type of error.  The 12-18 version just stops working and Windows shows its "cudaminer has stopped working" junk.  The latest 64-bit versions don't work at all.  They throw some "application was unable to start successfully" error.

This rig has 4GB RAM, Core2 Duo CPU, Intel MB (granted, its from around 2008).  I have 4 other rigs running various NVidia cards and none of them have these errors. 

Here are a few of the cudaerrors.  These just repeat thousands of times until you kill cudaminer or the OS locks up.

cudaError 30 (unknown error) calling 'cudaMemcpyAsync(hash, context_hash[stream][thr_id], mem_size, cudaMemcpyDeviceToHost, context_streams[stream][thr_id])'
(D:/Christian/Documents/Visual Studio 2010/Projects/CudaMiner/sha256.cu line 446)

cudaError 30 (unknown error) calling 'cudaEventRecord(context_serialize[stream][thr_id], context_streams[stream][thr_id])'
(D:/Christian/Documents/Visual Studio 2010/Projects/CudaMiner/salsa_kernel.cu line 932)

cudaError 30 (unknown error) calling 'cudaStreamQuery(context_streams[stream][thr_id])'
(D:/Christian/Documents/Visual Studio 2010/Projects/CudaMiner/salsa_kernel.cu line 938)

cudaError 30 (unknown error) calling 'cudaStreamWaitEvent(context_streams[stream][thr_id], context_serialize[(stream+1)&1][thr_id], 0)'
(D:/Christian/Documents/Visual Studio 2010/Projects/CudaMiner/salsa_kernel.cu line 926)


Thanks in advance for any and all help!

I'm not sure @ what time over the evening, but I did have a cudaminer ( more likely driver ) failure, I awoke to a cmd window full of scrolling text ( its never legible so I cant past the error ). I'll keep an eye on it and see if I can post any valuable info. Christian, I do have a question, as stated previously, is there any way to send cudaminer a stop command, yet still have the text in the cmd window, so that if it is a cudaminer issue I can post the screen shot here? Its weird when it does it ( typically is a driver failure ), and it just keeps scrolling over and over, and is un legible.

Also, Im not sure if its possible , however does any one know of a way to automatically recover from a driver failure, so that when the drivers actually recover ( its usually fast ) , that the miner would kick back into action @ the last previous settings?

Not only am I running the new nVidia drivers, & new cudaminer, but also the latest version of GPU tweak 2.5.4.2 ( I had been using an older one ) so I now have unfortunately 3 variables :S.

lol

I want 4 750tis for my main rig...I'd love if you could do at least 2 way sli with them, but not possible

I'm gonna tweak one of these soon, a friend's getting one and letting me remote into his rig for proper tuning (not like those guys from the site Christian posted at did...no lookup gap lol)

That only makes 500000 ladies...

Damn it... Bought 3 with a portion of my Bitcoin haul from the Maxcoin launch...  

EDIT: And today I become a Tittiecoin MILLIONAIRE!! 

It certainly does put the new card in a bad light for scrypt-chacha coins at high N-factors.  The card is in the running for the next rig, but it's got some big hurdles to overcome for me to choose it (being so familiar with the AMD world and code), but at the end of the day speed and power consumption are serious factors in that choice well above the color of my brand...

quick CUDA 6.0 tests showed similar performance with Yacoin mining (lower case k kernel)
but about 30% less performance in scrypt mining using the nVidia submitted K kernel. Doh!

I cannot build a new cudaminer version with CUDA 6.0 (targeting Maxwell specifically) if I cannot get this performance regression on older architectures under control.

Christian

cudaminer.exe --algo=scrypt:2048 -l T12x20 -C 2 -o stratum+tcp://panda.nitro.org:3338

got "T kernel ignores the texture cache argument" giving me the same hash rate, so removed the launch config

cudaminer.exe --algo=scrypt:2048 -C 2 -o stratum+tcp://panda.nitro.org:3338
still got the t kernel ignores the .....

cudaminer.exe --algo=scrypt:2048 -lT12x20 -i 0 -C 2 -o stratum+tcp://panda.nitro.org:3338

got the T kernel ignores but had the same hash rate with some in ~+310 khash/s

cudaminer.exe --algo=scrypt:2048 -i 0 -C 2 -o stratum+tcp://panda.nitro.org:3338
still the same T kernel ignores ..... (result: -i 0 -C 0 -H 0) and hash rate average 310 with a few 318 in the beging. That said the autotuned launch configuration was the same T12x20.

Hope this helps and if have any ideas to increase hash would be glad to hear it.

 

this was not done well.

Yacoin needs lookup gap -L 2, 3, or 4 with mid range cards. I would expect the 750Ti to break 2.5 kHash/s.

Keccak needs huge launch configurations (like -l K1024x32) and the K kernel (T sucks for Keccak).

I followed the indication given there https://github.com/TheoRettisch/p2pool-panda
(apparently it should also work with vertcoin, didn't try tough)

How did you install the node?