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

[cbuchner1]

Seems my latest update gives the largest boost to Fermi architecture cards, and still some boost to Kepler devices.

[chanzilla]

Workaround for constant crash on exit.

Every time I went to exit the program (Ctrl-C) it would crash. Sometimes the driver would auto-recover, the screen would go black for a few seconds, then return, so I could answer the prompt and exit the program. Other times my driver would not auto-recover and I would get a black screen that lasted until I eventually hard rebooted my PC.

Anyway, I discovered that if you unplug or disable you network connection and wait for the timeout messages to appear, usually within 15 seconds, you can cleanly exit. While not an ideal solution, it sure beats taking a chance on getting the black screen and having to hard-reset your box, risking data loss, etc.

I agree with others that further CUDA optimizations would be the best use of the developer's time, so thought I would share this work-around for others who may also experience the "crash on-exits" problem.

Good tip. I just had to hard restart because this same issue. Most of the time the driver auto recovers.

[Lacan82]

Christian. is the -algo command suppose to work to switch? because it is skipping autotune and showing my gpu as (null)

[Wole]

560Ti 900 core clock:
2013-04-09: [2013-04-11 03:21:01] GPU #0:  107.78 khash/s with configuration 32x4
2013-04-10: [2013-04-11 03:32:00] GPU #0:  138.84 khash/s with configuration  64x2

Sweeet!

[grosminer]

CudaMiner 2013-04-09
GTX 660Ti 2GB
Windows 7
314.14 WHQL Driver
GPU Core OC to 1320 MHz
autotune launch config settings 148x2
130-140 khash/sec


CudaMiner 2013-04-10
GTX 660Ti 2GB
Windows 7
314.14 WHQL Driver
GPU Core OC to 1320 MHz
autotune launch config settings 96x3
140-150  khash/sec

Great update!

[wndrbr3d]

690GTX on 4-10:

autotune: 128x3 @ 161k/hash per GPU
no-autotune: 169x2 @ 158k/hash per GPU

chunked memory allocation still borked when left to run in SLI with both GPU's being fed by one instance, but runs fine if i fire up one instance per GPU.

670 with 4GB on 4-10:

autotune: 360x2 @ 169k/hash
no-autotune: 290x2 @ 165k/hash

it would seem that the trade off of more memory for fewer SP's in the 670 vs. one GPU in the 690 shows something interesting

[nst6563]

Anyone else run into problems when sli is enabled in the nvidia control panel? 
On my setup when I launch cudaminer with sli enabled, it just sits there after it says starting 'x threads' and the whole system seems very sluggish (4.6Ghz i7-3770k w/24GB ram and ssd's....definately not used to sluggish)

If I disable SLI and then run cudaminer, it seems to work fine. 

Is there any advantage to running in sli mode for mining?

[wndrbr3d]

Anyone else run into problems when sli is enabled in the nvidia control panel? 
On my setup when I launch cudaminer with sli enabled, it just sits there after it says starting 'x threads' and the whole system seems very sluggish (4.6Ghz i7-3770k w/24GB ram and ssd's....definately not used to sluggish)

If I disable SLI and then run cudaminer, it seems to work fine. 

Is there any advantage to running in sli mode for mining?

There's an issue with the chunked memory allocation routine when running in SLI mode. Leave SLI disabled and start an instance for each GPU and you should be good to go.

[gchil0]

I've confirmed on another machine containing a K20 that the Titan kernel is broken.  The regular kernel, though, shows significant improvement on my ancient 9600 GSO, the Tesla C1060's and the GTX 480's. Thanks!

[navigator]

GTX 460 768mb SC getting 104kh/s with 4-10 version and 56x2 config.

[gchil0]

BTW, thanks for the continued improvements to cudaminer!  I tossed a litecoin your way.

[xrobau]

So with the GTX570, I'm getting 179-180KHash, while using the desktop. It's JUST on the border of being annoyingly laggy. But is working quite nicely.

--Rob

[Lacan82]

So with the GTX570, I'm getting 179-180KHash, while using the desktop. It's JUST on the border of being annoyingly laggy. But is working quite nicely.

--Rob

I'm assuming that is at the standard clockrate?  I have pre-overclocked version. It is getting about 203

[cbuchner1]

I have been playing with options to improve system responsiveness. It involves choosing smaller launch configs and adding a few CPU sleep cycles before launching new CUDA work

Currently doing stratum mining for LTC on coinotron now, as wemineltc seems to have issues computing the correct payouts.

[Testarossa]

CudaMiner 1st version
GTX 295
Windows 7
313.96 Driver
Stock settings
autotune
75 khash/sec

CudaMiner 10/04
GTX 295
Windows 7
313.96 Driver
Stock settings
autotune
56 khash/sec

[nst6563]

I have been playing with options to improve system responsiveness. It involves choosing smaller launch configs and adding a few CPU sleep cycles before launching new CUDA work

Currently doing stratum mining for LTC on coinotron now, as wemineltc seems to have issues computing the correct payouts.

Would that slow down the hashrate?  If so, could you make that an option that could be disabled?

Right now I've got 3 of these nvidia cards in my system and it's completely responsive.  I'm using the onboard intel gpu for the primary display though so that may be why.  Once I switched to that, my gpu0 shot up another 10kh/s and was even with gpu2.

[peacefulmind]

I don't want to re-read the whole thread - does this support stratum?

Have a couple Titans clocked at 1GHZ and would be curious what they would do.  People seem to have great result so far!

[Lacan82]

I don't want to re-read the whole thread - does this support stratum?

Have a couple Titans clocked at 1GHZ and would be curious what they would do.  People seem to have great result so far!

have to use a proxy, but it runs great even with a proxy

[cbuchner1]

Have a couple Titans clocked at 1GHZ and would be curious what they would do.  People seem to have great result so far!

I have received reports that Titan support is presumably broken in the April 10th release. I would be interested to  get a confirmation of that, or a report that it actually works. I am definitely willing to fix it, if it is really broken.

UPDATE: I put the April 9th release up for download again, for those who want to use it on their Titan systems.

Christian

[TiWu]

Just tried the 2013-04-10 version on my GTX570. Upped the speed for me too.
Autotune output: http://pastebin.com/YgtHLWeC. Selected config 30x7 this time, promising around 180KH/s. After a while it is doing 175-178KH/s, pool reports 172.

GTX570 - 30x7
Win 8 x64
314.22 WHQL driver
stock: 732 core, 1900 shader

After 10 minutes, still no stales (yay!!!) 

Code:

[2013-04-11 20:03:51] GPU #0: GeForce GTX 570, 1491840 hashes, 179.38 khash/s
[2013-04-11 20:03:51] accepted: 50/50 (100.00%), 179.38 khash/s (yay!!!)

The 2013-04-09 version gave me config 120x2 most of the runs, and did 145-150KH/s.

I have an old PC laying around here too, with a 8800GTS 320MB.  Will try to get that online too.