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

[13G]

Also, what's up with people getting good results on their 780s using T/Z12x24 ? I get absolutely craptastic results with that!

That's like the 15x24 for some 780ti... yes me too I'd be curious to know... Is there different models (beside factory bios) ?

This is due to the new Titan kernel that the Nvidia engineer submitted. S/He stated that the best config for it will be SMX x 24. This is why this setting is preferred.
780    -l T12x24
Titan  -l T14x24
780Ti  -l T15x24

Please try all -H -C and -m settings before you write the changes off. This is KEY with this release.

I went from

Code:

-l Z14x24 -H 1 -C 1 -m 2

over to

Code:

-l T14x24 -H 1 -C 0 -m 0

(commit 114 to release) saw a very little kh/s change but found that the core could stably OC another 50MHz improving kh/s due to more OC head room

For scrypt on GTX TITAN i think this is little bit better:
-d 0 -i 0 -C 2 -m 1 -H 1 -l T14x24 (605khash/s @ 928Mhz)

[1714211328]

1714211328

[1714211328]

1714211328

[1714211328]

1714211328

[Mkilbride]

Can anyone help me with my GTX670 SLi's?

My old settings no longer work with the new version, just validate error after validate error.

cudaminer.exe -H 1 -m 1 -i 0 -l K7x32 -C 1 -o stratum+tcp://stratum-us.doge.hashfaster.com:3339 -O

I tried auto, that doesn't work either, and I tried the "Y" kernal, no go.

What should I do? The new version seems faster, so...

[cbuchner1]

Any ideas why both GPU's can't be used at the same time?

suspected nVidia driver bug. use two cudaminer instances. It's all in the README that no one bothers to read

[cbuchner1]

Autotune hotfix posted... this makes sure -l auto (or omission of -l) picks the advertised Kernels as described in the README.txt

[trell0z]

Also, what's up with people getting good results on their 780s using T/Z12x24 ? I get absolutely craptastic results with that!

That's like the 15x24 for some 780ti... yes me too I'd be curious to know... Is there different models (beside factory bios) ?

This is due to the new Titan kernel that the Nvidia engineer submitted. S/He stated that the best config for it will be SMX x 24. This is why this setting is preferred.
780    -l T12x24
Titan  -l T14x24
780Ti  -l T15x24

Please try all -H -C and -m settings before you write the changes off. This is KEY with this release.

I went from

Code:

-l Z14x24 -H 1 -C 1 -m 2

over to

Code:

-l T14x24 -H 1 -C 0 -m 0

(commit 114 to release) saw a very little kh/s change but found that the core could stably OC another 50MHz improving kh/s due to more OC head room

No matter what combination I use, the best I can get with T12x24 for my 780 is less than 1/3 of T12x20, ~170-185kh/s depending on settings.

[zer0k]

Any ideas why both GPU's can't be used at the same time?

suspected nVidia driver bug. use two cudaminer instances. It's all in the README that no one bothers to read

I am running two seperate instances
2 seperate dos windows. 2 seperate folders, 2 seperate commands d 0 and d 1

[Joohansson]

The "-l auto" does not pick the best settings for gtx670 (yes I have updated to latest patched version). The settings I've found in this thread gives much higher speeds, 300 compared to 200. Have not yet tried with my other cards.

[cbuchner1]

Any ideas why both GPU's can't be used at the same time?

suspected nVidia driver bug. use two cudaminer instances. It's all in the README that no one bothers to read

I am running two seperate instances
2 seperate dos windows. 2 seperate folders, 2 seperate commands d 0 and d 1

ok then I apologize. No idea what causes it, really.

[Mkilbride]

Your new version doesn't work for me at all, the fixed version - no matter what I do, it just closes automatically.

[cbuchner1]

The "-l auto" does not pick the best settings for gtx670 (yes I have updated to latest patched version). The settings I've found in this thread gives much higher speeds, 300 compared to 200. Have not yet tried with my other cards.

yes, autotune has its fair share of problems. most notably the Boost feature of Kepler cards can confuse it.

Christian

[trell0z]

Looking quite awesome on the overclocking front too, currently seems to be stable at 1241mhz core and default mem, at 1.137V! Working my my down to see where it crashes haha. Now at ~660kh/s

[cbuchner1]

Your new version doesn't work for me at all, the fixed version - no matter what I do, it just closes automatically.

I just tried mining with these binaries both under x86 and x64. Works for me.

if you run from a .bat file, add a pause statement at the end

[Stanr010]

Your new version doesn't work for me at all, the fixed version - no matter what I do, it just closes automatically.

I'd say double check your .bat config file. If it's shutting down instantly, that usually means it doesn't like something that it's trying to execute.  

[Mkilbride]

Since when was the pause command needed?

12-18 still seems to be the best version.

And that still didn't work.

Here is the thing, I've tried every .bat configuration known to miners and none work with the latest, but 12-18 works fine.

[manofcolombia]

Yay autotune sobered up
gave me K14x20 @325 khash on my 660 ti.
Still nothing beating K7x32 @340 khash

[cbuchner1]

The "-l auto" does not pick the best settings for gtx670 (yes I have updated to latest patched version). The settings I've found in this thread gives much higher speeds, 300 compared to 200. Have not yet tried with my other cards.

also browse the spreadsheets linked in the first posting. They will give you some tested configs (just beware of some differences between cudaminer releases).

[bugilt]

Sent 0.30926131 LTC To your motivation account.

Getting close to the hash rate 280x! Never thought you'd make it this far.

GTX780 650 kH/s watching videos at the same time.

[cbuchner1]

Getting close to the hash rate 280x! Never thought you'd make it this far.
GTX780 650 kH/s watching videos at the same time.

nVidia was helping ;-)

[Stanr010]

Getting close to the hash rate 280x! Never thought you'd make it this far.
GTX780 650 kH/s watching videos at the same time.

nVidia was helping ;-)

Gosh I'd think we should just donate a couple BTC to an Nvidia engineer or two and have them code Cudaminer from the ground up for their GPUs. With proper optimizations, I bet you 780's can reach the performance of 290's. 780 Ti's have already proved that they can play with 290x in mining too.

[cbuchner1]

Here is how launch configs from the cudaminer 2013-12-18 release translate to cudaminer 2014-02-02 release
to get equivalent performance. This can be handy if you find some older launch configs posted by others

 L b x w  ->                   sorry, legacy kernel was replaced by Fermi kernel. Autotune the F kernel.

 F b x w  ->  F b x w       ( no change to this one )

 K b x w  ->  k 4*b x w    (the previous K kernel is now named k and no. of blocks has to be quadrupled)

 T b x w ->  t 4*b x w     (the previous T kernel is now named t and no. of blocks has to be quadrupled)

 S b x w  ->                   Spinlock kernel is GONE.


These following kernels are new in the cudaminer 2014-02-02 release:  T (nVidia), K (nVidia), f (ported over from David Andersen's code).

Christian