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

[cbuchner1]

Hi there! Can you please plug your figures into this table for nvidia performance?
http://yacoinwiki.tk/index.php/Mining_Hardware_Comparison

It's a little early for posting this data publicly, as the feature is still being developed and there are no official Windows binaries available yet.

I did however add the scrypt-jane related google docs spreadsheet into the first posting.

[cbuchner1]

A GT 640 (GK107) 4GB at stock clocks will do something in the range of 1.65 kHash/s.

I wonder how this one performs with an N factor of 15.

The best case is that it achieves exactly half the hash rate as with N=14. Why? because for N=15 it's exactly twice the amount of work. This will affect CPUs likewise.

The usual case for most GPU models will be that the performance degrades by more than half, because the occupancy of the CUDA cores goes down (too many cores for fewer hashes to be computed simultaneously given the given available memory). Cards with 1GB and 2GB will be hit the hardest...

Cards with 4GB should barely see an impact for N = 14 --> 15

For cards with lots spare GPU cores (say, a GTX 780 or better) we can cut memory requirements in half and increase compute requirements instead (LOOKUP_GAP). It's on my TODO list.

[ManIkWeet]

A GT 640 (GK107) 4GB at stock clocks will do something in the range of 1.65 kHash/s.

I wonder how this one performs with an N factor of 15.

The best case is that it achieves exactly half the hash rate as with N=14. Why? because for N=15 it's exactly twice the amount of work. This will affect CPUs likewise.

The usual case for most GPU models will be that the performance degrades by more than half, because the occupancy of the CUDA cores goes down (too many cores for fewer hashes to be computed simultaneously given the given available memory). Cards with 1GB and 2GB will be hit the hardest...

Cards with 4GB should barely see an impact for N = 14 --> 15

For cards with lots spare GPU cores (say, a GTX 780 or better) we can cut memory requirements in half and increase compute requirements instead (LOOKUP_GAP). It's on my TODO list.

Nice, I have exactly that card... :3

[lolobubu]

HI everyone im new to mining with cudaminer. I currently am mining scrypt based coins with an MSI gtx 660 ti power edition and
an older gtx 460. I'm using some OC with afterburner and currently these are my stats.

http://gyazo.com/730066456565bff6c382f23168451bef
http://gyazo.com/3997db1d7d191733e257301057376da1
http://gyazo.com/4c187b5f53c44c9cb9876eb1786d654e

First of al can i get more hashspeed ? and is there an other kind of mining more profitable than scrypt mining for my nvidia cards ?

my flags -d 0,1 -C 1,2 -l K112x2,F52x2 -i 0,0

thanks for your time.

[cbuchner1]

and is there an other kind of mining more profitable than scrypt mining for my nvidia cards ?

well yes, scrypt-jane is currently about twice as profitable. It requires a binary built from the current cudaminer github repository.

Expect 3.2-3.5 kHash/s from your GTX 660 Ti. No idea about your GTX 460. Maybe the total is around 5 kHash? Look up on  the yac.coinmine.pl stats pages how much BTC per day that yields at current YAC->BTC exchange rates

Christian

[cbuchner1]

Some new card stats:

MSI GTX 650 OC (2 SMX, 2 GB GDDR5 RAM):            1.95 kHash/s at -l K15x1 -C 1
Gigabyte GTX 650 Ti OC (4 SMX, 2 GB GDDR5 RAM)   2.20 kHash/s at -l K15x1 -C 1

the latter card costs marginally more and would have reserves for some LOOKUP_GAP fun (making this card last through NFactor 15 I hope).

Both cards were tested on Linux and have a factory overclock. Temperatures stay reasonable at 68C, unlike some low profile GT 640 (GK107) or single slot GT 630 cards (Fermi) I have tested (->99 deg C!)

The main drawback is that these are double slot cards. So you need risers to run 5 cards on a mobo. If you care more about densely packed hashing power, then get GTX 660 Ti or GTX 760 cards instead. These will cost twice as much however.

Today's improvements:

I fixed a serious problem causing validation errors in Kepler and Test (X-) kernels with -C 1 just now. I also redesigned some of the timer code for interactive mode. It will take less CPU now.

It may still make sense to increase the default batch size -b 1024 to something like -b 4096 or -b 8192 when running interactive mode with scrypt-jane. This costs a bit of display smoothness, but adds more hash/s.

Christian

[psc]

It may still make sense to increase the default batch size -b 1024 to something like -b 4096 or -b 8192 when running interactive mode with scrypt-jane. This costs a bit of display smoothness, but adds more hash/s.

Thanks for the tip ! - increased my 770M hashrate from 1.1 to 1.33 on windoze (-b 4096)

[Flo354]

Noob question but what is the utility of the -b option and what it mean ?

With my GTX 770 4Gb, it is necessary to use the -b option, and if yes, how can i know what is the best value ?

[cbuchner1]

Noob question but what is the utility of the -b option and what is mean ?

With my GTX 770 4Gb, it is necessary to use the -b option, and if yes, how can i know what is the best value

The best (in terms of fastest) is to use the same value as N. N is currently 32768 for Yacoin. Let me explain what is going on:

scrypt-jane is running a for loop like this (which will take a loong time to complete, in the order of quarter to half a second. The GPU is fully unresponsive during that time.

   for (i=0; i < 32768; ++i) { do a lot of work and memory access }  // run once

-b 1024 instead runs 32 shorter for loops like this, with small pauses inbetween when interactive mode is enabled. This is the same workload as regular scrypt hashing per loop. This is why I made this the default.

   for (i=0; i < 1024; ++i) { do a lot of work and memory access }  // run 32 times

-b 4096 runs 8 for loops like this, which is an OK intermediate between the two extremes. This might be a good compromise for display smoothness (if you're not planning on watching movies that is).

   for (i=0; i < 4096; ++i) { do a lot of work and memory access }  // run 8 times


In the future, interactive mode may auto-determine the batch size to hit a desired target frame rate _exactly_.

Christian

[Flo354]

Noob question but what is the utility of the -b option and what is mean ?

With my GTX 770 4Gb, it is necessary to use the -b option, and if yes, how can i know what is the best value

The best (in terms of fastest) is to use the same value as N. N is currently 32768 for Yacoin. Let me explain what is going on:

scrypt-jane is running a for loop like this (which will take a loong time to complete, in the order of quarter to half a second.

   (i=0; i < 32768; ++i) { do a lot of work and memory access }

-b 1024 instead runs 32 shorter for loops like this, with small pauses inbetween when interactive mode is enabled. This is the same workload as regular scrypt-jane hashing per loop. This is why I made this the default.

   (i=0; i < 1024; ++i) { do a lot of work and memory access }

-b 4096 runs 8 for loops like this, which is an OK intermediate between the two extremes. This might be a good compromise for display smoothness if you're not planning on watching movies that is.

   (i=0; i < 4096; ++i) { do a lot of work and memory access }


In the future, interactive mode may auto-determine the batch size to hit a desired target frame rate _exactly_.

Christian

Okay, thank you I understand !

[lolobubu]

and is there an other kind of mining more profitable than scrypt mining for my nvidia cards ?

well yes, scrypt-jane is currently about twice as profitable. It requires a binary built from the current cudaminer github repository.

Expect 3.2-3.5 kHash/s from your GTX 660 Ti. No idea about your GTX 460. Maybe the total is around 5 kHash? Look up on  the yac.coinmine.pl stats pages how much BTC per day that yields at current YAC->BTC exchange rates

Christian

hey thanks for your time,
where exactly can i find the binary ? and what do with it? im a little new in this enviroment what do i have to change in my .bat to run it propperly

thanks will donate you some YAC as soon as i mine some.

[cbuchner1]

where exactly can i find the binary ? and what do with it? im a little new in this enviroment what do i have to change in my .bat to run it propperly

thanks will donate you some YAC as soon as i mine some.

I am not publishing binaries yet because the feature "is not done yet". I like my features well-done, not medium rare
Maybe you will find someone in this thread who is willing to share a binary that he compiled from recent sources.

If you are familiar with Visual C++ 2010/2012/2013 (Windows) or gcc and autotools (Linux) you can also try yourself.

Christian

[lolobubu]

where exactly can i find the binary ? and what do with it? im a little new in this enviroment what do i have to change in my .bat to run it propperly

thanks will donate you some YAC as soon as i mine some.

I am not publishing binaries yet because the feature "is not done yet". I like my features well-done, not medium rare
Maybe you will find someone in this thread who is willing to share a binary that he compiled from recent sources.

If you are familiar with Visual C++ 2010/2012/2013 (Windows) or gcc and autotools (Linux) you can also try yourself.

Christian

Oh i see, im not yet a pro at .NET and C++  anyone else here can help me out with a compiled working binary ?
will donate some YAC as soon as i can make some

[ManIkWeet]

Quite easy:
Get Visual Studio Pro 2010 (Trial if you must)
Get Cuda SDK 5.5
Get the source code from github (google cudaminer)
Extract source code in it's own folder
Get libraries from first post by Christian (almost 50mb)
Extract libraries to the same folder that the source code folder is in, so not to the source code folder but one up.
Open the .sln file in the source code with Visual Studio.
Select which version you want to compile, and compile it.
If it errors:

Go to your Microsoft Visual Studio 10.0\VC\bin folder and rename or relocate cvtres.exe and do the same with another cvtres.exe under Microsoft Visual Studio 10.0\VC\bin\amd64, then run Visual Studio 2010 and try compiling.

If it doesn't work, just change everything back.

[psc]

Oh i see, im not yet a pro at .NET and C++  anyone else here can help me out with a compiled working binary ?
will donate some YAC as soon as i can make some

Windows one was indicated here https://bitcointalk.org/index.php?topic=167229.msg4427833#msg4427833

[lolobubu]

Ok i seem to have been able to make everything work with my msi 660 ti PE and 460gtx
http://gyazo.com/009de91760e8553055556d7229edfcc4
-d 0,1 -C 0,0 -l K14x1,F7x1 -b 8192,8192 -i 0,0

these numbers seem right to you ppl ?

[skyvahaerie]

Ok, i finally got scrypt-jane running, but im not sure if its working properly. These are the parameters im running: cudaminer.exe -a scrypt-jane -b 8192 -i 1 -o stratum+tcp://yac.coinmine.pl:9088 -O user.1:password -C 2
http://s1.directupload.net/images/140116/v9yktimq.png

Is it normal that there are several minutes between the valid hashes? Or is my gpu just to slow due to the fast block generation of YAC?
I also got the problem that if i try to set the launch configuration manually, cudaminer terminates with a generic error message.

Any help is appreciated.

Thanks in advance.

[manofcolombia]

Ok i seem to have been able to make everything work with my msi 660 ti PE and 460gtx
http://gyazo.com/009de91760e8553055556d7229edfcc4
-d 0,1 -C 0,0 -l K14x1,F7x1 -b 8192,8192 -i 0,0

these numbers seem right to you ppl ?

My ti (Same one) gets 2.30 with 13x1

14x1 crashes my drivers I would think this is due to dual 1440p monitors its running

[lolobubu]

Ok i seem to have been able to make everything work with my msi 660 ti PE and 460gtx
http://gyazo.com/009de91760e8553055556d7229edfcc4
-d 0,1 -C 0,0 -l K14x1,F7x1 -b 8192,8192 -i 0,0

these numbers seem right to you ppl ?

My ti (Same one) gets 2.30 with 13x1

14x1 crashes my drivers I would think this is due to dual 1440p monitors its running

thanks i did get 0.20 Khs increase on the 660

[cdogster]

It may still make sense to increase the default batch size -b 1024 to something like -b 4096 or -b 8192 when running interactive mode with scrypt-jane. This costs a bit of display smoothness, but adds more hash/s.

Thanks for the tip ! - increased my 770M hashrate from 1.1 to 1.33 on windoze (-b 4096)

Using -b 8192 I was able to increase my hashrate from 1.11 to 1.26 on my GTX 650.  Doubling to 16384 didn't seem to have an effect.