arCUDAminer 64 bit v1.0c - PTS GPU miner for Windows

[archit]

Original topic over at:https://bitsharestalk.org/index.php?topic=2214.0

This doesn't work at all on AMD cards, if you are looking for that go here - https://bitsharestalk.org/index.php?topic=2296.0

Requirements:
Nvidia GPU with Compute capability higher than 20
CPU with AVX/SSE4 (Not tested with AMD)
Latest Nvidia driver

I have only included the binaries for now, I will release the source code after I get a donation of 10 PTS. The miner mines 500 rounds for user and then 15 rounds to support developer. Default -o is ypool.net. Use highperf if you have more than 1.2 GB of GPU memory. Please note that SLI support is not tested yet.

Usage:
arCUDAminer.exe -u archit.pts_1 -p x -m avx -g 2

-g User only if more than 1 GPU eg.2
-u Worker
-p Worker password
-m avx or sse4(lower caps)

Why this?

This doesn't work at all on AMD cards, if you are looking for that go here - https://bitsharestalk.org/index.php?topic=2296.0

Requirements:
Nvidia GPU with Compute capability higher than 20
CPU (Post 1.0c any processor works)
Latest Nvidia driver

I have only included the binaries for now, I will release the source code after I get a donation of 10 PTS. The miner mines 500 rounds for user and then 15 rounds to support developer. Default -o is ypool.net. Use highperf if you have more than 1.2 GB of GPU memory. Please note that SLI support is not tested yet.

Usage:
arCUDAminer.exe -u archit.pts_1 -p x -m avx -g 2

-g User only if more than 1 GPU eg.2
-u Worker
-p Worker password
-m avx or sse4 or sph(lower caps)
-o Address of pool

Why this?

• 64 bit
• extremly small fee
• Point to your pool of choice

Please donate on
PTS:Pn9b5y2FRpAn4V94gTLjxJ9eSQkeuuU6GP

https://www.dropbox.com/s/glecca0a5srhpzt/arCUDAminer1.0c.rar

How much you donate can be seen from the live workers page on ypool
If you get a malloc error, make sure you have more than 800 MB of free video memory[/list]

[1715501792]

1715501792

[1715501792]

1715501792

[1715501792]

1715501792

[Flashman]

Interesting, that's 2.0 up not 20 up right?

Seems like it should be 4xx upward apart from the 420

https://developer.nvidia.com/cuda-gpus

[archit]

Yes, that was a typo

[fran2k]

Nice to see some work on NVIDIA GPU Mining.

I would like to see some stuff working better over NVIDIA.

[n00b2013]

cool work ! btw what card ur using for that ?

[barkinos98]

YEAAHH
Should kick some ass of intel with the 780, knew i was right buying this!

[nirom]

So how much kh/s can I expect from my good ol' 9500 GT?

[glongsword]

My 2x GTX 465's can't mine with either the original or highperf version.  I've tried all combinations of SLI on/off, sse4 and avx, -g 1 and 2, and I always get "Initializing.  Device has 961875968 free of 1073741824 total bytes of memory.  Could not malloc dev_data (2)"

They both have 1GB of memory.

[Flashman]

So how much kh/s can I expect from my good ol' 9500 GT?

Won't work for this, those only have a compute capability of 1.0 .... otherwise I'd be trying it out, only have access to a 9600GT and a 280GT that needs a fix.

(Oh btw, balls to the wall overclocked on cgminer, I've only ever got about 30kh outa my 9600GT)

[Snard]

For 10 PTS you are release source to everyone or the donator?

[SnakeJG]

I get the following on a Quadro 1000M on 64 bit win 7:


╔══════════════════════════════════════════════════╗
║  arCUDAminer (v0.4)                              ║
║  author: jh,dga and ported by archit             ║
║  http://ypool.net                                ║
╚══════════════════════════════════════════════════╝
Launching miner...
Using 1 threads
Will mine 15 round after 500 to support developerUsing AVX
Initializing.  Device has 2084675584 free of -2147483648 total bytes of memory
Connected to server using x.pushthrough(xpt) protocol
New block data - height: 41634 tx count: 18
Error in kernel exec (30)
Could not memcpy dev_data (30)
Could not memcpy dev_data (30)
Could not memcpy dev_data (30)
Could not memcpy dev_data (30)

[archit]

For 10 PTS you are release source to everyone or the donator?

donator

Do you have the CUDA toolkit installed?

[illodin]

Tried this with my GTX 660, and apparently I'm earning about $0.01 / hour with this. Cool? And oh, ypool's minimum amount to withdraw is 1 PTS, so it will take about 100 days of mining before I can withdraw my 1 PTS heh.

[SnakeJG]

I have the latest drivers, but I haven't downloaded the toolkit. I was able to run cudaminer on the card, so CUDA works in some way, but I will download the toolkit (800+ MB!) and get back to you if that fixes it or not.


EDIT: downloading the toolkit didn't help, but it also didn't install all of the components because I don't have MS Studio.

[archit]

I have the latest drivers, but I haven't downloaded the toolkit. I was able to run cudaminer on the card, so CUDA works in some way, but I will download the toolkit (800+ MB!) and get back to you if that fixes it or not.


EDIT: downloading the toolkit didn't help, but it also didn't install all of the components because I don't have MS Studio.

It seems that Quadro 1000M isn't working since someone else also reported the exact same error

[glongsword]

Before you donate, be aware of this earlier post of a CUDA PTS miner:

https://bitcointalk.org/index.php?topic=404915.0

[Supercomputing]

Before you donate, be aware of this earlier post of a CUDA PTS miner:

https://bitcointalk.org/index.php?topic=404915.0

Windows version has more features than the open source Linux version. Also, I think that the OP is offering the Visual Studio project ready to compile.

[fragamemnon]

Reporting issues.
I'm mining @ ypool on a GTX 680; running Windows 8.
Sharerate is being calculated precisely (~100sh/h)
After the first donation round, when it 'switches' back to user mining, ypool's counter for my account stops taking into account shares submitted past that point. If I restart the miner, all is fine again.

Someone able to check this? I have the feeling it doesn't switch back to the user defined credentials after donation round.

[Mapin]

Are those spikes for the GPU utilization expected or is sth bugged?

http://thumbnails111.imagebam.com/30114/d45ebc301135806.jpg

[archit]

Use the newer version from bitsharestalk forum. I forgot to update the thread over here

[Mapin]

If this was an answer to the spikes question: arCUDA 1.0 still gives me exactly the same gpu spikes.

EDIT: After initializing I also get a message like -965346328 free of 0 total bytes of memory.

[archit]

If this was an answer to the spikes question: arCUDA 1.0 still gives me exactly the same gpu spikes.

EDIT: After initializing I also get a message like -965346328 free of 0 total bytes of memory.

What parameters? And can you copy the all the text that is written out

[Gwynbleidd]

Around 800CPM with 680 gtx  Want more!

[Mapin]

Thx for careing archit =)

http://thumbnails112.imagebam.com/30142/e18196301418115.jpg

Tried non-highperf and avx too for the same results. (regarding the spikes) Had it once starting with a little less spikes too for maybe 1 min achieved ~20-30% more collisions.

Running on a 750M (Kepler) with 4GB.

[archit]

Thx for careing archit =)



Tried non-highperf and avx too for the same results. (regarding the spikes) Had it once starting with a little less spikes too for maybe 1 min achieved ~20-30% more collisions.

Running on a 750M (Kepler) with 4GB.

That would be fixed soon.

EDIT: Should be fixed with 1.0c

[Mapin]

First of all, thx for the update on this forum, and for your work in general.

Results with 1.0c are ~10%up from 1.0b GPU is constantly at 90%.

BUT:
At startup i still get the same -"99847394853" of 0 bytes-message and the FB is still spiking with does not really look optimized or is there an intrinsic reason for this to happen?

http://thumbnails101.imagebam.com/30199/f11ba7301980006.jpg

[archit]

First of all, thx for the update on this forum, and for your work in general.

Results with 1.0c are ~10%up from 1.0b GPU is constantly at 90%.

BUT:
At startup i still get the same -"99847394853" of 0 bytes-message and the FB is still spiking with does not really look optimized or is there an intrinsic reason for this to happen?

That should be because of a printing mistake I did. A new version would coem soon pushing GPU usage to 100% but I still haven't been able to figure out your spikes

[Mapin]

Any kind of debugging Information I might sent to you?

[archit]

Any kind of debugging Information I might sent to you?

Nope, as I have no idea what's causing these spikes for you

[ballot]

btw u decreased pts's price 50% just for 3500$ bounty 

[archit]

btw u decreased pts's price 50% just for 3500$ bounty 

Don't blame it on os

[NineLives]

Trying the miner.

Keeps crashing is i use anything but sse4.

And using sse4 gives me this:  

Code:

Could not malloc dev_data (2)

[archit]

Trying the miner.

Keeps crashing is i use anything but sse4.

And using sse4 gives me this:  

Code:

Could not malloc dev_data (2)

What about sph?

[bittick]

ok, whats the best price/CPM ratio card on the market atm?

[Flashman]

Just had GT-440 fall in my lap, what should I expect out of it? .... in comparison to a middling CPU say.