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

[djm34]

I can get 4 motherboards for that

What's the point then?

Who buys these things???

People bad in math, I guess

[cbuchner1]

how about this 4x port extender?

http://www.aliexpress.com/store/product/PCI-e-PCIe-Express1X-to-4port-1X-multiplier-switch-riser-card-adapter-cable-for-diy-bitcoin/825827_1444567810.html

much cheaper. But only PCIe 1x times 4.

[bigjme]

I can't see the point in them. Their so close together you would still need extra extensions. So their of if you need long lengths but other then that,  a waste

[djm34]

I can't see the point in them. Their so close together you would still need extra extensions. So their of if you need long lengths but other then that,  a waste

Combined with USB risers would be ok I guess (however I don't understand how you plug it to the motherboard, using one of the pcie1 ?)

[deadthings]

there are several proprietary stratum extensions made by the
1gh pool that are a bit messy to merge. 

Will the miner you're working on be compatible with regular stratum like heavycoinpool.com?

[reznor]

I want new cudaminer for Maxwell , i think this card ( Nvidia 750ti ) not go at 100% , i think this GPU is more fast and go up to 400 kh/s or more , but cudaminer is not optimizer .

how would you know.

nVidia has their engineers looking into mining performance and they haven't come up with any better code yet.

Christian i read some post over nvidia dev zone. Do you think nvidia devs will release a driver with some potential hashrate increase using this GTX750 architecture?

By the way whats the difference of using GTX 750ti 1GB and GTX 750ti 2GB , does it affect the hashrate when using more vram?

[ivanlabrie]

I want new cudaminer for Maxwell , i think this card ( Nvidia 750ti ) not go at 100% , i think this GPU is more fast and go up to 400 kh/s or more , but cudaminer is not optimizer .

how would you know.

nVidia has their engineers looking into mining performance and they haven't come up with any better code yet.

Christian i read some post over nvidia dev zone. Do you think nvidia devs will release a driver with some potential hashrate increase using this GTX750 architecture?

By the way whats the difference of using GTX 750ti 1GB and GTX 750ti 2GB , does it affect the hashrate when using more vram?

You want 2gb for the alternate algorithm coins, scrypt won't make a difference, but that's a bit shortsighted.

[cbuchner1]

there are several proprietary stratum extensions made by the
1gh pool that are a bit messy to merge.  

Will the miner you're working on be compatible with regular stratum like heavycoinpool.com?

I think the heavycoinpool stratum code will differ.

We can insert our extra GPU code in pretty much any cpu miner code very quickly.

Christian

[Number6]

Well so far I am not too impressed with the GTX 750ti cards, one of mine already went belly up in under 2 weeks use. Anyone else have any issues with them going dead yet, or is this just an outlier?

[RbelMonstr]

Well so far I am not too impressed with the GTX 750ti cards, one of mine already went belly up in under 2 weeks use. Anyone else have any issues with them going dead yet, or is this just an outlier?

It could (hopefully) be a bad apple. Can you RMA it, or did it go up in flames?

[deadthings]

Well so far I am not too impressed with the GTX 750ti cards, one of mine already went belly up in under 2 weeks use. Anyone else have any issues with them going dead yet, or is this just an outlier?

It could (hopefully) be a bad apple. Can you RMA it, or did it go up in flames?

No matter what, after 2 weeks, it will still be RMA'able, even if you OC'd its tits off.

[phm]

just showing where we are with Heavycoin at the moment... 2400 kHash/s per GTX 780 Ti
at only just above half the TDP power of 250 Watts...

Code:

2014-03-14 23:02:24] Starting Stratum on stratum+tcp://hvcpool.1gh.com:5333
[2014-03-14 23:02:24] 3 miner threads started, using 'heavy' algorithm.
[2014-03-14 23:02:26] accepted: 1/1 (100.00%), 5470 khash/s (yay!!!)
[2014-03-14 23:02:29] accepted: 2/2 (100.00%), 5933 khash/s (yay!!!)
[2014-03-14 23:02:42] accepted: 3/3 (100.00%), 6777 khash/s (yay!!!)
[2014-03-14 23:02:45] accepted: 4/4 (100.00%), 6755 khash/s (yay!!!)
[2014-03-14 23:02:48] accepted: 5/5 (100.00%), 6778 khash/s (yay!!!)
[2014-03-14 23:02:50] accepted: 6/6 (100.00%), 7294 khash/s (yay!!!)
[2014-03-14 23:02:50] accepted: 7/7 (100.00%), 7278 khash/s (yay!!!)
[2014-03-14 23:02:54] accepted: 8/8 (100.00%), 7267 khash/s (yay!!!)
[2014-03-14 23:02:55] accepted: 9/9 (100.00%), 7204 khash/s (yay!!!)
[2014-03-14 23:03:00] accepted: 10/10 (100.00%), 7288 khash/s (yay!!!)
[2014-03-14 23:03:15] accepted: 11/11 (100.00%), 7295 khash/s (yay!!!)
[2014-03-14 23:03:19] accepted: 12/12 (100.00%), 7311 khash/s (yay!!!)
[2014-03-14 23:03:24] accepted: 13/13 (100.00%), 7243 khash/s (yay!!!)
[2014-03-14 23:03:24] accepted: 14/14 (100.00%), 7243 khash/s (yay!!!)
[2014-03-14 23:03:25] accepted: 15/15 (100.00%), 7043 khash/s (yay!!!)
[2014-03-14 23:03:25] accepted: 16/16 (100.00%), 7043 khash/s (yay!!!)
[2014-03-14 23:03:28] accepted: 17/17 (100.00%), 7260 khash/s (yay!!!)
[2014-03-14 23:03:32] accepted: 18/18 (100.00%), 7279 khash/s (yay!!!)
[2014-03-14 23:03:34] accepted: 19/19 (100.00%), 7251 khash/s (yay!!!)

|===============================+======================+======================|
|   0  GeForce GTX 780 Ti  Off  | 0000:01:00.0     Off |                  N/A |
| 49%   81C    P0   137W / 350W |     94MiB /  3071MiB |     73%      Default |
+-------------------------------+----------------------+----------------------+
|   1  GeForce GTX 780 Ti  Off  | 0000:02:00.0     Off |                  N/A |
| 43%   79C    P0   136W / 350W |    191MiB /  3071MiB |     80%      Default |
+-------------------------------+----------------------+----------------------+
|   2  GeForce GTX 780 Ti  Off  | 0000:03:00.0     Off |                  N/A |
| 47%   80C    P0   134W / 350W |    101MiB /  3071MiB |     75%      Default |
+-------------------------------+----------------------+----------------------+
                                                                               

NOTE: this isn't cudaminer. It's a separate project, so don't expect to find this in the cudaminer trunk.
It might be merged in later on, but there are several proprietary stratum extensions made by the
1gh pool that are a bit messy to merge.  This is the main reason we started with a fresh fork.

Very nice numbers, I guess I share mine as well now, they are for 4 x R9 290:

Code:

[2014-03-15 06:43:39] accepted: 122/227 (53.74%), 15721 khash/s (yay!!!)
[2014-03-15 06:43:40] thread 13: 48234496 hashes, 1017 khash/s
[2014-03-15 06:43:40] accepted: 123/228 (53.95%), 15827 khash/s (yay!!!)
[2014-03-15 06:43:41] thread 1: 17825792 hashes, 1052 khash/s
[2014-03-15 06:43:42] thread 9: 58720256 hashes, 957.14 khash/s
[2014-03-15 06:43:42] accepted: 124/229 (54.15%), 15855 khash/s (yay!!!)
[2014-03-15 06:43:44] thread 5: 62914560 hashes, 971.25 khash/s
[2014-03-15 06:43:45] thread 2: 4194304 hashes, 664.38 khash/s
[2014-03-15 06:43:45] thread 9: 2097152 hashes, 614.34 khash/s
[2014-03-15 06:43:46] accepted: 125/230 (54.35%), 15204 khash/s (yay!!!)
[2014-03-15 06:43:46] accepted: 126/231 (54.55%), 15204 khash/s (yay!!!)
[2014-03-15 06:43:50] thread 0: 58720256 hashes, 1016 khash/s
[2014-03-15 06:43:50] thread 8: 11534336 hashes, 942.77 khash/s
[2014-03-15 06:43:51] accepted: 127/232 (54.74%), 15207 khash/s (yay!!!)
[2014-03-15 06:43:52] thread 8: 1048576 hashes, 495.55 khash/s

I think it still has some bugs, the pool rejects my shares when diff is low, but after target switches to lower value everything works fine. Also not sure if these numbers are correct, the pool shows 13-14 MH/s.

I'm going to open source this miner when your is out, good luck with more optimizations. Power usage on R9 290 when mining this is about 100W/card.

[gfgrimm]

Man that is one expensive riser but it sure sounds awesome.

What is the difference between atx version and external version? Don't they both need a regular power supply to run?

I am thinking hard about them. Seems a lot simpler to hook up and run. I am just so confused about do I need a powered or non powered riser. Running at this or that speed. This makes it seem like it is just a "plug and play" type version.

Simpler is better for me.

ATX means you can use a 12V/5V FDD connector from an ATX PSU...external version means that you only need to provide 12V with the same connector (I think). I believe the details are in the datasheet on the product page.

[gfgrimm]

I can get 4 motherboards for that

What's the point then?

Who buys these things???

People bad in math, I guess

I'm paying about $190 for the sku that allows me to directly connect 4 GPUs and use an ATX PSU to power the adapter boards.

My math says that for $450 + cost of GPUs, I can get 6 cards running just by running 6 cards per system with x1 risers.
My math says that for $820 + cost of GPUs, I can get 12 cards running on one system by using two of the x4 splitters with a single system.

Tells me I'm saving $80 per 12 GPU's I want to scale.

Probably close enough to not matter, but I just think it's cool to get 12 cards working on one system. Plus, the supplier said that volume discounts are applicable.

To be honest...I could probably design a clone of their host and adapter boards pretty easily...I wouldn't have a problem having PLX sending me their silicon. If I was serious about mining, that's probably what I'd do. If I didn't have a nice office job, it's probably something that would be fun to try. It's just an over-glorified high speed inter-poser board. I'm sure PLX has a reference design that takes the guesswork out of the schematic and layout. It may be interesting to do a BOM scrub of the host/adapter boards, contact the suppliers, and see how much it costs in materials. If only I had the time. If you were semi-serious about GPU mining and wanted to fill a small warehouse, designing your own PCIe extender assembly with PLX silicon would probably be a good move given then triviality of designing the host/adapter boards.

Christian, that other x1 splitter you're linking to splits the bandwidth of a single x1 PCIe link between four cards. Given that cudaminer chokes an x1 slot with one card already, splitting that four ways sounds like a horrible idea. From what I can tell running one instance of cudaminer for a 750 Ti with 5x24 config uses about 68% of a PCIe 2.0 x1 interface bandwidth. Of course, if you find a way to severely reduce the BW required by Cudaminer, then yes, that cheaper splitter could work just fine.

The PLX based splitter that I linked two is actually an x4 to four x1 bridge. Big difference.

I'm doing all of this more just for fun than to try and make a buck. I consider all money spent on anything related to mining basically lost.

[antonio8]

I can get 4 motherboards for that

What's the point then?

Who buys these things???

People bad in math, I guess

I'm paying about $190 for the sku that allows me to directly connect 4 GPUs and use an ATX PSU to power the adapter boards.

My math says that for $450 + cost of GPUs, I can get 6 cards running just by running 6 cards per system with x1 risers.
My math says that for $820 + cost of GPUs, I can get 12 cards running on one system by using two of the x4 splitters with a single system.

Tells me I'm saving $80 per 12 GPU's I want to scale.

Probably close enough to not matter, but I just think it's cool to get 12 cards working on one system. Plus, the supplier said that volume discounts are applicable.

To be honest...I could probably design a clone of their host and adapter boards pretty easily...I wouldn't have a problem having PLX sending me their silicon. If I was serious about mining, that's probably what I'd do. If I didn't have a nice office job, it's probably something that would be fun to try. It's just an over-glorified high speed inter-poser board. I'm sure PLX has a reference design that takes the guesswork out of the schematic and layout. It may be interesting to do a BOM scrub of the host/adapter boards, contact the suppliers, and see how much it costs in materials. If only I had the time. If you were semi-serious about GPU mining and wanted to fill a small warehouse, designing your own PCIe extender assembly with PLX silicon would probably be a good move given then triviality of designing the host/adapter boards.

Christian, that other x1 splitter you're linking to splits the bandwidth of a single x1 PCIe link between four cards. Given that cudaminer chokes an x1 slot with one card already, splitting that four ways sounds like a horrible idea. From what I can tell running one instance of cudaminer for a 750 Ti with 5x24 config uses about 68% of a PCIe 2.0 x1 interface bandwidth. Of course, if you find a way to severely reduce the BW required by Cudaminer, then yes, that cheaper splitter could work just fine.

The PLX based splitter that I linked two is actually an x4 to four x1 bridge. Big difference.

I'm doing all of this more just for fun than to try and make a buck. I consider all money spent on anything related to mining basically lost.

Does this trick windows into thinking there is only 1 gpu or will it read 4 gpus?

Only reason asking I can see having 2 of these for a total of  8 750ti's. Plus keeping 1 main card in my rig to run my 3 monitors and still mine like I am currently doing. That is of course if Windows 7 would read 3 cards total or 9 cards total. I know Windows 7 Won't run 9 cards.

[becker]

I can get 4 motherboards for that

What's the point then?

Who buys these things???

They are for rich people like bon jovi.

[gfgrimm]

I can get 4 motherboards for that

What's the point then?

Who buys these things???

People bad in math, I guess

I'm paying about $190 for the sku that allows me to directly connect 4 GPUs and use an ATX PSU to power the adapter boards.

My math says that for $450 + cost of GPUs, I can get 6 cards running just by running 6 cards per system with x1 risers.
My math says that for $820 + cost of GPUs, I can get 12 cards running on one system by using two of the x4 splitters with a single system.

Tells me I'm saving $80 per 12 GPU's I want to scale.

Probably close enough to not matter, but I just think it's cool to get 12 cards working on one system. Plus, the supplier said that volume discounts are applicable.

To be honest...I could probably design a clone of their host and adapter boards pretty easily...I wouldn't have a problem having PLX sending me their silicon. If I was serious about mining, that's probably what I'd do. If I didn't have a nice office job, it's probably something that would be fun to try. It's just an over-glorified high speed inter-poser board. I'm sure PLX has a reference design that takes the guesswork out of the schematic and layout. It may be interesting to do a BOM scrub of the host/adapter boards, contact the suppliers, and see how much it costs in materials. If only I had the time. If you were semi-serious about GPU mining and wanted to fill a small warehouse, designing your own PCIe extender assembly with PLX silicon would probably be a good move given then triviality of designing the host/adapter boards.

Christian, that other x1 splitter you're linking to splits the bandwidth of a single x1 PCIe link between four cards. Given that cudaminer chokes an x1 slot with one card already, splitting that four ways sounds like a horrible idea. From what I can tell running one instance of cudaminer for a 750 Ti with 5x24 config uses about 68% of a PCIe 2.0 x1 interface bandwidth. Of course, if you find a way to severely reduce the BW required by Cudaminer, then yes, that cheaper splitter could work just fine.

The PLX based splitter that I linked two is actually an x4 to four x1 bridge. Big difference.

I'm doing all of this more just for fun than to try and make a buck. I consider all money spent on anything related to mining basically lost.

Does this trick windows into thinking there is only 1 gpu or will it read 4 gpus?

Only reason asking I can see having 2 of these for a total of  8 750ti's. Plus keeping 1 main card in my rig to run my 3 monitors and still mine like I am currently doing. That is of course if Windows 7 would read 3 cards total or 9 cards total. I know Windows 7 Won't run 9 cards.

Windows will see a PLX PCIe bridge and then see the four cards connected to that bridge. We'll find out soon how the operating systems affect the qty of cards to have drivers loaded. I'm too lazy to try and look it up because there are too many people that are on these forums that are full of crap when they cite limitations of this or that. The extender is in the mail and I've got 9 cards at my disposal so we shall see.

I assume I'll run into a BIOS issue before anything else. Likely run out of BIOS PCIe address space. I'm not too familiar with this stuff.

[becker]

Can I make a humble cudaminer request here?

Okay, Thanks...

Would it be possible, upon error, that cudaminer would automatically exit the terminal?

I ask this, because I can get cudaminer running my 750ti's for hours with higher settings, and not hot, but it will randomly start scrolling erros, stating error in salsa, or sha256, etc... Now, I can lower the settings, and lose like 20 KH/s / card  and run stable. But the error, takes awhile, and well, it'd be simple to just have windows scheduler, set to launch cudaminer anytime a terminal is not active for it, and it would just check like every minute.

So, if cudaminer exited on error, I could get more Khash, and only lose a minute of hashing every few hours.

Thanks!!

[cbuchner1]

So, if cudaminer exited on error, I could get more Khash, and only lose a minute of hashing every few hours.

Thanks!!

it would be easy to add an exit() statement into the error checking macro.   not sure if that would exit cleanly without crashing the driver though.

But something like this should be feasible, e.g. setting a global error condition flag that makes the main thread initiate a shutdown.

Christian

[cbuchner1]

Very nice numbers, I guess I share mine as well now, they are for 4 x R9 290:

Code:

[2014-03-15 06:43:40] accepted: 123/228 (53.95%), 15827 khash/s (yay!!!)
[2014-03-15 06:43:42] accepted: 124/229 (54.15%), 15855 khash/s (yay!!!)
[2014-03-15 06:43:46] accepted: 125/230 (54.35%), 15204 khash/s (yay!!!)
[2014-03-15 06:43:46] accepted: 126/231 (54.55%), 15204 khash/s (yay!!!)
[2014-03-15 06:43:51] accepted: 127/232 (54.74%), 15207 khash/s (yay!!!)

I think it still has some bugs, the pool rejects my shares when diff is low, but after target switches to lower value everything works fine. Also not sure if these numbers are correct, the pool shows 13-14 MH/s.

I'm going to open source this miner when your is out, good luck with more optimizations. Power usage on R9 290 when mining this is about 100W/card.

Great effort there too. We see stratum issues as well - need to start the miner a couple of times before it starts smoothly

We have some great optimization ideas in mind, due to the way the coin was designed. Today is optimization day.
Factor 2-3 more speed should be easy to accomplish for us.

Christian