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

[yellowduck2]

I think it's inevitable that in some point in time Nvidia and AMD will actually start hiring people to optimize their performance on cryptos. If mining is contributing so much to their sales that they are releasing new cards obviously optimized for mining (AMD Tonga, probably 750ti as well) then they can sure as hell hire some developers as well. Maybe they will do it anonymously or maybe even out loud but I am pretty sure it's going to happen.

The big scrypt crypto boom (LiteCoin, DogeCoin) is over. This is when nVidia actually had engineers look into scrypt performance. cudaMiner even incorporates some performance enhancements from nVidia engineers.

Haha nice   Cant say I am suprised though as AMD's market share went from 40% to probably 70-80% in Q4 and Q1
Yeah, that boom is over for now but as scrypt is taking it's last breaths we are entering a new era with new algos emerging so all the more reasons for optimization. Nothing is for sure but I higly doubt that was the laast mining boom, there is just too much room to grow for cryptos  
Christian you should ask Nvidia for endorsment   If there was Nvidia miner for XMR,BBR and every other new algo out there they would probably sell thousands of cards for mining every week  

maybe Christian should get commission for every gtx 750 ti sold worldwide. Motivation right there. 1 million dollar annual income should be easy. Forget mining. Aim for Nvidia Chief of Crytocurrencies

[1714139444]

1714139444

[1714139444]

1714139444

[1714139444]

1714139444

[cbuchner1]

maybe Christian should get commission for every gtx 750 ti sold worldwide. Motivation right there. 1 million dollar annual income should be easy.

huh? I don't think nVidia shareholders would like this much.

[yellowduck2]

maybe Christian should get commission for every gtx 750 ti sold worldwide. Motivation right there. 1 million dollar annual income should be easy.

huh? I don't think nVidia shareholders would like this much.

It's only 1 model. Commission 0.01% - 1%. I think it's a good deal for them since it will lead them new sales. Considering mining boom for Nvidia , Even 0.01% commission will be a lot, possibly stack up to 1 million dollars in a year. It's possible that alt coin mining (as a whole) might over take SHA mining or even become bigger in the future.

[S_tring]

I made some changes to my profit calc this weekend. So here it is:



v0.4: https://github.com/KBomba/cuda-profit-calc/releases/tag/v0.4b
Changes since v0.2:
- Massive speed upgrade
- Added fiat/day option (USD, EUR, GBP & CNY) using Coindesk
- Excludes very low volume coins now when "healthy coins" is checked.
- Added Poloniex
- Added option to weigh all exchanges into the BTC price, like this: (exchange1_price * exchange1_percentageOfTotalVol) + (exchange2_price * exchange2_percentageOfTotalVol) + ... Gives a better view on the real price if a coin is well spread among exchanges.
- Fixed PoolPicker
- General improvements and fixes

PS: If you guys think I should open another thread for this tiny app, please tell

Great work, running fine.

Are you working on integrating this into an auto switcher for ccMiner?

[yellowduck2]

I made some changes to my profit calc this weekend. So here it is:



v0.4: https://github.com/KBomba/cuda-profit-calc/releases/tag/v0.4b
Changes since v0.2:
- Massive speed upgrade
- Added fiat/day option (USD, EUR, GBP & CNY) using Coindesk
- Excludes very low volume coins now when "healthy coins" is checked.
- Added Poloniex
- Added option to weigh all exchanges into the BTC price, like this: (exchange1_price * exchange1_percentageOfTotalVol) + (exchange2_price * exchange2_percentageOfTotalVol) + ... Gives a better view on the real price if a coin is well spread among exchanges.
- Fixed PoolPicker
- General improvements and fixes

PS: If you guys think I should open another thread for this tiny app, please tell

In this thread is better. Too many threads to monitor

Profit in BTC is missing.

[tuanvie]

hi all.
I want to ask, I use nvidia gt 520 and want to try mine X11 does this work?
I've tried using cudaminer can only run on scrypt

Yes, although you need an older version for that. Use the ccminer 21 version from the V1.02-x13 download http://cudamining.cc/url/releases

ok thank you for the info friend,
I will immediately download and try

[Madychoux]

I made some changes to my profit calc this weekend. So here it is:



v0.4: https://github.com/KBomba/cuda-profit-calc/releases/tag/v0.4b
Changes since v0.2:
- Massive speed upgrade
- Added fiat/day option (USD, EUR, GBP & CNY) using Coindesk
- Excludes very low volume coins now when "healthy coins" is checked.
- Added Poloniex
- Added option to weigh all exchanges into the BTC price, like this: (exchange1_price * exchange1_percentageOfTotalVol) + (exchange2_price * exchange2_percentageOfTotalVol) + ... Gives a better view on the real price if a coin is well spread among exchanges.
- Fixed PoolPicker
- General improvements and fixes

PS: If you guys think I should open another thread for this tiny app, please tell

Thanks for your KillerApp !!!!

Could you add "energy cost" to get real earning ?

[Bombadil]

In this thread is better. Too many threads to monitor

Profit in BTC is missing.

Well spotted, friend! I directly fixed it and released it again: https://github.com/KBomba/cuda-profit-calc/releases/tag/v0.4.1b

Thanks for your KillerApp !!!!

Could you add "energy cost" to get real earning ?

I was thinking about this already. But still doubting about how to implement it properly in the GUI ^^"

Great work, running fine.

Are you working on integrating this into an auto switcher for ccMiner?

Well, if I have some more free time. ATM I like the profit calc part of it a little bit more. But its already a nice base to start with, even though I have the command prompt handling and gpu monitoring part already, just need to link everything up, but that has already proven to be a PITA. But as you all know, donations might be the motivator to overcome that

[leofur]

I made some changes to my profit calc this weekend. So here it is:



v0.4: https://github.com/KBomba/cuda-profit-calc/releases
Changes since v0.2:
- Massive speed upgrade
- Added fiat/day option (USD, EUR, GBP & CNY) using Coindesk
- Excludes very low volume coins now when "healthy coins" is checked.
- Added Poloniex
- Added option to weigh all exchanges into the BTC price, like this: (exchange1_price * exchange1_percentageOfTotalVol) + (exchange2_price * exchange2_percentageOfTotalVol) + ... Gives a better view on the real price if a coin is well spread among exchanges.
- Fixed PoolPicker
- General improvements and fixes

PS: If you guys think I should open another thread for this tiny app, please tell

EDIT: Oops, seems like I missed "BTC/day" ^^" Fixed it. Thx yellowduck2!

Do you know how many people have downloaded your app?

[Bombadil]

Do you know how many people have downloaded your app?

Github doesn't give that specific information but it does give this:

[tbearhere]

when ccminer go in loop,(i can't even read too fast) it mean that the gpu are crashed?

hey amph could you tell me what the plugs are that some use on the graphics cards to make them work without video load ? thz

[reorder]

But end result is that people and their support of open-source end up making more reason for people to take their innovation closed source until it is obsolete or nearly there.

Think about how much money you would be making if you had added the code to your AMD miners and mined with that for a month on your own... the market would still be there for the masses, and you would be making even more for having added to your advantage.

I think the AMD improvements were mostly caused by a new driver release, and adaptations of the code to make it run well on this release.

AMD could still gain from adding a bitsliced Groestl code. However I chose to publish only a code version that isn't trivially portable to AMD.

Christian

Come on, Canright published the sbox in GF(2^2) back in 2005

[cbuchner1]

Come on, Canright published the sbox in GF(2^2) back in 2005

GF(2^8)... we based our code on his work. What wasn't trivial was the Supermix function of Groestl. It is bigger and has more state than the AES mix function.
The S-Box was the easy part.

[reorder]

Come on, Canright published the sbox in GF(2^2) back in 2005

GF(2^8)... we based our code on his work. What wasn't trivial was the Supermix function of Groestl. It is bigger and has more state than the AES mix function.

MixBytes? On AMD, in the absence of pshufb-like instruction, you are better off doing it non-bitsliced (in fact, bitslicing back and forth just for sbox).

[cbuchner1]

Come on, Canright published the sbox in GF(2^2) back in 2005

GF(2^8)... we based our code on his work. What wasn't trivial was the Supermix function of Groestl. It is bigger and has more state than the AES mix function.

MixBytes? On AMD, in the absence of pshufb-like instruction, you are better off doing it non-bitsliced (in fact, bitslicing back and forth just for sbox).

No Sir, we stay in the bitsliced domain. In the double groestl hash of GroestlCoin we stay in bitsliced even during two rounds of hashing.

https://github.com/cbuchner1/ccminer/blob/master/groestl_functions_quad.cu

[reorder]

Come on, Canright published the sbox in GF(2^2) back in 2005

GF(2^8)... we based our code on his work. What wasn't trivial was the Supermix function of Groestl. It is bigger and has more state than the AES mix function.

MixBytes? On AMD, in the absence of pshufb-like instruction, you are better off doing it non-bitsliced (in fact, bitslicing back and forth just for sbox).

No Sir, we stay in the bitsliced domain. In the double groestl functon of GroestlCoin we stay in bitsliced even during two rounds of hashing.

https://github.com/cbuchner1/ccminer/blob/master/groestl_functions_quad.cu

Indeed, but it is slower than bitslice-sbox-bitslice back on AMD where you have enough registers, no (easily accessible) warp shuffle and no byte permutation instruction to speed up rotates. I have implemented it both ways, so it is not just a theory

[cbuchner1]

Indeed, but it is slower than bitslice-sbox-bitslice back on AMD where you have enough registers, no (easily accessible) warp shuffle and no byte permutation instruction to speed up rotates. I have implemented it both ways, so it is not just a theory

congratulations. Is that code already public?

[zelante]

Indeed, but it is slower than bitslice-sbox-bitslice back on AMD where you have enough registers, no (easily accessible) warp shuffle and no byte permutation instruction to speed up rotates. I have implemented it both ways, so it is not just a theory

omg... nvidia losing this battle (not war )?

[reorder]

Indeed, but it is slower than bitslice-sbox-bitslice back on AMD where you have enough registers, no (easily accessible) warp shuffle and no byte permutation instruction to speed up rotates. I have implemented it both ways, so it is not just a theory

congratulations. Is that code already public?

No, not doing this mistake again However, performance on GCN 1.1 is close to that of t-tables one with beta driver.

[cbuchner1]

omg... nvidia losing this battle (not war )?

but not currently on the power efficiency front. We might lose, but we do so very efficiently