Topic: [VTC] Lyra2RE Vertcoin optimized miner, NOW WITH SOURCES! - page 5. (Read 26923 times)

this is link for pallas kernel

https://dl.dropboxusercontent.com/u/40353042/sgminer-lyra2re.zip

https://codeload.github.com/badman74/Binaries/zip/V5_0_mod


mod sgminer for lyra2

A link in this thread would be very useful ☺

Can you post me a link of 64 bit windows sgminer you are using?

I found

Look at my avatar :-)
Jokes aside, if someone provides a windows build, I'll be grateful.
I could do it myself but it takes time, my machines are all on linux.

Thanks for sharing. I haven't looked at the code yet, but perhaps there is something I can use to improve the speed on the NVIDIA cards.

and where we can download sgminer.exe for your sources?

So here it is, the full miner sources.
It includes a modified sgminer based on the djm34 version and my opencl files (some are slightly enhanced, some are heavily modified).
The kernel is not compatible with stock sgminer: contrary to what is usually done on multi algo kernels, I've made this one monolitic (instead of the original split version).

https://dl.dropboxusercontent.com/u/40353042/sgminer-lyra2re.zip

If you feel you can donate to: 1H7qC5uHuGX2d5s9Kuw3k7Wm7xMQzL16SN
Or just mine a bit with the provided example.bat which loads lyra2re.conf

Many thanks and have fun!

I was forgot about that



Of course, all here are interested. Please release them, thanks.

Yeah, you might as well. Someone else may be able to improve on your hard work.

I have a working djm34 sgminer modded sgminer, which does 1.5mhs at 163watts for 290.

Therefore, the card is only used at 50%. And, it is clear 2mhs plus is pretty doable.

I, myself, am not a coder; so it is all double dutch to me

I'm good at economic theory and theoretical models, so I make my profits from buying and selling crypto currencies

see the first lines of the OP.

What sort of hash are you getting?

if people still interested I'll release the miner sources.

any news?

Farms with expensive power, rent, loans, high taxes and customs are going to be thrown out of business, assume that those who stayed in mining already optimized expenses.

Thank you for the elaborations even though those are not evidence in the proper sense of the word.
Sending you a PM containing some extra information not strictly related to this thread.

Yeah, I show you some maths evidence:

We will use current difficulty of 101 and not the peak of 132, because large mining organisations can arrange sufficient GPU's are mining a coin so it less profitable, yet profitable to them on their Bios mods and modded software. They can build up huge positions - small miners will simply switch to other algo coins on profitability.

CCminer 750Ti Lyra2 output is 740KHs.

Woking out profitability calculation is: 1000 divided by (Difficulty divided by 1000) times block ratio (block reward divided by 50) = coins yield per 1ghs of hashpower in 24 hours. Coin yield per 1ghs times current price (0.00004BTC) = profitability number for comparison purposes. 1GHs profitability times 0.00074 (740KHs) = your earnings per 1 750TI

1000 / 0.101 x 1 =9900VTC.
9900VTC x 0.00004 = 0.396BTC earned on 1GHs Lyra2 hashpower
0.396 x 0.00074 = 0.00029304BTC or $0.07 you earning on 1 card per 241 hours

To compare this to X11 profitability, we need to convert into X11 profitability
750TI X11 output is 3200KHs
Calculation is
Lyra2 haspower divided X11 haspower = ratio of difference.
Ratio of difference times profitability number for comparison purposes = X11 profitability

740 / 3400 = 0.23125
0.396 x 0.23125 = 0.091575BTC X11 equivalent
  
What is Darkcoin profitability:
7 day average difficulty of blocks found at miningpoolhub is 4120
1000 / 4.120 x 0.0682 = 16.55DRK
16.55 x 0.019 = 0.31BTC
0.31 x 0.0032 = 0.000992BTC or $0.25 on a single 750TI

So, VTC profitbility on CCminer converted into X11 profitability is 0.09157BTC and Dash (DRK) is 0.31BTC. Therefore, CCminer owners will have switched to other more profitable algo coins.

This leaves sgminer (official version) of sgminer is:

800khs on a 280x = 0.0003168BTC or $0.08
X11 output is 6mhs on 280x = 0.00186BTC or $0.465

So, small miners will not be mining Lyra2 VTC. They will mine another coin and then swap it for VTC.

This leaves the large mining operations using AMD GPU's. They have their own modded bios for GPU's and coders to mod software or write their own mining software. So, the question is: How much have they squeezed out of sgminer?

On Neoscrypt, Wolf0 has recoded it to squeeze an extra 90% e.g. 325KHs has increased to 590KHs, a 81% increase on the official SGminer.

Wolf0, is part-time coder, the large mining organisations can hire coders on a full-time basis or there owner is a coders, until every performance gain has been squeezed out of mining software.

Therefore, a large mining organisation can be achieving double or more hash output per a card then a small miner. They can use this to build huge positions in coins over years via selling the coin down to get rid of smaller miners and when a block halving happens, they control of large proportion and can set the new market price making decent profits.

This applies only to GPU mining and not ASIC mining, with ASIC mining the manufacturers of the equipment pay for mining software development. Therefore, little can be squeezed out of this equipment via software modding.

AMD got $59 million in GPU sales over the past 3 years and made about $5.9 million plus in profits from crypto currencies and spent $0 on developing mining software for their cards.

Nvidia got $7 million in GPU sales in the last year, but has spent $0 on developing mining software for their cards.

This is why GPU mining is more better proposition for the large mining operations; they are the only people funding optimizations in software for mining these algorithms. But, it is so much more strategic and require patience wait 1 year or more. Therefore, not something the less talented or impatient businessmen get involved in.

That's good idea, thank you! Indeed, if big farm operators are instant dumpers with perfect arbitrage, it's possible to compare unknown best implementations of different algorithms.