[ANN] sgminer v5 - optimized X11/X13/NeoScrypt/Lyra2RE/etc. kernel-switch miner

[thevictimofuktyranny]

Darkcoin is hitting new highs and X11 profits are pretty much the same as before.

Disappointing.

You expected people to just not notice and not hop on X11 until the diff made the returns shitty again?

Yes but now its the #3 minable coin. It should of made GPU more profitable again with the large market cap.

I guess there are millions of GPUs out there hashing away.

Darkcoin has about 70% of GPU hash at moment, it fluctuates between 50%-70%. It is 124GHs at the moment.

Or, it is the equivalent of 20,667 280X's or 38,750 750TI's. 90% of GPU's are AMD, so that nvidia comparison is hypothetical

Besides, Darkcoin is still good for rises in price up to $5.18 this year, I did a post looking into the degree of undervaluation on Darkcoin at the beginning of February based on GPU investments from the large mining groups and master-node investment (1000DRK per masternode): I bought extra DRK.  

[1714785844]

1714785844

[1714785844]

1714785844

[1714785844]

1714785844

[1714785844]

1714785844

[1714785844]

1714785844

[1714785844]

1714785844

[thevictimofuktyranny]

I'll repost it from February, with a number of improvements, which Thomas Paine famously recommended doing in The Age of Reason. I'm not incline to causing such controversies myself and I just apply it to what I wrote in February.


American are saying they're paying 14 cent for electricity and Chinese are saying they're paying 10 cent for electricity. Finally, in Iceland electricity cost is around 9 cent; not the rumoured 3-5 cents. Thirdly, I have updated figures for the current Fiat Currency prices.

"As a digression, for anyone investing in DRK.

The current ROI on a 4 card 750TI PC ($880) hashing machine on Darkcoin is:

POW 62.5% mining versus Masternode deduction of 37.2% on block reward, not everything goes through masternodes, effective block reward is 3.41DRK for miners.

China 10 cent for electricity and BTC price is $260

Average difficulty for 3 year breakeven on investment $880 = (Difficulty at 3600) 0.023BTC or $5.98.

Should you factor in 20% per year profit margin for the miner, yes, it is a miserly profit margin

0.0276BTC or $7.18.

These calculations won't really bite home until 2016, when old Litecoin GPU machines (to many) fail in large numbers, miners will either need to increase their selling price or quit GPU mining altogether (DRK supports up to 70% of GPU cards mining today). The latter would monopolize DRK into miners (who know what the are doing) and save in DRK. Many will refuse to sell their DRK on the exchanges until Darkcoin owners realize their long-term ambitions. This will push the price up this year and next year.  

Equally, there is 7% block reduction per annum and these calculations should be adjusted each year. Current Difficulty spike today is 4200, but this temporary spike disappears when miners figure out this lesson: save in DRK whenever it is undervalued, just redo the numbers yourself.  

[thevictimofuktyranny]

I've based that on publicly released open source software or leaked mods widely available, we all know that their are many mods to boost hashing power on Neoscrypt, Lyra2, X13 and so on.

[adaseb]

So 3 years to break-even with those 750ti GPUs?

WOW. Much worse then buying an BTC Asic.

[thevictimofuktyranny]

So 3 years to break-even with those 750ti GPUs?

WOW. Much worse then buying an BTC Asic.

Well, its very strategic on GPU's, because you can resell GPU's for decent money at anytime................plus the Litecoin transition out of GPU mining left the GPU alt coins over saturated with GPU mining hardware..............it has taken 6 months to replace GPU Litecoin with another alt coin (DRK)................profitability will rise this year and next..............Darkcoin is limited to 21 million coins ..........it will last a lot longer then GPU Litecoin, it will need to pass $28 per DRK before anyone considers making any ASIC mining hardware for it.........it has 10x lower daily output then LTC, plus an annual reduction of 7%..........so it could be many years of profitable GPU mining on DRK........allowing larger miners to make successful long term investments.

However, by 2018 and 2019 there will be more alt coins reaching maturity for GPU mining (1st Block Halvings or Reductions)

Here is top four, which varies with profit switching multi-pools (X11 averages are):
1) 124GHs (Peak is 144Ghs) Darkcoin Block Reduction 7% per annum 74%-----$66 million of GPU hardware.
2) 10.5GHs Vertcoin Block Halving 2018 6.3% -------$0.56 million of GPU hardware
3) 10GHs Startcoin Block Halving 2016 6%-----------$0.53 million of GPU hardware
4) 8.5GHs Feathecoin Block Halving 2019 5%-----------$0.45 million of GPU hardware

Other GPU crypto-currencies with weak mining support equal about 13Ghs 8%------$0.69 million.  Noteworthy coin with weaker mining support is Spreadcoin, with less then 20 million coins and a talented developer crew.

Total GPU hash power is approximately: 167GHs

You can see the headache left when GPU Litecoin ended unexpectedly; you've had $68 million of GPU mining hardware mining alt coins only worth $25 million dollars. This time around the larger miners won't make the same mistakes and nor will everyone else

[thevictimofuktyranny]

So 3 years to break-even with those 750ti GPUs?

WOW. Much worse then buying an BTC Asic.

Well, its very strategic on GPU's, because you can resell GPU's for decent money at anytime................plus the Litecoin transition out of GPU mining left the GPU alt coins over saturated with GPU mining hardware..............it has taken 6 months to replace GPU Litecoin with another alt coin (DRK)................profitability will rise this year and next..............Darkcoin is limited to 21 million coins ..........it will last a lot longer then GPU Litecoin, it will need to pass $28 per DRK before anyone considers making any ASIC mining hardware for it.........it has 10x lower daily output then LTC, plus an annual reduction of 7%..........so it could be many years of profitable GPU mining on DRK........allowing larger miners to make successful long term investments.

However, by 2018 and 2019 there will be more alt coins reaching maturity for GPU mining (1st Block Halvings or Reductions)

Here is top four, which varies with profit switching multi-pools (X11 averages are):
1) 124GHs (Peak is 144Ghs) Darkcoin Block Reduction 7% per annum 74%-----$66 million of GPU hardware.
2) 10.5GHs Vertcoin Block Halving 2018 6.3% -------$0.56 million of GPU hardware
3) 10GHs Startcoin Block Halving 2016 6%-----------$0.53 million of GPU hardware
4) 8.5GHs Feathecoin Block Halving 2019 5%-----------$0.45 million of GPU hardware

Other GPU crypto-currencies with weak mining support equal about 13Ghs 8%------$0.69 million.  Noteworthy coin with weaker mining support is Spreadcoin, with less then 20 million coins and a talented developer crew.

Total GPU hash power is approximately: 167GHs

I doubt it'll need to pass $28 a coin before people decide they're gonna make ASICs for X11, mainly because unlike Scrypt (even the way it was misused for LTC's PoW), X11 is fucking EASY to do on hardware. Most hash functions are designed to be fast, as well as lend themselves to hardware implementations. Scrypt is not a hash function, it's a KDF, and was designed to be an all-around pain in the ass.

It costs money to make the silicon design and you have to insure the yield on the wafer is high enough to manufacture ASICs equipment at a profit; there is 10 time less DRK each day compared to LTC, plus masternodes take another 3.75 away.  Therefore, you are looking at 14 times lower volume of coins available compared to LTC. Plus, each year there there is 7% reduction. Furthermore, DRK's developers has done 3-4 hard forks in the last year, so Wolf0, it will need a mighty high price to justify the inherit higher risks of engaging in process of producing ASIC equipment.  

[adaseb]

So 3 years to break-even with those 750ti GPUs?

WOW. Much worse then buying an BTC Asic.

Well, its very strategic on GPU's, because you can resell GPU's for decent money at anytime................plus the Litecoin transition out of GPU mining left the GPU alt coins over saturated with GPU mining hardware..............it has taken 6 months to replace GPU Litecoin with another alt coin (DRK)................profitability will rise this year and next..............Darkcoin is limited to 21 million coins ..........it will last a lot longer then GPU Litecoin, it will need to pass $28 per DRK before anyone considers making any ASIC mining hardware for it.........it has 10x lower daily output then LTC, plus an annual reduction of 7%..........so it could be many years of profitable GPU mining on DRK........allowing larger miners to make successful long term investments.

However, by 2018 and 2019 there will be more alt coins reaching maturity for GPU mining (1st Block Halvings or Reductions)

Here is top four, which varies with profit switching multi-pools (X11 averages are):
1) 124GHs (Peak is 144Ghs) Darkcoin Block Reduction 7% per annum 74%-----$66 million of GPU hardware.
2) 10.5GHs Vertcoin Block Halving 2018 6.3% -------$0.56 million of GPU hardware
3) 10GHs Startcoin Block Halving 2016 6%-----------$0.53 million of GPU hardware
4) 8.5GHs Feathecoin Block Halving 2019 5%-----------$0.45 million of GPU hardware

Other GPU crypto-currencies with weak mining support equal about 13Ghs 8%------$0.69 million.  Noteworthy coin with weaker mining support is Spreadcoin, with less then 20 million coins and a talented developer crew.

Total GPU hash power is approximately: 167GHs

You can see the headache left when GPU Litecoin ended unexpectedly; you've had $68 million of GPU mining hardware mining alt coins only worth $25 million dollars. This time around the larger miners won't make the same mistakes and nor will everyone else

Yes I guess you are correct

According to:
https://bitinfocharts.com/darkcoin/


Only 2584 DRK were mined in the last 24 hours and the block reward for miners is only 2.80 or so which is only around $7000

Right now the average hash is 100Ghash/s which if everybody is mining with an 280X or equivalent leads to 16,666 GPUs or so. So only $0.40 per day per 280X equivalent video card. Very sad considering back in January 2014 an 280X was pulling in $10/day.

[adaseb]

Cross-post from main VNL thread.

Okay, it's not perfect, but it'll never be. Sources that I've written or modified have been heavily commented to help out devs who are learning. The OpenCL was entirely written by me, and SGMiner itself was modified by me not just to run the modified kernel, but to clean it, remove algorithms besides WhirlpoolX for simplicity, and add comments. The horrid lib SPH is no longer used at all, instead, a SHA-256 implementation that was actually already there and never even compiled is used for several parts of SGMiner that used to call into lib SPH, and the CPU Whirlpool-512 implementation is a rather clean one I found here. That is also where I found seemingly the only copy of Whirlpool-512 done without tables, on which I based my bitsliced implementation. Another modification I made was to check the OpenCL version at runtime and use clCreateCommandQueueWithProperties() if the version is 2.0 or greater; it replaces clCreateCommandQueue(), and the latter is now deprecated and should not be used, but even the current official SGMiner isn't updated. Anyways, go look at the source for yourself. It will be here momentarily.

Screenshot of final hashrates, although you may find better configs (NSFW): https://ottrbutt.com/miner/whirlpoolxwolfpub-03222015.png

Win64 binaries have been made for release, of course. As you can verify on Bitcoin-OTC's database, my key ID is 0x88CBE71ADD5FB10F - check your GPG sigs. Statically compiled, linked against the beta version 3.0 of the AMD APP SDK, stripped, and signed. The cURL version used is a bit old, but it doesn't matter. ADL headers were whatever version I happened installed on my Linux machine - the headers themselves are portable, and the binary was cross-compiled. Archive contains the binary, the CL file, and a README.

https://ottrbutt.com/sgminer/whirlpoolx/win64/wolf-sgminer-whirlpoolx-03222015-win64.zip
https://ottrbutt.com/sgminer/whirlpoolx/win64/wolf-sgminer-whirlpoolx-03222015-win64.zip.sig

Don't know what I am doing wrong but I get the old 60Mhash/s rate for my 270x

[adaseb]

Cross-post from main VNL thread.

Okay, it's not perfect, but it'll never be. Sources that I've written or modified have been heavily commented to help out devs who are learning. The OpenCL was entirely written by me, and SGMiner itself was modified by me not just to run the modified kernel, but to clean it, remove algorithms besides WhirlpoolX for simplicity, and add comments. The horrid lib SPH is no longer used at all, instead, a SHA-256 implementation that was actually already there and never even compiled is used for several parts of SGMiner that used to call into lib SPH, and the CPU Whirlpool-512 implementation is a rather clean one I found here. That is also where I found seemingly the only copy of Whirlpool-512 done without tables, on which I based my bitsliced implementation. Another modification I made was to check the OpenCL version at runtime and use clCreateCommandQueueWithProperties() if the version is 2.0 or greater; it replaces clCreateCommandQueue(), and the latter is now deprecated and should not be used, but even the current official SGMiner isn't updated. Anyways, go look at the source for yourself. It will be here momentarily.

Screenshot of final hashrates, although you may find better configs (NSFW): https://ottrbutt.com/miner/whirlpoolxwolfpub-03222015.png

Win64 binaries have been made for release, of course. As you can verify on Bitcoin-OTC's database, my key ID is 0x88CBE71ADD5FB10F - check your GPG sigs. Statically compiled, linked against the beta version 3.0 of the AMD APP SDK, stripped, and signed. The cURL version used is a bit old, but it doesn't matter. ADL headers were whatever version I happened installed on my Linux machine - the headers themselves are portable, and the binary was cross-compiled. Archive contains the binary, the CL file, and a README.

https://ottrbutt.com/sgminer/whirlpoolx/win64/wolf-sgminer-whirlpoolx-03222015-win64.zip
https://ottrbutt.com/sgminer/whirlpoolx/win64/wolf-sgminer-whirlpoolx-03222015-win64.zip.sig

Don't know what I am doing wrong but I get the old 60Mhash/s rate for my 270x

The miner is meant for 14.12, that said, try lowering xI to 2048 or 1024; I just heard on IRC that a worksize of 128 works well for 270X cards. I personally managed around 115MH/s with it, using a worksize of 256, xI of 2048, and 2 GPU threads. Also, double check that you're setting GPU threads to 2.

Yes I tried everything, with xI of 1024, 2048 and worksize of 128.

Will try the 14.12 cataylst version.

[pallas]

I also figured out the Catalyst 15.3 Beta seems to be even better for this miner.

Any way to run on linux? Extracting libopencl from Ubuntu vivid package?

[pallas]

I also figured out the Catalyst 15.3 Beta seems to be even better for this miner.

Any way to run on linux? Extracting libopencl from Ubuntu vivid package?

Running Catalyst 15.3? If you can rip it out of shitbuntu's package they got from AMD, awesome. Also, please upload it somewhere.

I've extracted the libs and rebuilt the (modified) kernel with those, it became 104 bytes fatter and hashrate went from 377 to 374 on a 290x@936 MHz; don't know how it works on your stock kernel, though...
Still want the files?

[Neosaan]

Super!!!! thank you so much !!!
280x earned by 260mh\c   ,  270x ... and not given the addition of speed ..... what am I doing wrong?

[dhsc19]

Cross-post from main VNL thread.

Okay, it's not perfect, but it'll never be. Sources that I've written or modified have been heavily commented to help out devs who are learning. The OpenCL was entirely written by me, and SGMiner itself was modified by me not just to run the modified kernel, but to clean it, remove algorithms besides WhirlpoolX for simplicity, and add comments. The horrid lib SPH is no longer used at all, instead, a SHA-256 implementation that was actually already there and never even compiled is used for several parts of SGMiner that used to call into lib SPH, and the CPU Whirlpool-512 implementation is a rather clean one I found here. That is also where I found seemingly the only copy of Whirlpool-512 done without tables, on which I based my bitsliced implementation. Another modification I made was to check the OpenCL version at runtime and use clCreateCommandQueueWithProperties() if the version is 2.0 or greater; it replaces clCreateCommandQueue(), and the latter is now deprecated and should not be used, but even the current official SGMiner isn't updated. Anyways, go look at the source for yourself. It will be here momentarily.

Screenshot of final hashrates, although you may find better configs (NSFW): https://ottrbutt.com/miner/whirlpoolxwolfpub-03222015.png

Win64 binaries have been made for release, of course. As you can verify on Bitcoin-OTC's database, my key ID is 0x88CBE71ADD5FB10F - check your GPG sigs. Statically compiled, linked against the beta version 3.0 of the AMD APP SDK, stripped, and signed. The cURL version used is a bit old, but it doesn't matter. ADL headers were whatever version I happened installed on my Linux machine - the headers themselves are portable, and the binary was cross-compiled. Archive contains the binary, the CL file, and a README.

https://ottrbutt.com/sgminer/whirlpoolx/win64/wolf-sgminer-whirlpoolx-03222015-win64.zip
https://ottrbutt.com/sgminer/whirlpoolx/win64/wolf-sgminer-whirlpoolx-03222015-win64.zip.sig

Hi Wolf0,

Any chance you would compile the sgminer for linux 64bit?

[smolen]

Anyways, now I'm curious, so I decided to take a look at what's really inside the x64 version of libOpenCL.a from the 3.0 beta version of the APP SDK. What I found seems to confirm this. I dumped the external symbols with a my cross-build of GNU nm, and got this: http://pastebin.com/EPPijWKu

The capital I means that it's a reference to another symbol. Not surprising - the only time I've ever seen symbols beginning in "__imp_" is when they were located in a DLL. The U means that symbol is undefined, and the T means that the symbol can be found in the text (code) section. The reference to another symbol is completely expected, the undefined reference is a bit odd, but one in the text section? That doesn't make sense to me, so I dumped the section data (run through grep to prune all but the important bits). That is here: http://pastebin.com/0i4aEXpk

Now, it's been a while since I've poked around in the PE format, but I don't get this. Each .text section is 8 bytes, which isn't really enough room for anything, let alone code. If anyone has an idea, let me know, cause that shit's confusing.

Anyway, I guess I've established the shit we need definitely isn't in the libOpenCL.a library...

Looks like usual import library. IMHO those 8 bytes gets into executable at link time (or is it just empty dataless placeholder?), at load time dl fills it with resolved function address, at execution time code performs indirect jump by this address (dl patches not JMP code  in read-only text segment, but read-write data, so library image could have different base address for each process and still be shared). I thought that gnu linker works directly with .so files but nothing prevents it from this scheme. Well, in PE there are fancy things like forwarded exports but sure that's not the case.
EDIT: yes, something strange with sections, I'll try to feed this library to IDA, it's good at such things

[smolen]

That's not linux, sorry, I've messed everything up. Looked at .a dump and talked about dl, but that's windows PE import library, by mingw or cygwin. No actual code there. Well, you could have some fun with IDA and actual .dll, a good way to kill a week or two

[thevictimofuktyranny]

New MinerControl pack available at Crypto Mining Blog for Yamp and Nicehash:

http://cryptomining-blog.com/4549-miner-control-1-6-4-pack-for-amd-radeon-r9-280x/

Adjust for you cards hashing output and put you wallet address in and remember to just add Wolf0's X11 (50% boost), X13 (50% boost) and X15 (25% boost) modded bins and make sure those 3 modded algos have 64 worksize.

Bins download from:

http://www.reddit.com/r/DRKCoin/comments/2o1yoz/rewritten_x11_binaries/

Using bigger worksize for qubit (128) and whirpool (256) get higher hash output on a R9 290

You can donate to Crypto Mining Blog a few Litecoins or mBTC as Thank You for their hard work

[Arnault59]

Hi, sorry for this noob question but ten months I didn't try mining, I forget many things. I have three 280 TRIX, I downloaded sgminer v5 for window but unable to write properly the .bat to mine DigiByte groestl on hub pool. Can someone help please ?

[thevictimofuktyranny]

Hi, sorry for this noob question but ten months I didn't try mining, I forget many things. I have three 280 TRIX, I downloaded sgminer v5 for window but unable to write properly the .bat to mine DigiByte groestl on hub pool. Can someone help please ?

Try this update version of SGminer 5.1.1. Precompiled for Windows

http://cryptomining-blog.com/4535-updated-windows-binary-of-sgminer-5-1-1-with-fixed-lyra2re-support/

You are better off using Configuration file.

[confudido]

Mining x11, gpu0 hashes but no submitted shares and no hardware errors, gpu1 is ok.
Tried switching the two gpus but still gpu0 doesn't work (i.e. the gpu in the top pci slot).
I'm using wolf0's binaries.
Mining other algos works fine.
Also tried older sgminer versions but no change.
Any clues?

Any updates or fixes?  I'm having the exact same issue.  Thanks!

[troky]

Mining x11, gpu0 hashes but no submitted shares and no hardware errors, gpu1 is ok.
Tried switching the two gpus but still gpu0 doesn't work (i.e. the gpu in the top pci slot).
I'm using wolf0's binaries.
Mining other algos works fine.
Also tried older sgminer versions but no change.
Any clues?

I'd say the problem is between Wolf's x11 bins and new AMD drivers (14.12 and 15.3beta).
Standard x11 kernel works with any driver version. Sgminer version is irrelevant here.

Wolf, do you have any explanation?