Author

Topic: CCminer(SP-MOD) Modded NVIDIA Maxwell / Pascal kernels. - page 698. (Read 2347641 times)

sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
And Decred is basically to worthless to mine now after the dual miner hit.
If you guys make a dual miner, please combine it with a different algo... Like Neoscrypt (although more memory hard).

I thought of a dual etherum miner a long time ago, and I also suggested this to Genoil here on bitcointalk. This was before decred came, and I was thinking of a dual  Keccak(Maxcoin) + etherum.. You need a simple algo with no memory access to take advantage of the memory pipeline.
legendary
Activity: 1764
Merit: 1024
remember that with pascal or new amd, all those miner software will be useless

I'm hoping we'll reach an age where miners will collectively gather funds to pay for miners and developers to create miners and not the other way around...

Oh well, at least one can dream.

im with you on that mate ...

the dbm system is STILL functioning and running - so if miners want to donate by mining to the dev of their choice - they still can ...

its still going through nicehash - as unfortunately to date - i have not been able to get a dev to build the multipools i wanted to incorporate this donation system ...

we can STILL dream ...

#crysx

Too bad a dev fee isn't exactly like that without infrastructure and they already exist now...
sr. member
Activity: 364
Merit: 260
--- ChainWorks Industries ---
remember that with pascal or new amd, all those miner software will be useless

I'm hoping we'll reach an age where miners will collectively gather funds to pay for miners and developers to create miners and not the other way around...

Oh well, at least one can dream.

im with you on that mate ...

the dbm system is STILL functioning and running - so if miners want to donate by mining to the dev of their choice - they still can ...

its still going through nicehash - as unfortunately to date - i have not been able to get a dev to build the multipools i wanted to incorporate this donation system ...

we can STILL dream ...

#crysx
legendary
Activity: 1764
Merit: 1024
And Decred is basically to worthless to mine now after the dual miner hit.

If you guys make a dual miner, please combine it with a different algo... Like Neoscrypt (although more memory hard).
legendary
Activity: 2002
Merit: 1051
ICO? Not even once.
remember that with pascal or new amd, all those miner software will be useless

I'm hoping we'll reach an age where miners will collectively gather funds to pay for miners and developers to create miners and not the other way around...

Oh well, at least one can dream.
legendary
Activity: 1484
Merit: 1082
ccminer/cpuminer developer
only closedCL binary sellers like you Wolf are affected :p
legendary
Activity: 1764
Merit: 1024
remember that with pascal or new amd, all those miner software will be useless

Oh yeah? Kernels don't matter with new GPUs? XD

not that, but you need again to pay for a better release, also it's not like going from 750 to 970, both were maxwell, this is a new tech

Change between 2XX series and Fury wasn't a big deal. As long as they're CUDA compliant they'll work fine and just need a bit of tuning.
legendary
Activity: 3248
Merit: 1072
remember that with pascal or new amd, all those miner software will be useless

Oh yeah? Kernels don't matter with new GPUs? XD

not that, but you need again to pay for a better release, also it's not like going from 750 to 970, both were maxwell, this is a new tech
legendary
Activity: 1764
Merit: 1024
remember that with pascal or new amd, all those miner software will be useless

Oh yeah? Kernels don't matter with new GPUs? XD
legendary
Activity: 3248
Merit: 1072
remember that with pascal or new amd, all those miner software will be useless
sr. member
Activity: 364
Merit: 260
--- ChainWorks Industries ---
Yes. I am merging all the kernals into one superminer. The sp-mod ultimate.

nice ...

please let me know what the time frame is - and the cost left over for the ultimate ...

tanx sp ...

#crysx
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer
Yes. I am merging all the kernals into one superminer. The sp-mod ultimate.
sr. member
Activity: 364
Merit: 260
--- ChainWorks Industries ---


1. 0.1BTC: Pentablake +100-120% (3 releases)
2. 0.1BTC: Cryptonight +10% (one release) (reported bugs on compute 5.2)
3. 0.1BTC: Spreadcoin +10-20% (with full sourcecode / linux compatible) (9 releases)
4. 0.1BTC: All nicehash algos optimized. 0-10% (5 releases)(x11,x13,x15,nist5,quark,lyra2v2,neoscrypt)
5. 0.1BTC: decred +18-25% (7 releases) (Full sourcecode(linux) 0.4BTC)
6. 0.2BTC: Vcash(+13%+decred(+18-25%) (0.1 btc discount for the decred buyers) (4 releases)

would you look at combining the whole lot into one? ...

it would be a much better way - and a much more simplified system if it could be done ...

#crysx
sp_
legendary
Activity: 2954
Merit: 1087
Team Black developer

1. 0.1BTC: Pentablake +100-120% (3 releases)
2. 0.1BTC: Cryptonight +10% (one release) (reported bugs on compute 5.2)
3. 0.1BTC: Spreadcoin +10-20% (with full sourcecode / linux compatible) (9 releases)
4. 0.1BTC: All nicehash algos optimized. 0-10% (5 releases)(x11,x13,x15,nist5,quark,lyra2v2,neoscrypt)
5. 0.1BTC: decred +18-25% (9 releases) (Full sourcecode(linux) 0.4BTC)
6. 0.2BTC: Vcash(+13%+decred(+18-25%) (0.1 btc discount for the decred buyers) (6 releases)
legendary
Activity: 1797
Merit: 1028
WHAT PRIVATE MINERS ARE CURRENTLY AVAILABLE?  --

@sp_  Please list the most current private miners available.  Currenlty, I stopped mining DECRED with my 750ti rig and switched to your open-source v1.5.79, and mining mostly X11, Quark, Qubit, and Neoscrypt.  My other rigs are still mining ETH.  I have a couple of your private miners and am interested in the sourcecode.  Having been a long-time contributor, and also a Linux miner, I want to be able to compile and run the modified kernels under Linux.       --scryptr
sr. member
Activity: 438
Merit: 250
And only one person is interested in the dual miner Claymore made for Eth/Dcr and thinks there should be similar for Nvidia?

Is this just really hard to do, which is why no one wants to talk about it? Nvidia is crippled on Eth right now because of the memory bus... which means there is even more GPU power that could be devoted to mining Dcr.

GCN 1.0 cards are even worse these days. While compute 5.2 cards on win7/8 and linux will keep a steady hashrate until the DAG hits 2GB, GCN 1.0 have been on the downslope since the beginning. Hashrate is already down by 30%. when DAG hits 2GB it will be 50%. The good thing about the Claymore miner is that while memory latency increases with increasing DAG size, there's more time to spend on DCR mining.



Doesn't chunking the DAG help that quite a bit? You could additionally do a very small TMTO if it's worth it.

I haven't been able to get chunking working on AMD yet. The test program that I wrote to create the graph above has a much better hashrate when chunking is applied, but it doesn't really mine anything. When I apply chunking in ethash, I don't get any speed improvement, nor do I get valid results...
member
Activity: 117
Merit: 10
You have Kachur's SIMD, though.
I'm not quite good at decompiling and asm stuff.

I know - but didn't you get your GOST bin from him? Or did you do a different SIMD?
Nope, i never had his sibcoin bin.
member
Activity: 117
Merit: 10
You have Kachur's SIMD, though.
I'm not quite good at decompiling and asm stuff.
member
Activity: 117
Merit: 10
Yes, the result is, but at the moment the gap is very small, but still progress is being made

That's why I will release a sp-mod ultimate with all kernals optimized. Decred donators wil get a 0.1 BTC discount. Profitabillity is changing from algo to algo.
I have optimized all the 17 algos from x1 to x17, When I merge my kernals into the latest 1.7.6 branch other unkown algos will be faster as well. M2m, c11, sib etc..


Why is sib so slow on the GPU? On the CPU it's .73 x11, on GPU it's only .23.

Current opensource GOST implementations are pretty slow. I made a faster one for AMD and I will look into the cuda one soon.

I did one, too - WAY better. The current GOST one is dead slow.

what speed you achieved?

Looks like ~4.65MH/s on SIB for a 280X, but it's been a long while since I looked at it.
Smth around 5.2-5.25(280X) with a pretty high clocks(someone may consider them as high, for me it is ok): 1100/1500. In my case other algos are limiting the speed.
sr. member
Activity: 438
Merit: 250
And only one person is interested in the dual miner Claymore made for Eth/Dcr and thinks there should be similar for Nvidia?

Is this just really hard to do, which is why no one wants to talk about it? Nvidia is crippled on Eth right now because of the memory bus... which means there is even more GPU power that could be devoted to mining Dcr.

GCN 1.0 cards are even worse these days. While compute 5.2 cards on win7/8 and linux will keep a steady hashrate until the DAG hits 2GB, GCN 1.0 have been on the downslope since the beginning. Hashrate is already down by 30%. when DAG hits 2GB it will be 50%. The good thing about the Claymore miner is that while memory latency increases with increasing DAG size, there's more time to spend on DCR mining.

Jump to: