CCminer(SP-MOD) Modded GPU kernels.

[sp_]

Vanilla (V-cash) sp-mod #6 sendt to the donators. Please test and report your numbers compared to the alexis Kernal.

Compute 5.0 have a default intensity of 29.
Compute 5.2 have a default intensity of 30.

I get faster hashrates on all my cards. gtx 750,750ti,950,960,970,980,980ti

Compiled with the 364.51 driver. Cuda 7.5 32 bit.

[1714948413]

1714948413

[1714948413]

1714948413

[sp_]

About xvc, your latest (cuda 7.5 32bit exe) on windows 7 x64,
clocks ~1466:
970 ~3160
980 ~3890

hmm. slower than version #3. Can you try #6?

[sp_]

Can you disable L1 cache with -dlcm=cg while compiling? I dont know if it's already done since i dont use windows and the compilation parameters on my git (for windows) are broken. Thanks in advance

Interesting. I will try this.

[alexis78]

Can you disable L1 cache with -dlcm=cg while compiling? I dont know if it's already done since i dont use windows and the compilation parameters on my git (for windows) are broken. Thanks in advance

Interesting. I will try this.

Always here to help SP
I didnt come with a bad attitude, you brought me here with a bad attitude by promising inexistent speeds.
Also, you can try -abi=no , though it's deprecated, it performs slightly better here (edit: by slightly i mean less than 0.02%)

[antantti]

About xvc, your latest (cuda 7.5 32bit exe) on windows 7 x64,
clocks ~1466:
970 ~3160
980 ~3890

hmm. slower than version #3. Can you try #6?

It is faster, numbers soon.

-edit-

What pool are you guys using?

[sp_]

You can use this pool

ccminer -a vanilla -o stratum+tcp://pool.v.cash:3001 -u  VafRjT9uiVUasRPVdQi4KiKJJYW81BuDLQ -p d=10

[alexis78]

@antantti
i'm on suprnova in order to check the reported hashrate through the graphs. Been using xCore's pool too (https://pool.v.cash/)
There should be no problem choosing a pool

[antantti]

Pools can be picky sometimes, quick tests showed me that xvc gives reliable results on some major pools.

xvc@1466 -i 30 #5

970 ~3170
980 ~3920

xvc@1466 -i 30 #6

970 ~3260
980 ~4030

All on windows 7 x64.

-edit-

Results with p0.

[nrg_wolf]

So Sp_
How fast is your DCR miner for the 970 now?

[alexis78]

xvc@1466 -i 30 #5
970 ~3170
980 ~3920

xvc@1466 -i 30 #6
970 ~3260
980 ~4030

All on windows 7 x64. Results with p0.

Here is a screenshot of my latest vcash code (Compiled less than an hour ago on CUDA7.5 CP5.2,5.0,3.5,3.0) which gave me a +0.7% boost



It reports an average of 3.303Gh/s on an average of 1453.3MHZ on my GTX970

And here are the latest binaries:
http://s000.tinyupload.com/index.php?file_id=07038368028057919944 from the above screenshot

I'd be glad if testers could validate my results

Also, for historical purposes and to make things clear, i'd be glad if testers could grab the ccminer for vcash compiled on 24th of March for CUDA7.5 CP5.2, 5.0, 3.5 from here:
https://v.cash/forum/threads/ccminer-faster-8-round-blake-algo-1-16x.282/page-2
(It's the last post)
And report their results on this one too

Edit2:
Testers should find their own intensities (i use 31 for 970 on vcash)

@antantti
Can you validate my kernel's hashrate on your machine too?
Some basic math makes me expect that you should see (1466/1453.3)*3.303 ~= 3.33Gh/s  _{(That's not the case for memory bounded algos.)}

which is almost +70MH/s than SP_MOD#6 or 2.1% faster

Another thing is that I saw everybody using my first (13th of February) built for cuda 6.5 - CP 3.5 to compare against the private kernels.
I asked from A friend / forum member to run my latest build in comparison to cuda6.5 cp3.5 and found it 1.48% faster on his 2 x 980tis (running at 1466MHz).
His hashrate actually went from 10.7-10.8Gh/s to 10.91Gh/s .

[alexis78]

So Sp_
How fast is your DCR miner for the 970 now?

He reached my throughput with his SP_MOD#8
Run the 1.5% fee miner and you'll get a good approximation on your machines
https://bitcointalk.org/index.php?topic=1290358.msg14476625#msg14476625

I've released it yesterday because, up to yesterday, i thought SP's results were legit and he was indeed 25-30% than tpruvot's git while he is only 15-20%
and after the latest commits by pallas and me, he is 6% faster than the publicly available
Probably he used the default intensities there too in order to increase his sales and not the actual throughput

[sp_]

Decred #4 is the fastest if you run it with -i 29.6. But the hashrate on the pool is lower because of a uint32_t overflow in the nonce per thread loop. I have corrected this in #5.

Default intensity for the 980ti is 31. But it runs faster with -i 31.9. with overclocking you can reach 3.3GHASH.

1. The difference between a Linux build and a windows builds can be as much as 5%.
2. A optimalization on the gtx 970, can be slower than on the 750ti.
3. I compare my builds to the cryptomining blog builds, and sometimes I build myself. (cuda 7.5 32 bit)

[tbearhere]

1. 0.1BTC: Pentablake +100-120% (3 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)

I will put these 4 in one build.(1.7.6 refork) the price will be 0.4BTC It includes unpublished decred #8, Vcash #5, Quark #6. Discounts for the donators. So If you already have bought decred and VCASH you only need 0.2BTC more.

I just sent you .1 BTC for
4. 0.1BTC: All nicehash algos optimized. 0-10% (5 releases)(x11,x13,x15,nist5,quark,lyra2v2,neoscrypt)
I can't afford .4 BTC

But you have already donated 0.3 BTC haven't you. (from the beginning of the SP-MOD project) Donators get discounts..

Yes I did and I sent you another .1 BTC  
Net amount: -0.10010213 BTC
Transaction ID: ce6839a959c74297112b0c24b34859223cfffeffd4016ea4c72c42bf3dac887a-000
Thank you
Set you a pm.

Sp  how are you doing on the final build?  .... waiting for the quark and x11 x13  nicehash  algo's  
And yes now it's  a total of .4 BTC I sent...thx
EDIT:There is a increase in the last one you sent for DECRED but I only tested it for a few minutes cause I'm working on my first conf file for something and it's taking me a long time.

[alexis78]

1. The difference between a Linux build and a windows builds can be as much as 5%.

so far, not on my algos. But for this one, credits go to tpruvot. He is the windows tech guy

2. A optimalization on the gtx 970, can be slower than on the 750ti.

You are right, but that will happen if you use architecture specific asm, just like the lop.b32 instructions i used to further optimize your already optimum keccak256 kernal.
You can check my github, run it on your machines, but i can't trust your results, whatever you'll claim. Sorry.

But the hashrate on the pool is lower because of a uint32_t overflow in the nonce per thread loop. I have corrected this in #8.

That's what happens when you copy-paste code you dont understand for over a year from algo to algo Mr. "3lit3".
Yet again, by the time you fixed your binaries and brought to your buyers a fixed product, this code was already uploaded in my github fixed, since Feb of this year.
And it's been already a week since the pallas comit on the open-sourced repo.


The reason it was not working in CUDA7.5 came a month later when you reported bigger hashrate in CUDA7.5 and I was trusting your results.

You made me look into it and i found the flaw which was preventing CUDA7.5 to translate the source the same way CUDA6.5 did.
I remember that chrysophylax said both private decred miners were reporting almost the same results, with yours being slightly faster.
But mine was already 15% faster in CUDA6.5 version rather than CUDA7.5 which was reporting less than 1% boost.

That said: From SP_VAPORMOD#1 to SP_VAPORMOD#3 you were selling vaporware.
And you just admitted that from SP_BROKENMOD#4 until, SP_MOD#8 you were selling broken code?  

3. I compare my builds to the cryptomining blog builds, and sometimes I build myself. (cuda 7.5 32 bit)

You do it wrong, but i wont charge you (the 3lit3) 0.4BTC to tell you how to measure the actual optimization of your code.

So, what you should do in your case?

At first you should set the hashrate of CUDA6.5 as the new bottom, or at least inform your buyers that it's optimum for CUDA7.5 builds but it still lacks some major performance from the open sourced CUDA6.5 implementation.
//This thing is called respect to the actual work of other devs, which work was brought in public for free, for the credits. Not for BTCs
You cant even claim that you missed my implementation, the code was open sourced, the binaries were available, even cryptomining blog wrote an article about that, and the same article was commented by you in this thread.
I dont see anything that could excuse you for this behavior since you claim to be an experienced developer "who built the fastest ccminer" (an 3lit3) and not just a kid who is just playing around, trying to learn.

Then, you have to find the best arguments, and the best toolkit for the comparison tests.
I could have used tpruvot's version in cuda6.5 (which was proven worse) just to claim a better throughput, but i didnt. Cuz that would be wrong in every aspect since his source works optimally in CUDA7.5.
I also compared the two versions passing the same arguments, e.g. using the same intensity. I didnt use tpruvot's defaults (-i 25 or 27) and pushed mine to it's limits (-i 31) without even trying tpruvot's kernel with -i 31.
That would lead me to falsified results, just like you were led to.
You can't disagree on the above because, an optimization is not just a higher intensity. With a higher intensity you can claim bigger throughput, but in no case you can claim faster code, because simply the code is the same.

The thing you did with intensity to claim a "15% faster than Provos Alexis kernal" was also a big time joke.
It is the end-users obligation to find the best intensity for their equipment, the dev can only RECOMMEND through the default intensities, cuz you know... variables in programming need to be initialized with a value.
If your buyers had run your code with your default intensity set at 31 on a gpu which was driving their screen, their system would freeze.
That thing you did with intensity was like taking an empty gas tank car, put some gasoline. Drive it and claim you are better than the X car company's engineers because you managed to get an infinite percentage of acceleration on their cars. You see my point?

How do you even claim to be "3lit3" and ask BTC if you cant figure this simple thing out.


Over 'n out

[djm34]

sp please stop scamming people and steal code from the real dev and report faster speed without even doing anything.
This is getting out of hand. You are basically selling as private opensource code...

If you don't want to work on new code rather than stealing people work, may-be it is time for you to leave...

And please review your linkedin profile (before someone comment on it): You did not invent anything.

[scryptr]

I DRANK BEER LAST NIGHT--

Having earned the BitCoin to pay for it with my mining, I enjoyed an excellent Scotch Ale last night.  I woke up early today to read Alexis' post, and now am confused!  WHO indeed has the fastest miner!

My donations have gonee to sp_ and his rough-and-rugged enhanced coding.  Now, here is a newbie saying in fiery style that HIS code is the greatest?  Well, perhaps sp_ should just mod it so that the common miners like myself can put it to use, even while thoroughly hung-over.

Geez buleez, I am ready to go back to bed already.  As soon as I can figure out which private miner is the "all-in-one"  with Alexis' fixes, Epsylon3's smooth base code, all the algos, and sp_ 's assembly code enhancements, I'll plunk down another donation.  Not that I haven't donated to Pallas, Epsylon3, or a couple other true contributors.  Just get it right and tell me where it is.  And toss a bone to the noob, Alexis.  He deserves a few BTC for his contribution.

Back to bed.  My gear might finally ROI before the ETH and DECRED markets bottom out.  Adding another chunk of BTC to the coders might not be wasted.  DECRED is still my first choice for the 750ti because of the issues it has with ETH's DAG file.  Is somebody saying that VanillaCoin actually pays better?  Or what?       --scryptr

[alexis78]

I DRANK BEER LAST NIGHT--

Having earned the BitCoin to pay for it with my mining, I enjoyed an excellent Scotch Ale last night.  I woke up early today to read Alexis' post, and now am confused!  WHO indeed has the fastest miner!

My donations have gonee to sp_ and his rough-and-rugged enhanced coding.  Now, here is a newbie saying in fiery style that HIS code is the greatest?  Well, perhaps sp_ should just mod it so that the common miners like myself can put it to use, even while thoroughly hung-over.

Geez buleez, I am ready to go back to bed already.  As soon as I can figure out which private miner is the "all-in-one"  with Alexis' fixes, Epsylon3's smooth base code, all the algos, and _ 's assembly code enhancements, I'll plunk down another donation.  Not that I haven't donated to Pallas, Epsylon3, or a couple other true contributors.  Just get it right and tell me where it is.  And toss a bone to the noob, Alexis.  He deserves a few BTC for his contribution.

Back to bed.  My gear might finally ROI before the ETH and DECRED markets bottom out.  Adding another chunk of BTC to the coders might not be wasted.  DECRED is still my first choice for the 750ti because of the issues it has with ETH's DAG file.  Is somebody saying that VanillaCoin actually pays better?  Or what?       --scryptr

The noob you are talking about has 4 years of experience in CUDA, since the CP2.1 era.
Also, the noob you are talking about was from the first contributors to open source CudaMiner.
You see me here just because the idiot you trust, dared to say "15% faster than Provos Alexis kernal",
where, even after his 6th release, he's still actually slightly slower.
I wouldn't care less where you send your btcs.
If you like being scammed by an idiot, it's not my problem.

I dont even care if his performance is actually faster on the other kernels.
And at this point i dont trust anything he claims.
I Only know two things:
1. He has been slower than my decred implementation bragging that he's an 3lit3 who brought 30% speed increase, while in the same time he was using broken copy-pasted code from a 2011 bitcoin implementation
If Pallas didnt push a faster public release, the donators to SP_MOD private kernels would still be mining with a broken and slower code
2. His vcash ccminer private implementation is still slower than the publicly released one even after his claims about 15% boost on it

That said. I wont push my latest +0.7% vcash commit on github. At least not before his vcash SP_MOD#7

[Vaccinus]

he also scammed me 0.05 for a boost on cryptonight, that was lie because all he did was increasing the intensity, in fact i have the same hash as he claim with just an intensity increase lol

[scryptr]

I DRANK BEER LAST NIGHT--

Having earned the BitCoin to pay for it with my mining, I enjoyed an excellent Scotch Ale last night.  I woke up early today to read Alexis' post, and now am confused!  WHO indeed has the fastest miner!

My donations have gonee to sp_ and his rough-and-rugged enhanced coding.  Now, here is a newbie saying in fiery style that HIS code is the greatest?  Well, perhaps sp_ should just mod it so that the common miners like myself can put it to use, even while thoroughly hung-over.

Geez buleez, I am ready to go back to bed already.  As soon as I can figure out which private miner is the "all-in-one"  with Alexis' fixes, Epsylon3's smooth base code, all the algos, and _ 's assembly code enhancements, I'll plunk down another donation.  Not that I haven't donated to Pallas, Epsylon3, or a couple other true contributors.  Just get it right and tell me where it is.  And toss a bone to the noob, Alexis.  He deserves a few BTC for his contribution.

Back to bed.  My gear might finally ROI before the ETH and DECRED markets bottom out.  Adding another chunk of BTC to the coders might not be wasted.  DECRED is still my first choice for the 750ti because of the issues it has with ETH's DAG file.  Is somebody saying that VanillaCoin actually pays better?  Or what?       --scryptr

The noob you are talking about has 4 years of experience in CUDA, since the CP2.1 era.
Also, the noob you are talking about was from the first contributors to open source CudaMiner.
You see me here just because the idiot you trust, dared to say "15% faster than Provos Alexis kernal",
where, even after his 6th release, he's still actually slightly slower.
I wouldn't care less where you send your btcs.
If you like being scammed by an idiot, it's not my problem.

I dont even care if his performance is actually faster on the other kernels.
And at this point i dont trust anything he claims.
I Only know two things:
1. He has been slower than my decred implementation bragging that he's an 3lit3 who brought 30% speed increase, while in the same time he was using broken copy-pasted code from a 2011 bitcoin implementation
If Pallas didnt push a faster public release, the donators to SP_MOD private kernels would still be mining with a broken and slower code
2. His vcash ccminer private implementation is still slower even after his claims about 15% boost on the publicly released one

YOUR RATING IS NEWBIE--

And I am glad that you have good coding experience.  Just as long as sp_ can enhance your miner with assembly code, and make it faster, I am happy.

Now, as to whether the code is modded, or the code is original, or the code is "only an Intensity (-i) increase",  I don't care.  Sorry that you are so angry, I was just hoping others would donate for your code.  If you contribute Open Source, others can mod it, that is the idea idea behind Open Source.

Don't get your panties in a bunch.  I just want a faster, better mining software.  By the way, medical doctors can advertise these days, at least in the USA.  It is no longer considered an ethical violation to promote their medical services.  Honestly, sp_'s bragging is slightly 3l1t3st, and from the days of telophone-linked bulletin boards, Commodore 64's, Amiga coding, and "scrollers" (I hacked a few scrollers) when ANSI-taggers had flamewars online, but I guess you just want to be MORE 3l1t3.  Put a fiery ANSI artwork in your tagline, I guess you are 3l1t3 too.  It is not like you are selling medicine.

If the snake-oil kills the pain, it must be good!  Just as long as it works.  I need some.......       --scryptr

[Nikolaj]

1. The difference between a Linux build and a windows builds can be as much as 5%.

so far, not on my algos. But for this one, credits go to tpruvot. He is the windows tech guy

2. A optimalization on the gtx 970, can be slower than on the 750ti.

You are right, but that will happen if you use architecture specific asm, just like the lop.b32 instructions i used to further optimize your already optimum keccak256 kernal.
You can check my github, run it on your machines, but i can't trust your results, whatever you'll claim. Sorry.

But the hashrate on the pool is lower because of a uint32_t overflow in the nonce per thread loop. I have corrected this in #8.

That's what happens when you copy-paste code you dont understand for over a year from algo to algo Mr. "3lit3".
Yet again, by the time you fixed your binaries and brought to your buyers a fixed product, this code was already uploaded in my github fixed, since Feb of this year.
And it's been already a week since the pallas comit on the open-sourced repo.


The reason it was not working in CUDA7.5 came a month later when you reported bigger hashrate in CUDA7.5 and I was trusting your results.

You made me look into it and i found the flaw which was preventing CUDA7.5 to translate the source the same way CUDA6.5 did.
I remember that chrysophylax said both private decred miners were reporting almost the same results, with yours being slightly faster.
But mine was already 15% faster in CUDA6.5 version rather than CUDA7.5 which was reporting less than 1% boost.

That said: From SP_VAPORMOD#1 to SP_VAPORMOD#3 you were selling vaporware.
And you just admitted that from SP_BROKENMOD#4 until, SP_MOD#8 you were selling broken code?  

3. I compare my builds to the cryptomining blog builds, and sometimes I build myself. (cuda 7.5 32 bit)

You do it wrong, but i wont charge you (the 3lit3) 0.4BTC to tell you how to measure the actual optimization of your code.

So, what you should do in your case?

At first you should set the hashrate of CUDA6.5 as the new bottom, or at least inform your buyers that it's optimum for CUDA7.5 builds but it still lacks some major performance from the open sourced CUDA6.5 implementation.
//This thing is called respect to the actual work of other devs, which work was brought in public for free, for the credits. Not for BTCs
You cant even claim that you missed my implementation, the code was open sourced, the binaries were available, even cryptomining blog wrote an article about that, and the same article was commented by you in this thread.
I dont see anything that could excuse you for this behavior since you claim to be an experienced developer "who built the fastest ccminer" (an 3lit3) and not just a kid who is just playing around, trying to learn.

Then, you have to find the best arguments, and the best toolkit for the comparison tests.
I could have used tpruvot's version in cuda6.5 (which was proven worse) just to claim a better throughput, but i didnt. Cuz that would be wrong in every aspect since his source works optimally in CUDA7.5.
I also compared the two versions passing the same arguments, e.g. using the same intensity. I didnt use tpruvot's defaults (-i 25 or 27) and pushed mine to it's limits (-i 31) without even trying tpruvot's kernel with -i 31.
That would lead me to falsified results, just like you were led to.
You can't disagree on the above because, an optimization is not just a higher intensity. With a higher intensity you can claim bigger throughput, but in no case you can claim faster code, because simply the code is the same.

The thing you did with intensity to claim a "15% faster than Provos Alexis kernal" was also a big time joke.
It is the end-users obligation to find the best intensity for their equipment, the dev can only RECOMMEND through the default intensities, cuz you know... variables in programming need to be initialized with a value.
If your buyers had run your code with your default intensity set at 31 on a gpu which was driving their screen, their system would freeze.
That thing you did with intensity was like taking an empty gas tank car, put some gasoline. Drive it and claim you are better than the X car company's engineers because you managed to get an infinite percentage of acceleration on their cars. You see my point?

How do you even claim to be "3lit3" and ask BTC if you cant figure this simple thing out.

Over 'n out

sp please stop scamming people and steal code from the real dev and report faster speed without even doing anything.
This is getting out of hand. You are basically selling as private opensource code...

If you don't want to work on new code rather than stealing people work, may-be it is time for you to leave...

And please review your linkedin profile (before someone comment on it): You did not invent anything.

Respect. I totally agree that this thing is getting out of hand. Keep on the great work guys

I've just tried the 1.5% fee miner for decred, gtx 970@1650GH/s @1342MHz with I ~ 28 (better to use 26, but I should make some more efficiency analysis)