CCminer(SP-MOD) Modded GPU kernels.

[sambiohazard]

I am usually paid for what I develop and open-source...

And after you open source it, somebody works for free and optimize your work and get a few beers in donations. I don't understand your problem.. Miners needs the software to be up do date to earn money. Old kernals are not profitable any more.

ahem...

https://bitcointalk.org/index.php?topic=826901.msg12425383#msg12425383

https://bitcointalk.org/index.php?topic=826901.msg12424197#msg12424197

[1714670280]

1714670280

[1714670280]

1714670280

[1714670280]

1714670280

[1714670280]

1714670280

[1714670280]

1714670280

[sp_]

I am usually paid for what I develop and open-source...

And after you open source it, somebody works for free and optimize your work and get a few beers in donations. I don't understand your problem.. Miners needs the software to be up do date to earn money. Old kernals are not profitable any more.

ahem...
https://bitcointalk.org/index.php?topic=826901.msg12425383#msg12425383
https://bitcointalk.org/index.php?topic=826901.msg12424197#msg12424197

Your card is trottleing, and I have already told you what to do about it.
I have reduced the default intensity abit in release 68. Please try it out.

[sambiohazard]

but relatively djm34's miner is still faster in same environment. You got your variables wrong. I think you should lower intensity to a decent base, as you have done with new release. People can always try up from a lower bound and find best for their own hardware. Highest intesity for your h/w breaks the miner for people like me. You using highest intensity, now that says something about your motivations & preferences.

EDIT: i would like to add that i appreciate your work & donate whenever i can but after release 56 i had to stop using your miner. I hope we all get this sorted, i am currently using djm34's miner even though it not most optimized but it most stable for me.

[sp_]

Here are the changeset's between release 56 and 57:
Wich algo are you mining?
The only change in the miner logic except the cuda kernals is the "stop mining when disconnecting fix" by Pallas


https://github.com/sp-hash/ccminer/commits/windows?page=7

[sambiohazard]

i mostly mine quark or lyra2v2 and nothing is broken but i guess you optimized your miner or increased default intensity to a point where my cards started throttling. I am just saying that default intensity should be such that anyone  who runs the miner gets it running w/o throttling/crashing. Anyone who is not a noob can find best for their own cards for sure but it breaks or at least makes experience worse for noobs. May be set the intensity by putting your cards on reference clock from nvidia & post results from both default intensity as well as your personal highest hashrate using highest intensity you can get. This will define a range for intensity within which people can find best for their own setup. I hope this makes sense.

[pallas]

between those commits there is my little modification to stop gpus when disconnected (it has been edited by sp back then):

/* prevent gpu scans before a job is received */
if ((have_stratum && work.data[0] == 0 || network_fail_flag) && !opt_benchmark)
{
   sleep(1);
   continue;   
}

Let's see if it has something to do with solo mining problems; try changing it like this:

if (have_stratum && work.data[0] == 0 && !opt_benchmark)

[sp_]

I have added it now.

[djm34]

I am usually paid for what I develop and open-source...

And after you open source it, somebody works for free and optimize your work and get a few beers in donations. I don't understand your problem.. Miners needs the software to be up do date to earn money. Old kernals are not profitable any more.

I have no problem with you optimizing my work, as long as an actual optimization takes place which hasn't always been the case lately (I tested what you did on lyra2rev2 side algorithms (blake,bmw,cubehash) and I found absolutely no difference while the numbers you reported were by itself equivalent to half what you would get if you had removed these algos (minus cubehash...), which from my stand point  is more or less equivalent as saying that what I release isn't optimized. So at least make sure it is true before saying it... (especially when you say it almost every days... hey I understand that you need to motivate donators, but please don't do it repeatedly on my back  )

ps: I don't deny that you did a good job with quark and x11

[sp_]

I am usually paid for what I develop and open-source...

And after you open source it, somebody works for free and optimize your work and get a few beers in donations. I don't understand your problem.. Miners needs the software to be up do date to earn money. Old kernals are not profitable any more.

I have no problem with you optimizing my work, as long as an actual optimization takes place which hasn't always been the case lately (I tested what you did on lyra2rev2 side algorithms (blake,bmw,cubehash) and I found absolutely no difference while the numbers you reported were by itself equivalent to half what you would get if you had removed these algos (minus cubehash...), which from my stand point  is more or less equivalent as saying that what I release isn't optimized. So at least make sure it is true before saying it... (especially when you say it almost every days... hey I understand that you need to motivate donators, but please don't do it repeatedly on my back  )
ps: I don't deny that you did a good job with quark and x11

Every small change I do might not do any difference, but when you add all the small changes they matter. The changes in blake,bmw,cubehash was so small compared to the lyra2v2 algo that it was almost not messurable, but my implementation use less assembly instructions. After I changed the lyra2v2 algo to use less registers my version is mining at around 4,4MHASH on the 750ti on standard clocks. I can overclock stable to 5,150.

Your version does 3,950.  (750ti) (4,6 with overclocking.) My modded kernal is 10% faster...

I do the tests in windows(x86)
You need to merge all the modded kernals together and adjust/increase the intensity to get the speedup...

[scryptr]

SECURECOIN--

I have downloaded the latest SecureCoin (SRC) wallet, version 0.9.0.0-g12be17d-beta.  It is now called "Securechain-wallet", and contains security upgrades to protect from the Heartbleed bug and etc.  It is synchronizing as I type this, and fairly quickly.

I plan to solomine SRC and test the latest git from sp_, commit 1092 with Pallas' patch.  My 6x 750ti FTW rig can generate 40Mh/s+ of Quark algo in the SRC net, and should hit about 1 block per hour.  SRC is worth about 1/2 of a Vertcoin (VTC).  Right now, CoinWarz rates SRC as more profitable than VTC.  The rapid rate for finding an SRC block should at least check the latest patch quickly.

--scryptr

[Slava_K]

Git 1092 dont work whith solo mining...

[pallas]

Git 1092 dont work whith solo mining...

I'm sad that it doesn't work but happy that it's not my fault
:-D
I assume we can revert the last commit, than, because it disabled the "don't send work to GPU while disconnected", which is quite useful.
Actually I just wanted someone to test it (to know if that part was to blame), not that it ended in the main fork...
I'd have tested it myself if I had the time to do it.

[scryptr]

Git 1092 dont work whith solo mining...

SOLO-MINING BUG--

Thanks for the information.  Which coin did you attempt to mine, and what hash rate did you send into the net?

I will wait for another patch before I mine on the SecureCoin (SRC) net.

--scryptr

[t-nelson]

Does anyone have a list of coins where soloing is known to be broken?  Maybe one with low diff?

Phoenixcoin has low diff and the blockchain should download fairly quick for the wallet.

I find blocks in a matter of mins. after starting r54, r55 & r56 gives more boooo's than r54 that I get none in.
-i value needs to be low -i 10 is a good starting point for solo mining phoenixcoin.
at 300+ kh/s I find a block in about every 15 mins most of the time on a single GTX960.

r68 dump log

Code:

[2015-09-23 02:44:50] NVAPI GPU monitoring enabled.
[2015-09-23 02:44:50] 1 miner thread started, using 'neoscrypt' algorithm.
[2015-09-23 02:44:50] Binding thread 0 to cpu 0 (mask 1)
[2015-09-23 02:44:50] JSON protocol request:
{"method": "getwork", "params": [], "id":0}


* Rebuilt URL to: http://127.0.0.1:3333/
*   Trying 127.0.0.1...
* TCP_NODELAY set
* Connected to 127.0.0.1 (127.0.0.1) port 3333 (#0)
* Server auth using Basic with user 'phoenixcoinpc'
> POST / HTTP/1.1

Authorization: Basic cGhvZW5peGNvaW5wYzp4

Host: 127.0.0.1:3333

Accept-Encoding: identity

Content-Type: application/json

Content-Length: 45

User-Agent: ccminer/1.5.67-git(SP-MOD)

X-Mining-Extensions: longpoll noncerange reject-reason

X-Mining-Hashrate: 0



< HTTP/1.1 200 OK

< Date: Wed, 23 Sep 2015 07:44:50 +0000

< Connection: keep-alive

< Content-Length: 311

< Content-Type: application/json

< Server: phoenixcoin-json-rpc/v0.6.6.0

< 

* Connection #0 to host 127.0.0.1 left intact
[2015-09-23 02:44:50] JSON protocol response:
{
   "error": null,
   "result": {
      "algorithm": "neoscrypt",
      "data": "02000000e2f3ceab75f7db954acf51383b48743fea1faa9f0216239bcc864ff0f1c66a0a221a2f1c9f713cd60c3d9cc41aa5fc9c5e52c37405c96a017a19442d5158802f6c580256549e071d000000003d010000",
      "target": "0000000000000000000000000000000000000000000000000000549e07000000"
   },
   "id": 0
}

[2015-09-23 02:44:50] JSON inval data
[2015-09-23 02:44:50] json_rpc_call failed, retry after 30 seconds

r54 dump log:

-- SNIP --

Thanks.  Looks like neoscrypt has a 168byte data size?  We're passing 128 to that hack hex2bin() function, so it fails.

[flipclip]

Phoenixcoin has low diff and the blockchain should download fairly quick for the wallet.

I find blocks in a matter of mins. after starting r54, r55 & r56 gives more boooo's than r54 that I get none in.
-i value needs to be low -i 10 is a good starting point for solo mining phoenixcoin.
at 300+ kh/s I find a block in about every 15 mins most of the time on a single GTX960.

r68 dump log

Code:

[2015-09-23 02:44:50] NVAPI GPU monitoring enabled.
[2015-09-23 02:44:50] 1 miner thread started, using 'neoscrypt' algorithm.
[2015-09-23 02:44:50] Binding thread 0 to cpu 0 (mask 1)
[2015-09-23 02:44:50] JSON protocol request:
{"method": "getwork", "params": [], "id":0}


* Rebuilt URL to: http://127.0.0.1:3333/
*   Trying 127.0.0.1...
* TCP_NODELAY set
* Connected to 127.0.0.1 (127.0.0.1) port 3333 (#0)
* Server auth using Basic with user 'phoenixcoinpc'
> POST / HTTP/1.1

Authorization: Basic cGhvZW5peGNvaW5wYzp4

Host: 127.0.0.1:3333

Accept-Encoding: identity

Content-Type: application/json

Content-Length: 45

User-Agent: ccminer/1.5.67-git(SP-MOD)

X-Mining-Extensions: longpoll noncerange reject-reason

X-Mining-Hashrate: 0



< HTTP/1.1 200 OK

< Date: Wed, 23 Sep 2015 07:44:50 +0000

< Connection: keep-alive

< Content-Length: 311

< Content-Type: application/json

< Server: phoenixcoin-json-rpc/v0.6.6.0

< 

* Connection #0 to host 127.0.0.1 left intact
[2015-09-23 02:44:50] JSON protocol response:
{
   "error": null,
   "result": {
      "algorithm": "neoscrypt",
      "data": "02000000e2f3ceab75f7db954acf51383b48743fea1faa9f0216239bcc864ff0f1c66a0a221a2f1c9f713cd60c3d9cc41aa5fc9c5e52c37405c96a017a19442d5158802f6c580256549e071d000000003d010000",
      "target": "0000000000000000000000000000000000000000000000000000549e07000000"
   },
   "id": 0
}

[2015-09-23 02:44:50] JSON inval data
[2015-09-23 02:44:50] json_rpc_call failed, retry after 30 seconds

r54 dump log:

-- SNIP --

Thanks.  Looks like neoscrypt has a 168byte data size?  We're passing 128 to that hack hex2bin() function, so it fails.

It looks like that bit of neoscrypt code was changed at commit: 43f4d827a032f1d89da5c957b1e823d7f54d60af (not sure if that will help).

[t-nelson]

Phoenixcoin has low diff and the blockchain should download fairly quick for the wallet.

I find blocks in a matter of mins. after starting r54, r55 & r56 gives more boooo's than r54 that I get none in.
-i value needs to be low -i 10 is a good starting point for solo mining phoenixcoin.
at 300+ kh/s I find a block in about every 15 mins most of the time on a single GTX960.

r68 dump log

-- SNIP --

r54 dump log:

-- SNIP --

Thanks.  Looks like neoscrypt has a 168byte data size?  We're passing 128 to that hack hex2bin() function, so it fails.

It looks like that bit of neoscrypt code was changed at commit: 43f4d827a032f1d89da5c957b1e823d7f54d60af (not sure if that will help).

It looks like bad data lengths are being passed all over the place...  I managed to coerce a good JSON string in the debugger.  Of course it was reject since the block was hours old, but at least getting close.

[qqqq]

Phoenixcoin has low diff and the blockchain should download fairly quick for the wallet.

I find blocks in a matter of mins. after starting r54, r55 & r56 gives more boooo's than r54 that I get none in.
-i value needs to be low -i 10 is a good starting point for solo mining phoenixcoin.
at 300+ kh/s I find a block in about every 15 mins most of the time on a single GTX960.

r68 dump log

-- SNIP --

r54 dump log:

-- SNIP --

Thanks.  Looks like neoscrypt has a 168byte data size?  We're passing 128 to that hack hex2bin() function, so it fails.

It looks like that bit of neoscrypt code was changed at commit: 43f4d827a032f1d89da5c957b1e823d7f54d60af (not sure if that will help).

It looks like bad data lengths are being passed all over the place...  I managed to coerce a good JSON string in the debugger.  Of course it was reject since the block was hours old, but at least getting close.

Will this understanding be affected on others algo ?

[scryptr]

SECURECOIN BLOCK FOUND--

I took my saved copy of sp_'s CCminer release dot 54 a began solo-mining with it against my SecureCoin wallet (v9.0) across my LAN.  I successfully found a block within an hour:


Linux 6x 750ti FTW 40Mh/s rig finds SRC block (Quark algo).

The "securecoin.conf" file, for the wallet:

==============cut line=================

rpcuser=yourname
rpcpassword=x
rpcallowip=192.168.1.*
rpcport=12568
daemon=1
server=1
addnode=24.222.191.37
addnode=71.187.248.95
addnode=71.77.233.192
addnode=54.234.172.139
addnode=86.2.171.203
addnode=208.77.217.7
addnode=208.77.217.8
addnode=us-east1.nodes.mywl.lt
addnode=us-ny1.nodes.mywl.lt
addnode=178.237.35.34:12567
addnode=209.188.16.220:12567
addnode=5.9.120.80:12567
ddnode=62.24.83.120:12567
addnode=78.27.191.182:12567
addnode=64.251.188.62:12567
addnode=108.53.178.39:12567
addnode=207.12.89.221:12567
addnode=193.87.209.44:12567
addnode=198.199.70.224:12567

==============cut line=================

The securecoin wallet is found (Win 7) in "...appdata/roaming/securecoin" under the login user's name.

My launch command line (Linux):

"./ccminer -a quark -i 23.9 -o http://192.168.1.111:12568 -u yourname -p x"

where the IP address matches the LAN IP of the Win 7 computer with the wallet.

I may move up to sp_ release dot 56, but the performance is so much better with release dot 68.  I hope the bug is squashed.

--scryptr

[qqqq]

It's very strange cause i tried rel 54 on SRC and it failed. Or maybe it was 56, testing 54 right now.

[bensam1231]

Does anyone have a list of coins where soloing is known to be broken?  Maybe one with low diff?

SecureCoin (SRC) was suggested.  It is a Quark coin. Another miner posted that it has the same error message on finding a block (crashing) as VertCoin (VTC).  CCminer will generate Quark hash with SRC; SRC has very rapid blocks at low difficulty.

Currently, I do not think solo-mining will work with any algo using CCminer.  

please precise ccminer from sp... my releases have no problem, tpruvot releases neither... so please don't generalize to every ccminer release.
That's in part why I am rather unhappy about the situation. myself and other have busted our asses on ccminer while the most publicised version which mostly uses the code we developed gets broken on a daily basis, feature gets removed, not even possible to run in debug mod without recompiling etc...).

no offence sp, it is time you get your shit together and stop releasing 4 untested releases a day...

PRECISE--

OK, you have a point, DJ.  I was posting in sp_'s thread, however.  I DO try to be clear about OS, version, hash rates, clock rates, spelling and punctuation, etc.  I've groaned about the bleeding edge speed before, myself, but sp_ has brought my 750ti cards from 5.5+ Mh/s to 6.1Mh/s mining Quark within the last 20 releases.

Maybe a few releases should focus on code clean-up and standardization rather than speed.  Future improvements would then be on a stronger base.       --scryptr

Considering DJM doesn't release anything, he doesn't really offer an alternative. I'd rather take a buggy and sometimes glitchy version with speed over one that's getting steamrolled by the private miners (usually released by the same person).

Also, everyone is getting their panties in a knot about solo mining. Who even solo mines now days? Only people who do have to have a relatively large farm for that to actually work (on the profitable coins). Perhaps consider charging them for a solo mining version. The amount of shits I have for this issue are about 0 as it is I'm sure for the majority of miners as we're all on pools... because, you know, we can't magic 5% of the network hashrate out of our asses.

Seems as though people are on SP's balls because profits are tight and he's offering a open source-public miner, which hurts the private miner business. For instance, DJM's toted L2V2 rewrite isn't even worthwhile anymore compared to Quark or other alternatives.

Without downplaying what sp_ has bring to us, the nvidia maxwell "small" miners communauty, he has built on top of what others like djm, truvot, tsiv,and before KlausT cbuchner had already built. Without them ccminer wouldn't be what it is.
On the other hand solomining is important for small miners : this allow us to benefit from recently launched coins or low profiles ones, often more profitable. Moreover, ethereum (rather shift, exp which have lower difficulties) mining is more profitable in solo, due to very fast blocks which penalized pools mining if you have not low latency network between your rig and the pool.
 

Yeah, no. If you can solo mine Ethereum right now you have a pretty big mining operation. Things balance out over the long term, but I tried mining it solo for a few days and never hit a block... That was before it took off and all the AMD miners hopped on it.

As far as new coins... I guess. If .61 for instance still works, you're not losing much by going back to it, especially for a low difficulty coin you're prospect mining. Usually coins get pools before they ever get an exchange. It doesn't take long. Still makes this a non-issue.

He definitely has built on top of what other coders have done, but that's part of being open source. You open source your ancient miner that is a year and a half old and someone builds off of it, you don't get to bitch about someone 'misusing' your code. It's always your decision to open source a miner and contribute. Some of the aforementioned names have contributed a lot more to the continued development of CCminer and I think they've also gained recognition for it (You're missing Pallas for instance).

The whole reason CCminer is so popular is because it gets love and attention from continued development. If other developers are jealous of this, maybe they should pull their dick out of their ass and try a different payment model - like donations or a % based fee. Or partner up with SP and work towards group donation or a % miner fee. It's all deserved. I've seen 'partners' in this thread receive donations too, much like SP, although I'm sure in a smaller amount.

I don't think the private miner coders get to bitch about SP having bugs in his release and what a 'terrible' version of their code it makes. Oh god, the humanity.

second: stop accusing other devs. to release private miners all the time, which is your excuse for everything
(you know very well it isn't true at least for me since you contacted me to have my private neoscrypt and I told you no.
I release only very rarely private miners and mostly on a trust basis (if the fee is large enough, I could trust you more... yeah well...  ))

Pretty much sums it up. You just reiterated everything I just said.

Sure, L2V2 was profitable for awhile, then it's not anymore after you decided to go private with the miner (which I'm sure you still ended up selling). No improvements to L2V2 speed happened afterwards, the only change was the difficulty change and some bug improvements. Neither of which effected me.

As far as NeoScrypt, why wouldn't I try to buy a private miner? I don't have the option of a good public one. In the case of SP's work, I do.

BTW there will never be fair rewards for the miners because you can't measure the value of each one's work. And anyway it's a handful of dollars worth.

Sure you can, a % based miner fee. A coop between developer, redistribution of the fee. Roughly maybe based off of initial work, speed improvements, and bug/feature fixes. You could make a system for figuring things out or let the community decide. A company would do this all on their own though.

The current issue may be under the guise of 'Sp doesn't fix bugs', but it's all about money. They know SP gets donations because his miner is popular. It's also popular because it gets continual updates and releases. He actually gives a shit about improving hash speeds too, instead of walling it off and selling it. Lo and behold, people like this and they praise him for it, offer donations, and talk about it. Go figure people like things people like?

I think this may actually be the start of a actual company or partnership between developers and maybe this scares some solo devs a bit.


Keep in mind SP credits everyone for their work AFAIK. It's pretty open ended as far as who has contributed, not only in the miner, but also in his updates.