[ANN] ccminer 2.3 - opensource - GPL (tpruvot)

[danspasiva]

Can anyone here help me? I simply cannot get the ccminer to work no matter what i try.  It turns on, mines like 15 shares, then crashes with the error message "illegal memory access"

Nothing i've tried works, even with lower intensity.

Running 6x 1070's on windows 10

Maybe try raising your virtual memory up to some crazy amount like 64GB or something.

tried this, nothing is working at the moment i keep getting illegal memory access after the first 10 minutes worth of mining and it crashes.  Literally about to beat my rig up with a baseball bat lol.  Anyone have any solutions to the  "illegal memory access" problem please?

[Amph]

Here lets go down the list.

Bitcore is 58% faster 1070 to 1080ti
x17 is 56% faster
Tribus is 65% faster
Skein itself is 65% faster
Blake14r(Decred) is 69% faster
Equihash is 51% faster
Library is 68% faster
Lyra2v2 is 69% faster

wrong on x17 you can do 70% faster, 20MH with a 1080ti versus 11-12MH

skein is 85% faster, again wrong, 550MH versus 1GH

equihash is more like 60%, 480 versus 780

lbry is 300 versus 500, so less than what you have

i don't know where you pulled those numbers but they are all wrong...i have both card and i tested with the best miner for 1080ti, the alexis 1.0 not ccminer 2.0

Check the link, it's from a database of 100s of cards. Your numbers are completely irrelevant. You're doing something wrong along the way or have a private miner. The database also lists the miner.

I told you EXACTLY where I pulled the numbers from. Myself, what to mine, cryptominingblog, and most importantly Epsylons database on yiimp. Like there is literally no other way you can try to prove those numbers. Hell those numbers WOULD be in the database on yiimp if you mined with those speeds... lol. What do you think it means what the miner reports "Pool asked for stats"? XD

uh, just look again at the bench on yiimp, skein is reported 1 giga for the 1080ti and the 1070 550MH, like i said which is 85% faster and not 65% faster like you said, so dunno what you are talking about

[sidog]

Been looking around but can't find the answer anywhere.   

I have a 7 GPU 1080 TI mining rig and would like to be able to control the power limit for all 7 cards via the ccminer --plimit parameter.    I'm currently on Windows 10 version 1703, page file set to 32-64gb, with a 4gb stick and intel g3900 celeron processor.   All is running fine and dandy, but I would like to explore the possibility of controlling the power for all 7 cards directly via the ccminer executable. 

As it stands, I can only control the first GPU using the --plimit parameter.  I cannot find any documentation regarding passing that parameter to apply to all (identical) cards.

[heyz]

Hello,

Does anyone know why I'm getting those errors? It's GTX 1070 card. And the MSI overclock settings are:
Core +80; Mem: +400; power: 60; Temp: 85; Fan auto. The setting works fine with other miners.


>ccminer-x64.exe -q -a groestl -o stratum+tcp://erebor.dwarfpool.com:3345 -u xxx.CAT-2 -p 1
*** ccminer 2.0 for nVidia GPUs by tpruvot@github ***
[2017-07-13 23:14:39] Starting on stratum+tcp://erebor.dwarfpool.com:3345
[2017-07-13 23:14:39] 4 miner threads started, using 'groestl' algorithm.
[2017-07-13 23:14:41] Stratum difficulty set to 16 (0.06250)
[2017-07-13 23:14:41] groestl block 1669538, diff 716.480
[2017-07-13 23:14:41] GPU #0: Intensity set to 19, 524288 cuda threads
[2017-07-13 23:14:41] GPU #1: Intensity set to 19, 524288 cuda threads
[2017-07-13 23:14:41] GPU #2: Intensity set to 19, 524288 cuda threads
[2017-07-13 23:14:41] GPU #3: Intensity set to 19, 524288 cuda threads
[2017-07-13 23:14:45] accepted: 0/1 (diff 8.507), 136.17 MH/s booooo
[2017-07-13 23:14:45] reject reason: expected string or buffer
[2017-07-13 23:14:47] accepted: 0/2 (diff 0.075), 136.23 MH/s booooo
[2017-07-13 23:14:47] reject reason: expected string or buffer
[2017-07-13 23:14:49] accepted: 0/3 (diff 0.113), 135.54 MH/s booooo
[2017-07-13 23:14:49] reject reason: expected string or buffer
[2017-07-13 23:14:55] accepted: 0/4 (diff 0.066), 135.53 MH/s booooo
[2017-07-13 23:14:55] reject reason: expected string or buffer
[2017-07-13 23:14:56] accepted: 0/5 (diff 0.260), 135.29 MH/s booooo
[2017-07-13 23:14:56] reject reason: expected string or buffer
[2017-07-13 23:14:57] accepted: 0/6 (diff 2.611), 135.39 MH/s booooo


Thanks beforehand!

[Epsylon3]

Been looking around but can't find the answer anywhere.   

I have a 7 GPU 1080 TI mining rig and would like to be able to control the power limit for all 7 cards via the ccminer --plimit parameter.    I'm currently on Windows 10 version 1703, page file set to 32-64gb, with a 4gb stick and intel g3900 celeron processor.   All is running fine and dandy, but I would like to explore the possibility of controlling the power for all 7 cards directly via the ccminer executable. 

As it stands, I can only control the first GPU using the --plimit parameter.  I cannot find any documentation regarding passing that parameter to apply to all (identical) cards.

--plimit=150,150,150,....

7x 1080ti wow lol

[bensam1231]

Here lets go down the list.

Bitcore is 58% faster 1070 to 1080ti
x17 is 56% faster
Tribus is 65% faster
Skein itself is 65% faster
Blake14r(Decred) is 69% faster
Equihash is 51% faster
Library is 68% faster
Lyra2v2 is 69% faster

wrong on x17 you can do 70% faster, 20MH with a 1080ti versus 11-12MH

skein is 85% faster, again wrong, 550MH versus 1GH

equihash is more like 60%, 480 versus 780

lbry is 300 versus 500, so less than what you have

i don't know where you pulled those numbers but they are all wrong...i have both card and i tested with the best miner for 1080ti, the alexis 1.0 not ccminer 2.0

Check the link, it's from a database of 100s of cards. Your numbers are completely irrelevant. You're doing something wrong along the way or have a private miner. The database also lists the miner.

I told you EXACTLY where I pulled the numbers from. Myself, what to mine, cryptominingblog, and most importantly Epsylons database on yiimp. Like there is literally no other way you can try to prove those numbers. Hell those numbers WOULD be in the database on yiimp if you mined with those speeds... lol. What do you think it means what the miner reports "Pool asked for stats"? XD

uh, just look again at the bench on yiimp, skein is reported 1 giga for the 1080ti and the 1070 550MH, like i said which is 85% faster and not 65% faster like you said, so dunno what you are talking about

Yeah, and those weren't up there when I posted that, nor are they legit benchmarks. It's the same user reporting multiple instances with the same GPU. Running a 1080ti at +50% TDP and crazy clocks for a limited run.

Also lets take 1030 and divide it by 580 (highest hashrate 1070 currently, which if you notice isn't just a giant block of a single type of card all ran on the same day) and you end up at 80%. If you remove those outliers, what's the next closest hashrate to the top for 1080ti? A EVGA GTX 1080ti running at 932.

932/580 = 60%.

Like I said, not taking into the one guy that ran his card at bleeding edge speed for five seconds before his rig caught fire and everything melted to the ground.

So uhhhh... like I said go be a dumbass somewhere else.

[Epsylon3]

yep, should be this kind of math which force the alts to go down when btc do too

[pallas]

The usual blah blah nonsense.
For memory light algos, just use the cuda cores count.

[bensam1231]

The usual blah blah nonsense.
For memory light algos, just use the cuda cores count.

The point was that's not a good metric and you don't get results which are 85% higher even when it's not a memory heavy algo (85% is the different in cuda cores). Originally people were making a case for buying 1080/tis before the pricing for 1070s got all messed up. $400x 1.65 = $660 not $800.

Now it's about on par, but you're sacrificing being able to mine other algos which depend more on memory. That's just NOW, it's possible for new algos to come out which utilize memory and for a 1080 to be crippled in them. I haven't seen a single algo where GDDR5X helps, even though it has more bandwidth then GDDR5.

[gzubeck]

Can anyone here help me? I simply cannot get the ccminer to work no matter what i try.  It turns on, mines like 15 shares, then crashes with the error message "illegal memory access"

Nothing i've tried works, even with lower intensity.

Running 6x 1070's on windows 10

Maybe try raising your virtual memory up to some crazy amount like 64GB or something.

tried this, nothing is working at the moment i keep getting illegal memory access after the first 10 minutes worth of mining and it crashes.  Literally about to beat my rig up with a baseball bat lol.  Anyone have any solutions to the  "illegal memory access" problem please?

I would try disconnecting all your garaphics cards and start with one primary one. I had a bad install of graphics drivers and some cards we're not working at all. Keep adding one at a time to make sure you don't have any defective equipment or risers. Also, you might want to post your bat file minus any keys to see what your starting up. In addition, what kind of power supply are you using? I'm asking because that could also shut things down if your power supply is not sufficient. I have 4 gtx 1080s with an 850 watt power supply and I would not want to add another card without chaining another power supply to my rig.

[gzubeck]

Hello,

Does anyone know why I'm getting those errors? It's GTX 1070 card. And the MSI overclock settings are:
Core +80; Mem: +400; power: 60; Temp: 85; Fan auto. The setting works fine with other miners.


>ccminer-x64.exe -q -a groestl -o stratum+tcp://erebor.dwarfpool.com:3345 -u xxx.CAT-2 -p 1
*** ccminer 2.0 for nVidia GPUs by tpruvot@github ***
[2017-07-13 23:14:39] Starting on stratum+tcp://erebor.dwarfpool.com:3345
[2017-07-13 23:14:39] 4 miner threads started, using 'groestl' algorithm.
[2017-07-13 23:14:41] Stratum difficulty set to 16 (0.06250)
[2017-07-13 23:14:41] groestl block 1669538, diff 716.480
[2017-07-13 23:14:41] GPU #0: Intensity set to 19, 524288 cuda threads
[2017-07-13 23:14:41] GPU #1: Intensity set to 19, 524288 cuda threads
[2017-07-13 23:14:41] GPU #2: Intensity set to 19, 524288 cuda threads
[2017-07-13 23:14:41] GPU #3: Intensity set to 19, 524288 cuda threads
[2017-07-13 23:14:45] accepted: 0/1 (diff 8.507), 136.17 MH/s booooo
[2017-07-13 23:14:45] reject reason: expected string or buffer
[2017-07-13 23:14:47] accepted: 0/2 (diff 0.075), 136.23 MH/s booooo
[2017-07-13 23:14:47] reject reason: expected string or buffer
[2017-07-13 23:14:49] accepted: 0/3 (diff 0.113), 135.54 MH/s booooo
[2017-07-13 23:14:49] reject reason: expected string or buffer
[2017-07-13 23:14:55] accepted: 0/4 (diff 0.066), 135.53 MH/s booooo
[2017-07-13 23:14:55] reject reason: expected string or buffer
[2017-07-13 23:14:56] accepted: 0/5 (diff 0.260), 135.29 MH/s booooo
[2017-07-13 23:14:56] reject reason: expected string or buffer
[2017-07-13 23:14:57] accepted: 0/6 (diff 2.611), 135.39 MH/s booooo


Thanks beforehand!

60% power is bleeding edge power reduction. I would be at 70% power especially for lbry, groestl etc. The other settings look fine so I don't see a problem there.

[gzubeck]

The usual blah blah nonsense.
For memory light algos, just use the cuda cores count.

The point was that's not a good metric and you don't get results which are 85% higher even when it's not a memory heavy algo (85% is the different in cuda cores). Originally people were making a case for buying 1080/tis before the pricing for 1070s got all messed up. $400x 1.65 = $660 not $800.

Now it's about on par, but you're sacrificing being able to mine other algos which depend more on memory. That's just NOW, it's possible for new algos to come out which utilize memory and for a 1080 to be crippled in them. I haven't seen a single algo where GDDR5X helps, even though it has more bandwidth then GDDR5.

I'm hoping and praying that nvidia fixes this driver issue in the next month as the gddr5x should be faster memory. Their doing a wonderful job and segmenting off the high end from the low end using the memory bug. Miraculously I think the bug will be fixed after amd introduces their new cards.

[gzubeck]

I am trying to mine LBRY on yiimp with ccminer 2.0 but according to the pool stats I did not find anything for 2 days even if the miner finds. Is there any special settings needed? I have 5 x 1080ti.

Tried yimp for lbry...got half the coins I get on suprnova...suprnova is like lbry central...

[danspasiva]

Can anyone here help me? I simply cannot get the ccminer to work no matter what i try.  It turns on, mines like 15 shares, then crashes with the error message "illegal memory access"

Nothing i've tried works, even with lower intensity.

Running 6x 1070's on windows 10

Maybe try raising your virtual memory up to some crazy amount like 64GB or something.

tried this, nothing is working at the moment i keep getting illegal memory access after the first 10 minutes worth of mining and it crashes.  Literally about to beat my rig up with a baseball bat lol.  Anyone have any solutions to the  "illegal memory access" problem please?

I would try disconnecting all your garaphics cards and start with one primary one. I had a bad install of graphics drivers and some cards we're not working at all. Keep adding one at a time to make sure you don't have any defective equipment or risers. Also, you might want to post your bat file minus any keys to see what your starting up. In addition, what kind of power supply are you using? I'm asking because that could also shut things down if your power supply is not sufficient. I have 4 gtx 1080s with an 850 watt power supply and I would not want to add another card without chaining another power supply to my rig.

turned out it was a bad GPU that im returning.  My other rig gets a completely different problem though, the miner will essentially just freeze and stop mining but look like nothing's wrong until you see that its actually just frozen.  Then when you close it, it causes the whole PC to freeze.  Any ideas?

[gzubeck]

Can anyone here help me? I simply cannot get the ccminer to work no matter what i try.  It turns on, mines like 15 shares, then crashes with the error message "illegal memory access"

Nothing i've tried works, even with lower intensity.

Running 6x 1070's on windows 10

Maybe try raising your virtual memory up to some crazy amount like 64GB or something.

tried this, nothing is working at the moment i keep getting illegal memory access after the first 10 minutes worth of mining and it crashes.  Literally about to beat my rig up with a baseball bat lol.  Anyone have any solutions to the  "illegal memory access" problem please?

I would try disconnecting all your garaphics cards and start with one primary one. I had a bad install of graphics drivers and some cards we're not working at all. Keep adding one at a time to make sure you don't have any defective equipment or risers. Also, you might want to post your bat file minus any keys to see what your starting up. In addition, what kind of power supply are you using? I'm asking because that could also shut things down if your power supply is not sufficient. I have 4 gtx 1080s with an 850 watt power supply and I would not want to add another card without chaining another power supply to my rig.

turned out it was a bad GPU that im returning.  My other rig gets a completely different problem though, the miner will essentially just freeze and stop mining but look like nothing's wrong until you see that its actually just frozen.  Then when you close it, it causes the whole PC to freeze.  Any ideas?

Same thing...either bad card or an incomplete install of one of your graphics cards...start by installing one graphics card and then add a couple more until you find the culprit (Bad card or riser), or the windows software completely installs all the cards.

[heyz]

Hello,

Does anyone know why I'm getting those errors? It's GTX 1070 card. And the MSI overclock settings are:
Core +80; Mem: +400; power: 60; Temp: 85; Fan auto. The setting works fine with other miners.


>ccminer-x64.exe -q -a groestl -o stratum+tcp://erebor.dwarfpool.com:3345 -u xxx.CAT-2 -p 1
*** ccminer 2.0 for nVidia GPUs by tpruvot@github ***
[2017-07-13 23:14:39] Starting on stratum+tcp://erebor.dwarfpool.com:3345
[2017-07-13 23:14:39] 4 miner threads started, using 'groestl' algorithm.
[2017-07-13 23:14:41] Stratum difficulty set to 16 (0.06250)
[2017-07-13 23:14:41] groestl block 1669538, diff 716.480
[2017-07-13 23:14:41] GPU #0: Intensity set to 19, 524288 cuda threads
[2017-07-13 23:14:41] GPU #1: Intensity set to 19, 524288 cuda threads
[2017-07-13 23:14:41] GPU #2: Intensity set to 19, 524288 cuda threads
[2017-07-13 23:14:41] GPU #3: Intensity set to 19, 524288 cuda threads
[2017-07-13 23:14:45] accepted: 0/1 (diff 8.507), 136.17 MH/s booooo
[2017-07-13 23:14:45] reject reason: expected string or buffer
[2017-07-13 23:14:47] accepted: 0/2 (diff 0.075), 136.23 MH/s booooo
[2017-07-13 23:14:47] reject reason: expected string or buffer
[2017-07-13 23:14:49] accepted: 0/3 (diff 0.113), 135.54 MH/s booooo
[2017-07-13 23:14:49] reject reason: expected string or buffer
[2017-07-13 23:14:55] accepted: 0/4 (diff 0.066), 135.53 MH/s booooo
[2017-07-13 23:14:55] reject reason: expected string or buffer
[2017-07-13 23:14:56] accepted: 0/5 (diff 0.260), 135.29 MH/s booooo
[2017-07-13 23:14:56] reject reason: expected string or buffer
[2017-07-13 23:14:57] accepted: 0/6 (diff 2.611), 135.39 MH/s booooo


Thanks beforehand!

60% power is bleeding edge power reduction. I would be at 70% power especially for lbry, groestl etc. The other settings look fine so I don't see a problem there.

Thanks for suggestion. I've bumped the power up to 85%, but still getting same errors. weird

[joblo]

Hello,

Does anyone know why I'm getting those errors? It's GTX 1070 card. And the MSI overclock settings are:
Core +80; Mem: +400; power: 60; Temp: 85; Fan auto. The setting works fine with other miners.

Your shares are being rejected by the pool. Try different coins/algos/pools to get a better definition of the problem.
Remove -q and add -D -P when troubleshooting.

You could also try asking dwarfpool what that error means, it's not a typical reject.

[sidog]

--plimit=150,150,150,....

7x 1080ti wow lol

muchas gracias   that worked like a charm.   

I'm running them at 65% TDP (1080 TI ZOTAC AMP EXTREME).  These suckers have max TDP of 320w!  But they run cool, quiet, and have been rock solid for a month.

[maxmad_x]

I heard there is a tpruvot 2.1 in the works .. is it true? I am fan of your work dude. Keep it up and keep it open sourced. Whenever I can I will send the donation Cheers

[bensam1231]

The usual blah blah nonsense.
For memory light algos, just use the cuda cores count.

The point was that's not a good metric and you don't get results which are 85% higher even when it's not a memory heavy algo (85% is the different in cuda cores). Originally people were making a case for buying 1080/tis before the pricing for 1070s got all messed up. $400x 1.65 = $660 not $800.

Now it's about on par, but you're sacrificing being able to mine other algos which depend more on memory. That's just NOW, it's possible for new algos to come out which utilize memory and for a 1080 to be crippled in them. I haven't seen a single algo where GDDR5X helps, even though it has more bandwidth then GDDR5.

I'm hoping and praying that nvidia fixes this driver issue in the next month as the gddr5x should be faster memory. Their doing a wonderful job and segmenting off the high end from the low end using the memory bug. Miraculously I think the bug will be fixed after amd introduces their new cards.

It's not a bug. It's a latency issue, which is why it doesn't work well with Dagger. Nvidia can't fix it unless they redesign their cards, which they aren't going to do. We need to wait for nextgen of Nvidia products and hopefully it wont be bugged with that.