[ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 120.

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: tarzanbigcity on September 03, 2014, 01:35:03 PM

I would much rather waste hdd space than waste money for power. As far as returns go, hdd mining pays off in spades. Once you reach ROI on it, it becomes free money because your not paying for the upkeep of the system by paying inflated power costs every month. A Hdd consumes what 6watts at peak I/O? It scales fairly inexpensively. Also your not limited to 6 PCIe lanes like you would if you were gpu mining. Multichannel raid adapters allow you to keep expanding until your physical space runs out.

So you are buying something you don't need in hope to make ROI... that's kind of silly to me
I usually don't buy vast amount of space, because I know that each years the price decrease roughly by a factor 2,
so technically you are not making any money at all, you are just buying large HDD in advance you don't even need ignoring the
fact that you could have bought them in a couple of year for a quarter of the price... Grin

tarzanbigcity

sr. member

Activity: 602

Merit: 250

I would much rather waste hdd space than waste money for power. As far as returns go, hdd mining pays off in spades. Once you reach ROI on it, it becomes free money because your not paying for the upkeep of the system by paying inflated power costs every month. A Hdd consumes what 6watts at peak I/O? It scales fairly inexpensively. Also your not limited to 6 PCIe lanes like you would if you were gpu mining. Multichannel raid adapters allow you to keep expanding until your physical space runs out.

Amph

legendary

Activity: 3248

Merit: 1070

Quote from: djm34 on September 03, 2014, 10:42:12 AM

Quote from: bigjme on September 03, 2014, 10:38:45 AM

If you can select how much space to use per drive and which drives to use, i would consider it.
But i don't fancy dedicating a load of drives purely to that

Not mentioning that now that the coin made it to this thread the profitability is probably dead Grin

very unlikely, because hdd consume next to nothing, compared to gpu, it's still very profitable and always will be, unless the value reach 0 or something

i mean 1 3tb hd consume 6w lol

bathrobehero

legendary

Activity: 2002

Merit: 1051

ICO? Not even once.

While you can have plots with specific sizes and directory linking might work, I think I'm too late to go further into this coin. I can't imagine it's not way overpriced. I'm not sure if the plots are just solely for PoW purposes or they are eventually getting filled with data as people rent this cloud storage but if it would I can't imagine my 2 Mbps upload speed would very much like it.

On another note, you guys might want to check up on good ol' maxcoin and get some before the reward decreases essentially 1/6th of what it is now even if it's only going to end up being one big pump.

Also, I'm curious how many of you will jump on the 900 series cards solely for mining considering how terrible the altcoin scene for miners is lately.

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: bigjme on September 03, 2014, 10:47:01 AM

very true, this thread is death to all profitable coins, and that probably isn't even a joke

too bad, it wasn't even an nvidia coin Grin

bigjme

sr. member

Activity: 350

Merit: 250

very true, this thread is death to all profitable coins, and that probably isn't even a joke

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: bigjme on September 03, 2014, 10:38:45 AM

If you can select how much space to use per drive and which drives to use, i would consider it.
But i don't fancy dedicating a load of drives purely to that

Not mentioning that now that the coin made it to this thread the profitability is probably dead Grin

bigjme

sr. member

Activity: 350

Merit: 250

If you can select how much space to use per drive and which drives to use, i would consider it.
But i don't fancy dedicating a load of drives purely to that

Amph

legendary

Activity: 3248

Merit: 1070

Quote from: djm34 on September 03, 2014, 08:30:21 AM

Quote from: yellowduck2 on September 03, 2014, 08:21:52 AM

once the HDD is fully plot , CPU will no longer be use ?

so now, we say fully plot rather than full ? Grin

edit: clearly the reason I am not attracted to that coin which just write some mindless crap to hdd... (I don't need a coin to do that I can do it myself... Grin

)
if at least what was written had some use...

with gpu you waste electricity, here you waste "space" Grin

Quote from: yellowduck2 on September 03, 2014, 08:21:52 AM

once the HDD is fully plot , CPU will no longer be use ?

it still used but the % is tiny

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: unitedminers on September 03, 2014, 09:28:47 AM

Yesterday I've tried to compile ccMiner like always ...

cd ccminer-djm34-0901
chmod +x configure autogen.sh

"This is easily fixed by adding at the end to Makefile.am"
AUTOMAKE_OPTIONS = subdir-objects

./autogen.sh
./configure
make

... but now I get the following error Huh

./cuda_helper.h(493): error: more than one instance of overloaded function "__shfl" matches the argument list:
   function "__shfl(int, int, int)"
   function "__shfl(float, int, int)"
   argument types are: (uint32_t, int)

./cuda_helper.h(494): error: more than one instance of overloaded function "__shfl" matches the argument list:
   function "__shfl(int, int, int)"
   function "__shfl(float, int, int)"
   argument types are: (uint32_t, int)

2 errors detected in the compilation of "/tmp/tmpxft_00001fc8_00000000-12_cuda_blake512.compute_50.cpp1.ii".
Makefile:1666: recipe for target 'heavy/cuda_blake512.o' failed
make[2]: *** [heavy/cuda_blake512.o] Error 2
make[2]: Leaving directory '/home/user/ccminer-djm34-0901'
Makefile:1229: recipe for target 'all-recursive' failed
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory '/home/user/ccminer-djm34-0901'
Makefile:453: recipe for target 'all' failed
make: *** [all] Error 2

You need to use cuda 6.5
I compiled it on aws (ubuntu 14) without any problem.
(using git clone rather than the zip file)
actually you can remove the line 489 to 497 in cuda_helper.h as it isn't used

unitedminers

full member

Activity: 241

Merit: 100

Yesterday I've tried to compile ccMiner like always ...

cd ccminer-djm34-0901
chmod +x configure autogen.sh

"This is easily fixed by adding at the end to Makefile.am"
AUTOMAKE_OPTIONS = subdir-objects

./autogen.sh
./configure
make

... but now I get the following error Huh

./cuda_helper.h(493): error: more than one instance of overloaded function "__shfl" matches the argument list:
function "__shfl(int, int, int)"
function "__shfl(float, int, int)"
argument types are: (uint32_t, int)

./cuda_helper.h(494): error: more than one instance of overloaded function "__shfl" matches the argument list:
function "__shfl(int, int, int)"
function "__shfl(float, int, int)"
argument types are: (uint32_t, int)

2 errors detected in the compilation of "/tmp/tmpxft_00001fc8_00000000-12_cuda_blake512.compute_50.cpp1.ii".
Makefile:1666: recipe for target 'heavy/cuda_blake512.o' failed
make[2]: *** [heavy/cuda_blake512.o] Error 2
make[2]: Leaving directory '/home/user/ccminer-djm34-0901'
Makefile:1229: recipe for target 'all-recursive' failed
make[1]: *** [all-recursive] Error 1
make[1]: Leaving directory '/home/user/ccminer-djm34-0901'
Makefile:453: recipe for target 'all' failed
make: *** [all] Error 2

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: gadado on September 03, 2014, 08:44:40 AM

Hmm scales with HDD capacity..

Is there a coin that scales with the Memory capacity too? (I know Memory Coin but that looks to be dead)
I have access to a PC with a lot of Memory.

shinycoin min mem requirement 32Gb (actually min mem requirement is strictly greater than 16Gb)

gadado

hero member

Activity: 965

Merit: 515

Hmm scales with HDD capacity..

Is there a coin that scales with the Memory capacity too? (I know Memory Coin but that looks to be dead)
I have access to a PC with a lot of Memory.

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: yellowduck2 on September 03, 2014, 08:31:44 AM

Quote from: djm34 on September 03, 2014, 08:30:21 AM

Quote from: yellowduck2 on September 03, 2014, 08:21:52 AM

once the HDD is fully plot , CPU will no longer be use ?

so now, we say fully plot rather than full ? Grin

not sure if HDD is full will throw up Cool

only if it is a roman HDD

yellowduck2

hero member

Activity: 868

Merit: 1000

Quote from: djm34 on September 03, 2014, 08:30:21 AM

Quote from: yellowduck2 on September 03, 2014, 08:21:52 AM

once the HDD is fully plot , CPU will no longer be use ?

so now, we say fully plot rather than full ? Grin

not sure if HDD is full will throw up Cool

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: yellowduck2 on September 03, 2014, 08:21:52 AM

once the HDD is fully plot , CPU will no longer be use ?

so now, we say fully plot rather than full ? Grin

edit: clearly the reason I am not attracted to that coin which just write some mindless crap to hdd... (I don't need a coin to do that I can do it myself... Grin

)
if at least what was written had some use...

yellowduck2

hero member

Activity: 868

Merit: 1000

once the HDD is fully plot , CPU will no longer be use ?

AizenSou

hero member

Activity: 938

Merit: 1000

Quote from: Amph on September 03, 2014, 07:08:15 AM

Quote from: AizenSou on September 03, 2014, 06:55:49 AM

Source is available and it's less than 1000 lines of code. Someone should be able to compile it in Windows for you. Sorry but I don't have Windows.

can you point me to the source? tnx

New miner in C
https://s3-ap-southeast-1.amazonaws.com/burst-mirror/dcct-pool-miner.tgz
Plot generator in C
https://bchain.info/plotgenerator.tgz
Plot range checker:
https://bchain.info/BURST/tools/overlap
Plot merge tool(significantly reduce disk stress, very useful if you generate your plot in java so you could replot your plots in bigger stagger size with significant higher speed)
https://bchain.info/merge.c
All credits to this dcct guy.

Quote from: djm34 on September 03, 2014, 07:15:44 AM

java isn't that bad... robots on Mars are operated in java

Java is not bad but for handling big memory task it's just not designed for it. LOL

bigjme

sr. member

Activity: 350

Merit: 250

Wait its a java applet that creates files? does it even need a wallet?

djm34

legendary

Activity: 1400

Merit: 1050

Quote from: Amph on September 03, 2014, 05:50:07 AM

Quote from: djm34 on September 03, 2014, 05:16:00 AM

Quote from: Amph on September 03, 2014, 04:50:01 AM

Quote from: yellowduck2 on September 03, 2014, 04:39:03 AM

Quote from: Amph on September 03, 2014, 04:37:42 AM

Quote from: yellowduck2 on September 03, 2014, 04:35:08 AM

Quote from: Amph on September 03, 2014, 04:34:07 AM

Quote from: yellowduck2 on September 03, 2014, 04:29:56 AM

Quote from: Amph on September 03, 2014, 04:28:20 AM

Quote from: bigjme on September 03, 2014, 04:23:01 AM

Honestly i have a good 50 or 60TB worth of empty drives sitting about, if its good to mine i may try it.
I think it will be more profitable over time as some plots that are generated at the start may be wanted much later on, unless i'm understanding it wrong

and SSD's are only limited by their writing life, which is pretty long

3TB = 5k burst, your 60 will do 100k day(right now) = 0.7 btc daily

If he can make a GPU miner to plot , i guess it could be 7 BTC per day for a while

why? he can just make plot faster, the the paid amount would be the same, he always mine with hdd not gpu

i am guessing faster plot = more payout = more btc ?

yeah but it's still 0.7 daily(with 60), you just make them early, without waiting so long for the cpu miner to finish

u mean the payout is fix no mater how fast the plot ? is it something like 1 x CPU and 100 x CPU will get the same number of coins per day with same HDD size ? i don't full understand how it works

if so, that means its paying based on space per day

basically it's like this:

you need to fill your terabyte first before mining, your space is your hashpower, but they must be filled completely if you want to use them all, the gpu miner only fill those tera faster, so you can mine early with all your power(in terabyte)

it's like a car and her gasoline, the more gas you have(terabyte) the more km you will do(coin mined), but you need to fill your call with it first, and this process take too much time with cpu, this is why we need a gpu "plotter"

First of, I must say that I have no interest in this coin so I didn't look much into detail...
But unless I am totally mistaken, gpu's never write directly to hdd...
Gpu usually does its his work transmit what it has done to the cpu who would eventually write to hdd. So if on burst the whole thing is just "writing to the disk" well this isn't a gpu thing at all...

You want to improve your perf on Burst: Better cpu, Better Mobo with more SataIII and better HDD supporting high speed transfer... for the rest...
So you need to build a disk server which can be access by many cpu at a time: means huge amount of memory for buffering and fast drive, fast cpu...

(at that level, I think you can do that as well on aws)

you may be like you have said, but the process can be speeded up with a c code instead of java

java isn't that bad... robots on Mars are operated in java

Topic: [ANN] cudaMiner & ccMiner CUDA based mining applications [Windows/Linux/MacOSX] - page 120. (Read 3426944 times)