Topic: [ANN][BURST] Burst | Efficient HDD Mining | New 1.2.3 Fork block 92000 - page 1241. (Read 2170889 times)

so, to continue.... 1.3Tb over 3 seperate pc's, for 2 days, found 1 block on the last machine I added with the smallest -lot of 50Gb.





 

passphrases is your login ... you can not change it without getting a new address ... they are bound to each other!

Sorry you misunderstood - I spent $70 figuring this out. It would cost $20 to generate 1 TB in 2.5 hours (instead of a couple of days so you are mining at full pelt quicker).  Once it is generated, it would cost around $15 a month for the server and $100 a month for storage.  And at current rates that would probably still be profitable

hi all!!
just to understand browser wallet  http://localhost:8125/.
i login into wallet browser with my passphrases.txt code.
when i do "copy numeric account id" i correctly found the number generated with run_generate.bat and written inside address.txt.
but i dont understand why my account ID "es.: BURST-XXXX-YYYY-ZZZZ" change every time i logoff wallet browser and i do a new login.
This behaviour is correct?
or i do something wrong?
thanks!!

It is, but you get use to it when it comes to solo mining crypto-currency. Seeing people with 30GB to 80 GB hdd's finding a block to 3 blocks within 48 hours is just surreal at times....

i guess that's the frustrating thing about this coin. it does seem like luck has a lot to do with it. it would be interesting to get a poll going to see what the average number of block found per miner is so far.

What is "Generate token" in the wallet home page!?

any exchanges now?

Thanks also paul. I would send you some, but no luck for the pass 72 hours plus mining. These guys with smaller hdd's are getting block left and right with 48 hours....

why u need NFS?,
1. first u create large instance 8 core or more
2. mount EBS to that instance
3. generate plot, while generating u r mining from it too (for 1 TB it will take about half day)
4. when plot generated, detach EBS, and shutdown + remove that instance
5. create new micro instance, attach that EBS and mining from it

this will cost you about 60$ / TB / Month

anyway, azure is cheaper, you can get it free, but i can't and won't tell you how

thanks for sharing. sent you some for your efforts.

At $70 per day, that's $2100 per month. You can nearly buy a mobo, a used 6/12 core (12/24 threads) Intel cpu, Ram and some HDD's for that price, give or take....LOL

If anyone is interested, the conclusion of my test to see if I could speed up plot generation by farming it out using cluster/cloud computing is:
Yes it can be done, but it is probably not worth it.  This is what I did using Amazon EC2...

First I created a single instance (2 core) to act as the server for the plot storage and also act as the wallet server / miner.
This instance had a 1TB Elastic Block Store volume attached to it (1TB is max EBS volume size on AWS). I made this volume available to the network using NFS

I then created a "plot generator" instance with much higher CPU capacity (C3.xlarge 8 core). I mounted the shared volume so it appeared like a local disk and started creating 10GB plots
I then created a few scripts so that upon startup the instance would auto-mount the NFS volume, figure out what the last used nonce range was, and start mining from the next 10GB plot
Finally, I created an image of this instance and spooled up 100 more instances like it (spot rates, nice and cheap).

To begin with it all looked great but I soon hit the bandwidth limit of the instance I was using as the server.  In essence the server didn't have enough network bandwidth to cope with all of the data being copied to it.  This restricted plot creation to a little over 100GB / hour.  Changing the instance type of the server instance to one with a "high" network performance got this up to around 400 GB/hour.  If I had have left it running I would have created all 1 TB of plots in 2.5 hours, at a cost of around $20.  This could have been repeated or even parallelized to create several 1TB volumes full of plots in a relatively short amount of time.

After all the fiddling about getting it working, I spent around $70 testing this stuff out, only to conclude it was too much hassle.  And as OP said earlier, the cost of EBS storage on AWS is too high to make this a viable long term mining option.  That said, had I have done all of this on day one, I would have scaled the process out and had 10TB+ up within the first few hours and dominated the mining.  Good to know I have the process licked for the next PoC coin

Unfortunately I STILL don't have any BURST as I don't have any spare capacity at home so haven't been able to mine at all  So if anyone would like to donate some coins to offset the cost of testing all of this, my address is BURST-WADY-CBZE-HSJU-2NH5G

Many thanks!

1201 - Thu Aug 14 2014 13:35:28 GMT+0200 - 67567251495709
1502 - Fri Aug 15 2014 13:35:04 GMT+0200 - 111403016327443

diff increased 65% in last 24h ... that means network hd capacity rises by the same factor ... correct?!


should make no difference ... but for every block, some data has to be read by miner from every plot file, so if you have a lot, thats slower than reading from one file ... but 1-10 or some more are no problem.

if you once need more space for 'real' data, its a good thing to have multiple plot files ... so you can simple delete one of them :-)




_{btw. selling some BURST: https://bitcointalksearch.org/topic/m.8330476}

So is it better to have one big plot on a hdd, or many plots?

Pre generated saves alot of power though so it really comes down to gpu speed

it is possible, but considering the size of each nonce 256KB compared to GPU style about 4-16Bytes, the possible combination is too huge.

assuming speed of generating hash on GPU is 1 MH/s, this means we can try 0.24 Giga-Nonce per block time

compare it if we have 4 TB disk, it store 1 Giga-Nonce which we can try all of them per block time, resulting
in storing nonce on disk is faster 4x than hashing it "on the fly" on GPU, ofcourse my assumption of 1 MH/s is wild guess...

before we have exact value how much nonce can be generated on GPU per seconds, we can't come to conclusion which one is more efficient, hash on the fly or pre-generate it on disk

Found a block about 26 hours since last found block on 800gb. 

Alias registration.

What is the matter with all these transactions to BURST-2222-2222-2222-22222?