[ANN][BURST] Burst | Efficient HDD Mining | New 1.2.3 Fork block 92000 - page 611.

Yanakitu Tenatako

sr. member

Activity: 328

Merit: 250

Quote from: majere on October 28, 2014, 12:39:06 PM

Quote from: Yanakitu Tenatako on October 28, 2014, 12:34:56 PM

Hmmm..... Must be than a CPU making that difference? I am using A4-7300.
Will get better one for testing. Hdds are fastest I could get. WD purple.

I cannot run latest version of your miner, it is crashing.

Mining process isn't CPU-intensive.

Blago's miner tries to read all files in parallel. You'll have to merge them into one file per HDD or use a different miner which reads them sequentially.

While reading plots CPU is locked at 100% as long as reading goeas on, after that, goes to almost 0%.
So, maybe reading of HDD actually IS CPU sensitive?

Just a thought....

Yanakitu Tenatako

sr. member

Activity: 328

Merit: 250

Quote from: majere on October 28, 2014, 12:39:06 PM

Quote from: Yanakitu Tenatako on October 28, 2014, 12:34:56 PM

Hmmm..... Must be than a CPU making that difference? I am using A4-7300.
Will get better one for testing. Hdds are fastest I could get. WD purple.

I cannot run latest version of your miner, it is crashing.

Mining process isn't CPU-intensive.

Blago's miner tries to read all files in parallel. You'll have to merge them into one file per HDD or use a different miner which reads them sequentially.

Blago gave an example on his picture, you can read he is having 8-11 files per hdd.
Why then I an getting dofferent woth same structure?
He is not using 1 file per hdd also.

majere

newbie

Activity: 44

Merit: 0

Quote from: Yanakitu Tenatako on October 28, 2014, 12:34:56 PM

Hmmm..... Must be than a CPU making that difference? I am using A4-7300.
Will get better one for testing. Hdds are fastest I could get. WD purple.

I cannot run latest version of your miner, it is crashing.

Mining process isn't CPU-intensive.

~~Blago's miner tries to read all files in parallel. You'll have to merge them into one file per HDD or use a different miner which reads them sequentially.~~
Sorry, not true.

Yanakitu Tenatako

sr. member

Activity: 328

Merit: 250

Generally, stagger size is ok?

Blago

sr. member

Activity: 416

Merit: 250

Quote from: Yanakitu Tenatako on October 28, 2014, 12:29:59 PM

Here is a structure of a 4Tb drive:

I made plots with GPUPlotGenerator 3.0.0:

gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101000000000 2048000 8192
gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101025000000 2048000 8192
gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101050000000 2048000 8192
gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101075000000 2048000 8192
gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101100000000 2048000 8192
gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101125000000 2048000 8192
gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101150000000 2048000 8192
gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101175000000 901120 8192

I assume stagger size is 8192 ?

It takes from 130 to 200 seconds for reading with Blago miner.
I have tried with 32Gb ram, does not makes any difference.

With 8Gb ram it wont even read plots. Miner is breaking system apart.

My 2nd rig have 8Gb RAM...
You can merge plots for fast reading. Use dcct's "optimizer", or https://burstforum.com/index.php?threads/plot-optimizer-v1-5-j6jq.268/

Quote from: Yanakitu Tenatako on October 28, 2014, 12:34:56 PM

I cannot run latest version of your miner, it is crashing.

PM me logs (if you not solo mode)

Yanakitu Tenatako

sr. member

Activity: 328

Merit: 250

Quote from: Blago on October 28, 2014, 12:24:39 PM

Quote from: Yanakitu Tenatako on October 28, 2014, 10:22:10 AM

Using Blago mi ner, for 26Tb plot, it takes 200 seconds (more or less) and CPU is @ 100% use.
Having 16Gb RAM.

IS it OK? I think it is long time.

11,5 Tb, 26 sec, AMD FX 6300, 16Gb RAM

Hmmm..... Must be than a CPU making that difference? I am using A4-7300.
Will get better one for testing. Hdds are fastest I could get. WD purple.

I cannot run latest version of your miner, it is crashing.

majere

newbie

Activity: 44

Merit: 0

Quote from: hero18688 on October 28, 2014, 12:09:26 PM

Is there any significant difference of mining performance between 5400rpm hd and ssd if they are of the same capacity?

Not really as long as HDD is being read in time without skipping the blocks. Better to buy more space instead. Big capacity USB2 devices might be too slow though. Some 5400 rpm hdds are faster than 7k btw.

Yanakitu Tenatako

sr. member

Activity: 328

Merit: 250

Here is a structure of a 4Tb drive:

I made plots with GPUPlotGenerator 3.0.0:

gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101000000000 2048000 8192
gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101025000000 2048000 8192
gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101050000000 2048000 8192
gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101075000000 2048000 8192
gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101100000000 2048000 8192
gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101125000000 2048000 8192
gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101150000000 2048000 8192
gpuplotgenerator.exe generate "D:\plots" 3000*************** 10101175000000 901120 8192

I assume stagger size is 8192 ?

It takes from 130 to 200 seconds for reading with Blago miner.
I have tried with 32Gb ram, does not makes any difference.

With 8Gb ram it wont even read plots. Miner is breaking system apart.

Pilotseye

full member

Activity: 210

Merit: 100

Quote from: hero18688 on October 28, 2014, 12:09:26 PM

Is there any significant difference of mining performance between 5400rpm hd and ssd if they are of the same capacity?

only the fact that you will never get ROI with SSD, so no need to put your text in supersize Wink

Blago

sr. member

Activity: 416

Merit: 250

Quote from: Yanakitu Tenatako on October 28, 2014, 10:22:10 AM

Using Blago mi ner, for 26Tb plot, it takes 200 seconds (more or less) and CPU is @ 100% use.
Having 16Gb RAM.

IS it OK? I think it is long time.

11,5 Tb, 26 sec, AMD FX 6300, 16Gb RAM

hero18688

sr. member

Activity: 392

Merit: 250

Is there any significant difference of mining performance between 5400rpm hd and ssd if they are of the same capacity?

majere

newbie

Activity: 44

Merit: 0

A simpler diagram for non-techies. Grin

The image is in public domain. Author's wallet: BURST-6MZ2-6MH9-U5AX-FNMJX

http://i61.tinypic.com/2rw5n5t.png

DMaster2008

member

Activity: 66

Merit: 10

Quote from: Yanakitu Tenatako on October 28, 2014, 10:22:10 AM

Using Blago mi ner, for 26Tb plot, it takes 200 seconds (more or less) and CPU is @ 100% use.
Having 16Gb RAM.

IS it OK? I think it is long time.

It depends on the stagger size. What staggers do you use?

Yanakitu Tenatako

sr. member

Activity: 328

Merit: 250

Using Blago mi ner, for 26Tb plot, it takes 200 seconds (more or less) and CPU is @ 100% use.
Having 16Gb RAM.

IS it OK? I think it is long time.

uray

hero member

Activity: 1400

Merit: 505

Quote from: burstcoin on October 28, 2014, 05:51:57 AM

Quote from: uray on October 28, 2014, 03:23:14 AM

Quote from: uray on October 28, 2014, 02:57:02 AM

Quote from: burstcoin on October 27, 2014, 08:00:17 PM

Quote from: uray on October 27, 2014, 10:44:49 AM

Quote from: daWallet on October 27, 2014, 08:33:40 AM

Burst arrived in Japan

http://burstcoin.jp/
http://wiki.burstcoin.jp/

if you want better diagram, i have this

...

or maybe dev can replace his diagram on OP with this one, i think its describe better ( just my suggestion )

That chart is really nice, and I would use it, however it has a few errors. prev gensig and accountid that mined the last block is hashed to get the new gensig. The gensig is then hashed with the height before mod 4096 to get the scoop number. The new gensig is the one hashed with the scoop to get the deadline.

maybe what i mean by prev gensig, is the new gensig by what u said, because i see it on miner and pool perspective, i never dealt with prevGensig and block generator accountId.

so, if i rename "prevGensig" with "newGensig" on the diagram, does it look correct?

revised diagram :

gensig % 4096 should say genhash % 4096. Other than that I think it looks correct.

and also dev, i want to create some infographics about burst advantage

we see that on the diagram we dont dealt with transaction or hashing the transaction during mining (or maybe during block creation too), how does it affect 51% attack that most PoW coins suffer, because selfist miner can selectively ignore or delay transaction or executing double spend when he got more than 50% hash power, does it affect burst too?

i know you had this conversation before, but its hard to look for and i am not really grasp the matter

uray

hero member

Activity: 1400

Merit: 505

Quote from: burstcoin on October 28, 2014, 05:51:57 AM

Quote from: uray on October 28, 2014, 03:23:14 AM

Quote from: uray on October 28, 2014, 02:57:02 AM

Quote from: burstcoin on October 27, 2014, 08:00:17 PM

Quote from: uray on October 27, 2014, 10:44:49 AM

Quote from: daWallet on October 27, 2014, 08:33:40 AM

Burst arrived in Japan

http://burstcoin.jp/
http://wiki.burstcoin.jp/

if you want better diagram, i have this

...

or maybe dev can replace his diagram on OP with this one, i think its describe better ( just my suggestion )

That chart is really nice, and I would use it, however it has a few errors. prev gensig and accountid that mined the last block is hashed to get the new gensig. The gensig is then hashed with the height before mod 4096 to get the scoop number. The new gensig is the one hashed with the scoop to get the deadline.

maybe what i mean by prev gensig, is the new gensig by what u said, because i see it on miner and pool perspective, i never dealt with prevGensig and block generator accountId.

so, if i rename "prevGensig" with "newGensig" on the diagram, does it look correct?

revised diagram :

gensig % 4096 should say genhash % 4096. Other than that I think it looks correct.

ah yeah, i forgot to rename that, will re-upload

burstcoin

sr. member

Activity: 280

Merit: 250

Quote from: uray on October 28, 2014, 03:23:14 AM

Quote from: uray on October 28, 2014, 02:57:02 AM

Quote from: burstcoin on October 27, 2014, 08:00:17 PM

Quote from: uray on October 27, 2014, 10:44:49 AM

Quote from: daWallet on October 27, 2014, 08:33:40 AM

Burst arrived in Japan

http://burstcoin.jp/
http://wiki.burstcoin.jp/

if you want better diagram, i have this

...

or maybe dev can replace his diagram on OP with this one, i think its describe better ( just my suggestion )

That chart is really nice, and I would use it, however it has a few errors. prev gensig and accountid that mined the last block is hashed to get the new gensig. The gensig is then hashed with the height before mod 4096 to get the scoop number. The new gensig is the one hashed with the scoop to get the deadline.

maybe what i mean by prev gensig, is the new gensig by what u said, because i see it on miner and pool perspective, i never dealt with prevGensig and block generator accountId.

so, if i rename "prevGensig" with "newGensig" on the diagram, does it look correct?

revised diagram :

gensig % 4096 should say genhash % 4096. Other than that I think it looks correct.

Nevril

member

Activity: 108

Merit: 10

Quote from: uray on October 27, 2014, 10:50:24 AM

Quote from: Nevril on October 27, 2014, 09:31:16 AM

So guys...

I'm currently developing a method to sustain the BURST price on the long run.
It isn't time to share any detail, I will in the (hopefully) near future. So, try to be patient...

What I currently need, is to know if there is a Bitcoin client which supports features like the Asset Exchange of BURST (in BTC obviously) or, even better, a crowdfounding mechanism with the possibility to the founders to also attach text messages to the donation.

Mind that the WIP method doesn't strictly depend on this, but it would make its implementation a little bit easier

you might want to look at XCP (Counterparty) it has asset exchange and it linked with bitcoin by nature
take a look at : http://www.blockscan.com/

Thanks for the answer uray.
I've seen XCP, it doesn't help me, actually it makes things a little bit more complicated.

I've given a look to the lighthouse project built on top of bitcoin, too. Nice one, too bad it's in alpha stage Sad

I think I'll propose a different (and hopefully easier) way of doing what I have in mind, probably using burstforum to keep everything in one thread with Q&A too and some "advertise" in this thread to let everyone know. We'll see where we get soon...-ish

uray

hero member

Activity: 1400

Merit: 505

Quote from: uray on October 28, 2014, 02:57:02 AM

Quote from: burstcoin on October 27, 2014, 08:00:17 PM

Quote from: uray on October 27, 2014, 10:44:49 AM

Quote from: daWallet on October 27, 2014, 08:33:40 AM

Burst arrived in Japan

http://burstcoin.jp/
http://wiki.burstcoin.jp/

if you want better diagram, i have this

...

or maybe dev can replace his diagram on OP with this one, i think its describe better ( just my suggestion )

That chart is really nice, and I would use it, however it has a few errors. prev gensig and accountid that mined the last block is hashed to get the new gensig. The gensig is then hashed with the height before mod 4096 to get the scoop number. The new gensig is the one hashed with the scoop to get the deadline.

maybe what i mean by prev gensig, is the new gensig by what u said, because i see it on miner and pool perspective, i never dealt with prevGensig and block generator accountId.

so, if i rename "prevGensig" with "newGensig" on the diagram, does it look correct?

revised diagram :

uray

hero member

Activity: 1400

Merit: 505

Quote from: burstcoin on October 27, 2014, 08:00:17 PM

Quote from: uray on October 27, 2014, 10:44:49 AM

Quote from: daWallet on October 27, 2014, 08:33:40 AM

Burst arrived in Japan

http://burstcoin.jp/
http://wiki.burstcoin.jp/

if you want better diagram, i have this

...

or maybe dev can replace his diagram on OP with this one, i think its describe better ( just my suggestion )

That chart is really nice, and I would use it, however it has a few errors. prev gensig and accountid that mined the last block is hashed to get the new gensig. The gensig is then hashed with the height before mod 4096 to get the scoop number. The new gensig is the one hashed with the scoop to get the deadline.

maybe what i mean by prev gensig, is the new gensig by what u said, because i see it on miner and pool perspective, i never dealt with prevGensig and block generator accountId.

so, if i rename "prevGensig" with "newGensig" on the diagram, does it look correct?

Topic: [ANN][BURST] Burst | Efficient HDD Mining | New 1.2.3 Fork block 92000 - page 611. (Read 2171065 times)