Topic: [ANN][BURST] Burst | Efficient HDD Mining | New 1.2.3 Fork block 92000 - page 946. (Read 2171083 times)

I'm just going off of the displayed nonces/minute.
The filesize on Windows appears to remain at 0kb (only actually displays the size if I kill the plotting, strange... but it appears to be plotting just fine)
I haven't sat down and actually timed linux or windows, but they seemed about the same (to me at least)

Right now I have the Dual Xeon (8cores on Windows) plotting a 2TB file and my AMD Phenom II (4cores on Linux) Plotting a 4TB file... we will see who wins

Watched the file size growing on the Linux (4core) box and it is growing at ~0.65GB/minute...
~0.65GB/minute * 60 = 39GB/hour
4000 / 39 = 102 hours for 4TB
102 hours = 4.25/days

Yeah, this may take a while

sure. I tell share your story on Burst Conference / Summit 2015.

shouldn't plot2 start from 405000? since 400,000 was already plotted?

Also just wondering, how do you calculate nonce/min? I got mine set a 1000 but its not doing 1000 per minute takes a couple of minutes for a 1000. How do you guys know which nonce to set? Trial and error? or just set it as high as possible?

Thanks.

So. Why my xeon with 8 core male just 1000 nonché a minute? I nave set 8 in generate.

To convert your bat on exe, use the free software "bat to exe converter", then to relaunch your exe when crash, use "knas restarter" free too.

On wich pool you are?

i guess java showing 3200 nonces/minute is the fail.
have you compared the filesize and time needed for creation?
i only plot on linux and i still get 2x faster plots compared on time needed to create the files.

Glad you enjoyed it. Figured I'd share with the community... and tell it at parties and whatnot.

AMD Phenom II X4 965 Black Edition Quad-Core 3.4GHz and 4GB RAM

I also tried on an old server board I had laying around with 8 cores (2 CPUs, both Quad-Core Xeons @ 1.6Ghz) and that is only slightly faster than the above system at ~3500 nonces/minute

You plot numbers would actually be this:
{WalletNumber} 0 400000 {stagger} {cores}
{WalletNumber} 400000 400000 {stagger} {cores}
{WalletNumber} 800000 400000 {stagger} {cores}
{WalletNumber} 1200000 400000 {stagger} {cores}

ex:
Plot1: 11111222223333344444 0 400000 5000
Plot2: 11111222223333344444 400000 400000 5000
Plot3: 11111222223333344444 800000 400000 5000
Plot4: 11111222223333344444 1200000 400000 5000
Plot5: 11111222223333344444 1600000 400000 5000
...and so on (Each of the above plots is ~100GB)

Cool Story Bro. 

What is your cpu and ram?

What kind of main processor do you have? How many nonces did you get per minute?

OMG, I just had THE worst time with linux because of the prospect of plotting ~2x faster.

So, here's how it went:
Was plotting around 3200 nonces/minute on Windows 7 (4 cores)
Decided to try my hand at Ubuntu...
Downloaded an image, burned it, booted it, installed it to a spare drive
Got Java installed and was feeling very accomplished...

Tried to figured out how to use dcct's plot generator (which failed miserably).
Posts in the forum stated that you need to "make" the plot app...
Much Googling led me to needing additional build tools needing to be installed to be able to "make" the program, which I don't see mentioned anywhere (sudo apt-get install build-essential gcc).
Made the app by doing a CD to the location of dcct's plot generator I downloaded, then issued the "make" command...
...and I get a wall of error text spitting out.
Much more Googling, led me to issue a "uname -a" command only to find out that my build was "i686" (which is apparently 32-bit only) instead of the required x86_64 (which is 64-bit).
*sigh*
all that time wasted...

Downloaded a new Ubuntu image, this time I made sure it was 64bit
Booted the new image and go to install, but something is strange...
The license agreement was missing data, Mouse would randomly stop working and I had ZERO network connectivity
{Completely baffled why the 32bit Ubuntu had no issues, but 64bit is FUBARd}
*Much more Googling ensues*
Tried a bunch of suggested things for networking, tried a wifi dongle... even manual settings. zero connectivity.
*MUCH Much more Googling ensues*
Found a post stating that "IOMMU" (whatever the heck that is) likely needs to be enabled in BIOS.
Found & enabled "IOMMU" in my BIOS
HOLY COW! The 64bit version now has NO ISSUES! Amazing!
{Now feeling super duper accomplished!}

I repeat all the original steps (installing java, build tools, etc)
I go to "make" dcct's plot generator again, and it fails... *sigh*
*more Googling reveals that it's because the secodary drive (my plot drive) is in read only mode
{WTF?! why would I NOT want to write to it?!}
*Googling is getting REALLY old at this point*
Finally found information on how to remount the drive as writable...
"make" the plot generator (and it finally worked this time!)
issued the commands to start plot generation....
and it is now plotting!!!!
{Much excitement!}
.......... SLOWER THAN WINDOWS! (@ 2400 nonces/minute)
*sigh*

TL;DR - Wasted the better part of an entire day trying to get "~2x faster plotting" working by using Linux instead of Windows. Ended up being much more work than it was worth, because the Linux install is plotting SLOWER than Windows.
Windows: ~3200 nonces/minute (4 cores)
Linux: ~2400 nonces/minute (4 cores)

It's your opinion that Linux rules, but that has not been my experience lately

I think I read in the readme file included in the miner, the bigger the plot the less stress when mining starts but I think you can plot 100GB at a time just need to set your nonce after every plot is done.

So first plot 0 400,000
    Second plot: 400,000 800,000
    Third plot:    800,000 1,200,000

and so on. 800,000=200GB.

We need PROP pool with deadline 100k-200k.

do i have to generate the plot in one go...can i generate 10 plots of 100 GB instead of single 1 Tb ..because it takes lot of time to generate a single one and if something happens in between and it closes..all time is wasted..

i'm still stuck on this "ram sucking" issue, i tried to disable superfetch, but still nothing

Well done! Everyone hop on the multi pool!!!

Not exactly. NTFS supports links like unix filesystems.
See mklink.exe command.
Its possible to gather files from 1000 partitions to one directory

It's just my critique, but thanks for contributing to the community...