blagodarenko / xplotter Goto Github PK

within the avx2_mshabal there are a lot of "+ 0" operations
this might be nice to read but the compiler often does not remove this operation and executes the +0 physically

eg.
https://stackoverflow.com/questions/11492081/does-c-compiler-optimize-0x

might not have a too big impact but it might be helpful

Pull Request submitted to fix issue

colored message is printing garbled at times.

1>XPlotter.cpp(18): warning C4477: 'printf' : format string '%s' requires an argument of type 'char *', but variadic argument 1 has type 'std::string'
1>XPlotter.cpp(24): warning C4477: 'printf' : format string '%s' requires an argument of type 'char *', but variadic argument 1 has type 'std::string'

Hey there!

First of all I want to thank you for developing and supporting this project.

v1.0 in release notes says:
You can turn off Xplotter at any time, and later may continue plotting (works only at NTFS)

But for me it doesn't work. It says that file exists and I should remove it first.

Am I doing anything wrong?

Thanks.

I have 8 x 4TB disks in 2 NASs, accessed via smb. When I run Xplotter without -n it correctly works out the nonces to plot the entire space, but then fails with "Not enough free space, reduce "nonces"... (code = 59)".

The disks report as 3.56TB free and I was able to plot 13.9 million nonces, leaving about 2-300 GB free. However, as time passes and I plot more drives the space Xplotter will let me plot reduces. The next one was 13.8 mio, now I can't even plot 12 mio any more.

Can we debug this problem, please?

I would just like to bring this to your attention. It appears they have tried to present your work as their own under the MIT licence. This is in direct violation of the GPL.

I would like to find out what commands do i need to use to enable POC2 plotting when using xplotter with out the burstcoin qbundle

Hey!

Could you add a license to your project?

Thanks.

Hello,

I've seen that no ne release has been posted since early 201 but in the master branch some commit has been made only a fe days ago.
Because i'm a little bit curious i've take a look at the Wplotter.cpp file and have found that the avx2 is mentionned and apparently working.

Any guide how to build the exe for avx2 ?

Best regards,

I would love to try this on Linux. Any chance we can get Linux source?

Thank You

Can we get a switch to move the file after it's plotted (for those of us that use a scratch drive before copying to slower write SMR drives)

for instance xplotter.exe -path X:\scratch\plots -move Q:\Plots

File gets moved to Q:\Plots after the plot is completed, then program terminates

I have mapped a network share and given all read write permissions to everyone but am still getting this error plotting: SetFileValidData error (code = 1314)

It creates the plot file instantly but seems like it is just filled with zeros and won't mine. Both machines are windows server 2016. Is this by design or am I supposed to be able to plot to shared drives?

Thanks!

Hi,

The plot filename shoud be : {Account ID}_{Start nonce}_{Nonces count}_{Stagger size}

But with XPlotter, it is : {Account ID}_{Start nonce}_{Nonces count}_{Nonces count}

I think there is a problem.

Have a good day !

Hi Blago,

I just discovered that the recently released version V1.1 is different from current codebase.

Example
The console output in provided executables is /CPU: (\d+)% done, \((\d+) nonces\/min\)/g,

while in the latest code the output is:

printf("\rCPU: %llu nonces done ", nonces_done + x);

The concrete difference is:
executable: CPU: 54% done
code: CPU: 13623 nonces done

Please, could you bring them in sync?

Btw: the AVX2 brought significant speed up

Hello, I've found a lot of different soft for this coin, but did not found any up to date for linux. Please help me, I need wallet, plotter, miner for linux, and normal docs how to use it. As I understand this plotter only for windows ? Thanks a lot.

Currently running in trouble because the length of the Paths variable is to small. Having more then 110 Drives connected to one node.

Sample:

"Paths":[
"C:\Mount\Burst_001\plots",
...
"C:\Mount\Burst_110\plots"
]

throws me an error. When I remove some mounting points its works fine.

While I'm plotting, one of the biggest bottlenecks I'm seeing is writing to my drives. I currently have two drives plotting at the same time right now, with equal resources 3 threads and 4GB RAM. While watching them plot, I notice that one drive will write very slowly for some reason, and then the plotter stops generating nonces while it waits for the scoops to finish writing. I feel like, while it's waiting for that drive to finish writing, that would be a good opportunity to crank up the threads for the other drive that is being plotted.

So basically, any time one plotting instance is stuck waiting for the scoops to finish writing, it will allow the other plotting instance to use those free threads.

Just bought a 4TB WD Mybook External HDD. It's factory formatted the exFAT file system.
I tried to run Xplotter from the AIO wallet on it and got an error msg like this :

"error creating stream for the file "

Did anyone else face this? Is it a problem unique to the Mybook?

Hi Blago,

another user encountered an issue with my autoplotter. For some reason XPlotter exited unsuccessfully, and returned a code only. It would be cewl if you could write your exceptions to stderr, such that it'll be easier to get the entire error messages.

The :stream ADS used to save plot progress isn't well documented, causing issues when using non-NTFS storage systems.

• In some limited testing, creating the stream appears to silently fail on any non-NTFS partition, so trying to resume plotting just truncates the file and starts over without any confirmation.
• Windows Explorer will show a vague warning message when trying to move finished plotfiles to non-NTFS destinations: Are you sure you want to copy this file without its properties? The file <filename> has properties that can't be copied to the new location., followed by some general information about the file. New miners won't know they can safely ignore the message for finished plotfiles, but not for unfinished ones (I think).

These should at least be documented in the README. I think it would also be possible to adopt the resume system cg_obup uses, which indicates incomplete plotfiles with a magic number at the end of the file. If I understand the plotfile format correctly, this would allow either program to resume the other's plotfiles, and completely eliminate the handling issues with ADS streams.

Another option would be to delete the stream after plotting is complete, and don't try to resume on filesystems that don't support ADS. This would solve both of the above issues.

I'm not very skilled in C++, but I could put together a simple PR if you don't have the time.

When I try to plot close to the total capacity of a network drive the plotter creates too large a file:

C:\tmp\XPlotter>XPlotter_sse.exe -id 18084828890988000920 -sn 400000000 -n 14800000
14800000-t 4 -mem 6G -path z:\

XPlotter v1.0 for BURST
programmers: Blago, Cerr Janror, DCCT

Checking directory...
Drive z:\ info:
Name: plots4
File system: NTFS
Serial Number: 3178277779
FILE_SUPPORTS_SPARSE_FILES: no
Bytes per Sector: 512
Sectors per Cluster: 2
The token does not have the specified privilege.
For faster writing you should restart plotter with Administrative rights.
Creating file: z:\18084828890988000920_400000000_14831352_14831352
Not enough free space, reduce "nonces"... (code = 59)

Note that -n is 14,800,000, but the number in the plot file is larger.

According to xplotter 1.1, my write speed is 65 MB/s. Yet writing 6 GB of previously plotted RAM to my HDD only takes ~51 seconds, which means, it's rather ~120 MB/s. Maybe some factor of two error in the calculations?
I also checked if the plotted RAM is incorrect, so I also stopped the time there: 26 seconds @ 28kN/min with 512 kByte/nonce equals 5.9 GByte. So all looks ok on this end.