ohager / burst-autoplotter Goto Github PK

I discovered that the instructionset-checker.exe depends on (legacy) DLLs (MVCP120.dll and MVCR120.dll). So, instruction set detection may fail silently and falls back always to SSE.

Plotting, better writing scoops, is damn slow as it seems that XPlotter is not spawned as admin

I'm not getting prompted to select threads and memory.

great tool, was recommended by a lot of people.

it would be nice to be able to specify a directory of a disk for plotting
Ex : Z:\My Drive\BURST.

Use travis and codecov!

First: Plotter is working, i.e. does not break plots - although shown results may be inconsistent.

On a testplot of 10 GiB in 3 files the last plot showed up with strange values (negative duration times, more than 100% done). I assume that the plot partitions have some issues, as it was expected to have
13648 nonces for each plot. For some reasons, the last plot resulted in larger value than that (maybe 14000) and therefore the plotter was still working and feeding the UI with > 100% values.

Result after plotting:

Validated Plots: Path: C:/Burst/plots/ 16107620026796983538_0_13648_13648 is VALID 16107620026796983538_13653_13648_13648 is VALID 16107620026796983538_27306_13648_13648 is VALID

the second plot should start at 13648 and third at 27296. Could be an indicator for wrong nonce calculations.

CI is up, now just missing CD

Apparently, the calculation of written nonces per plot is wrong.
I started a larger plotting (15 chunks) for 4TiB and the console shows for written nonces insanely large values, like writing 15000 nonces per update (4 secs).

I expect that the plotter works normally - at least the nonces/min speed seem reliable.

I made some tests with SPlotter and found out that its AVX2 version is faster than Blagos provided AVX2 version (Plotting 20000 Nonces - SPlotter: 116 secs, XPlotter: 161 secs),
Furthermore, the codebase is much better organized, and the printfs can be easily directed to stderr (Blago lost codes, and apparently my merge is not considered in v1.1)

Expected advantages:

• faster plotting
• better exception handling
• eliminates rounding problems with AVX versions, as written nonces is provided absolutely (instead of rough percentages)

From PM:

Hello, I am trying to use your burst auto plotter and I can't because my numeric id is only 19 digits long, not 20. Can you provide a fix for this? <<

see title

Would be nice to have an idea how long the plot will still take.

just a TODO

trying to plot, get this error :(

Additionally, to XPlotter, GPU plotter support could be interesting.

Make move plot asynchronous (again).

• Need to consider that while moving, a new plot can be launched already...that is we need consider sufficient diskspace
• if not sufficient, show hint that it can be moved only synchronously, otherwise do async'ly
• Nice to have: Indicator for move progress

i have used autoit for several projects and this could give a very easy and good looking GUI

File is written into current opened directory...must be written into installation directory.

As suggested in #12 , it should be possible to plot not only one disk, but multiple drives, such that the plotter needs to be started only once, and automatically plots several disks (may it sequentially, or even in parallel).

In case of parallel plotting, the amount of CPU/Threads can/should be split/splittable.

Running a 400GiB plot with 4 chunks the progress indicator always stuck around 0.96% (15768/4096000)

Seems that plotting still works, but no update on cli

-- that's the raw output from stdio!

Wouldn't it be cool, to have a small bot on telegram to check the current plotting status - especially if you're not at home?
Some features I can imagine:

• request ETA
• request current Status
• request a screenshot
• receive messages on certain events, like
  • finished single plot
  • finished total plot
  • errors

The current console updates are sloppy... and the ui is not really that neat, but functional.

Blessed is kind of a "reimplementation" of curses, although quite old, first impression are interesting. I'd like to play around with that.

Moving plots may cause exception which are not handled.

Murphy always gets me. every time. my numeric id is 18 digits long.

autoplotter says,
"> Use your numeric account ID (minimum 19 digits), dude!"

But, I'm using my legit numeric id.

Any advice?
Gary

node -v
v4.2.6

autoplot
/usr/local/lib/node_modules/burst-autoplotter/bin/command/command.js:5
const { printUsage, printUnknownCommand, printUnknownOption } = require("./usage");
^

SyntaxError: Unexpected token {
at exports.runInThisContext (vm.js:53:16)
at Module._compile (module.js:374:25)
at Object.Module._extensions..js (module.js:417:10)
at Module.load (module.js:344:32)
at Function.Module._load (module.js:301:12)
at Module.require (module.js:354:17)
at require (internal/module.js:12:17)
at Object. (/usr/local/lib/node_modules/burst-autoplotter/bin/autoplot.js:1:63)
at Module._compile (module.js:410:26)
at Object.Module._extensions..js (module.js:417:10)

Well, introducing logging that late is not a good sign...but it's worth doing it anyway

As plotting is a long-term process of several hours, or even days, would be nice to have some email notification when

• plot file finished
• entire plotting finished
• problems/exceptions

When a path on target drive doesn't exist the plotter doesn't work.

Workaround: Create <DRIVE>:/Burst/plots

As SMR disks are not very performant on plotting, it should be possible to plot a file on a fast disk (PMR, SSD) and then move it to slower disk.

• Need to consider available space on 'buffering' disk for that feature.
• Extend notifications when moving

For CLI scripting, it's necessary to make principal plotting parameters available via command line,

• account
• targetdisk
• plotdisk (move plot)
• plotsize
• chunks
• instructionset
• threads
• memory
• logenable

Maybe it'll be better to use a config file for that...

autoplot --config myconfig.json

this is less cumbersome than introducing dozens of arguments, and let's things more flexible for the future.

Currently, the entire structure is quite close to be spaghetti...no clear separations of responsibilities. Objective of this refactoring is:

• easier to maintain
• better testability
• resolve some bugs, i.e. duration estimates

A planned approach is applying reactive programming (using observable store) and break output into views.

Not sure if this is an error but it looks wrong to me. Each plot is at different percentages, almost never 100% before it starts a new plot. After all plots were created, it said it was finished but left each plot at different percentages complete. What would cause this? Is the text wrong or did it really not write all the nonces?

Need to accept large nonce numbers...currently, the UI complains e.g. 40000000.

The current version uses only AVX instruction set for plotting. Could be interesting to select AVX2, too (under extended settings maybe)

According to Blago, mem settings are an advanced feature - after a look at the XPlotter code I saw that mem settings is set automatically. So, could be shifted towards advanced setup. Same for CPU. Maybe in easy mode CPU/Threads could be set to (available Threads - 1) per default.

With all of this, more detailed output should be written - especially from XPlotters startup (reading initial settings from stdio)