-----------
DESCRIPTION
-----------
GPU plot generator for Burst coin.
Author: Cryo
Bitcoin: 138gMBhCrNkbaiTCmUhP9HLU9xwn5QKZgD
Burst: BURST-YA29-QCEW-QXC3-BKXDL

Based on the code of the official miner and dcct's plotgen.

If you like this software, support me ;)

-----------------------
HOW TO BUILD ON WINDOWS
-----------------------
Install msys/mingw.
Install OpenCL (available in the manufacturer SDKs) or find the include/lib files on your hard drive (some drivers include those files).

Modify the [OPENCL_INCLUDE] and [OPENCL_LIB] variables of the Makefile to the correct path.

cd <this directory>
make dist

The [bin] folder contains all the necessary files to launch the GPU plotter.

---------------------
HOW TO BUILD ON LINUX
---------------------
Install the build-essential and g++ packets.
Install OpenCL.

Modify the [OPENCL_INCLUDE] and [OPENCL_LIB] variables of the Makefile to the correct path.

cd <this directory>
make dist

The [bin] folder contains all the necessary files to launch the GPU plotter.

----------
HOW TO USE
----------
Display a full help message: ./gpuPlotGenerator
Display a per-command help message: ./gpuPlotGenerator help <command>
----
List the OpenCL platforms: ./gpuPlotGenerator list platforms
Choose your platform
List the devices bound to this platform : ./gpuPlotGenerator list devices <platformId>
Choose your device
Generate the plots with the selected platform/device pair : ./gpuPlotGenerator generate <platformId> <deviceId> ...

------------------
PERFORMANCE ISSUES
------------------
There is two available modes:
  - One graphic card
  - Multiple graphic cards (preferred way)

--
One graphic card
--
Facts:
  - The graphic card is bound to your display drawing loop.
  - When launched, an OpenCL kernel uses all the GPU resources.
  - To avoid screen freezes, the display driver kills every hanging kernel after a defined period (around 5 seconds most of the time).

So, the plot generation has to be split in smaller chunks to avoid being killed.
The chunks size are expressed in number of hashes (<hashesNumber> parameter). It ranges from 1 to 8160 (PLOT_SIZE / HASH_SIZE - HASH_SIZE).
Try with a low number and go up as far as you can.

Warning: As the plot generation runs at the same time as your display renders, you may experience some freezes or repetitive lack of responsiveness.

--
Multiple graphic cards (preferred way)
--
In this mode, all the non-bound graphic cards can be used to generate plots without any timeout issue.
The typical <hashesNumber> parameter will be 8160.

To generate multiple plots at the same time, just launch multiple processes on different cards.
Each process will still requires a certain amount of RAM on the CPU side.