Plotting Software

Video chia ram

Several Chia plotters are now available. The output (a plot) will be nearly identical for a given k-value and compression level. However, the hardware requirements are different for each plotter.

The three families of plotters include:

  • BladeBit – developed by Chia Network, Inc.
  • madMAx – developed external to CNI
  • ChiaPoS – the original plotter, developed by CNI

This page provides details about the plotters that exist within each of these families.

Plotters​

BladeBit CUDA​

A GPU plotter, included with Chia 2.0​

Plot capabilities

  • Type: Compressed or uncompressed
  • Size: k32 only (larger sizes to be added later)

Requirements

  • OS: Windows 11 or Debian/Ubuntu Linux (MacOS and other flavors of Linux will likely be supported in the future)
  • Memory:
    • For 100% in-memory plotting: 256 GB of DRAM
    • For RAM + disk plotting: 16 GB (experimental) or 128 GB (supported as of Chia 2.1.0)
  • Temporary Disk:
    • Not used for 256 GB version
    • SSD required for <= 128 GB of RAM
  • GPU: CUDA capability 5.2 (NVIDIA 10 series GPU or higher) with 8GB of GPU VRAM
  • Software: CUDA toolkit version 11.8 or later

More info

  • The newest BladeBit plotter, designed to work with CUDA-class GPUs
  • Creates compressed plots, up to C9 (75.2 GB)
  • The fastest Chia plotter for most hardware architectures that meet the minimum specs
  • The 256 GB version creates plots entirely in memory, so no temp disk is needed
  • Open-source, freely available, no dev fee
  • Can also be installed as a standalone build
  • A detailed analysis of BladeBit CUDA’s performance was undertaken by scienceofmining.com

CLI documentation

Example command

The following command will create a single plot with a compression level of 7. It will use the specified keys and contract address in case the farmer is located on a different machine. It will use the defaults for the remaining parameters:

BladeBit RAM​

An all-memory CPU plotter, included with Chia 2.0​

Plot capabilities

  • Type: Compressed or uncompressed
  • Size: k32 only (larger sizes to be added later)
  • More info

Requirements

  • OS: Windows, Mac, or Linux OS (64-bit is required); Intel and ARM (Apple Silicon) are both supported.
  • Memory: 416 GB of available RAM
  • Temporary Disk: Not used
  • GPU: Not used

More info

  • Designed to be used in high-memory servers that don’t have a GPU
  • Creates compressed plots, up to C9 (75.2 GiB)
  • Creates plots entirely in memory, so no temp disk is needed
  • Typically not as fast as BladeBit CUDA
  • Open-source, freely available, no dev fee
  • Can also be installed as a standalone build

CLI documentation

Example command

The following command will create a single plot with a compression level of 7. It will use the specified keys and contract address in case the farmer is located on a different machine. It will use the defaults for the remaining parameters:

Bladebit Disk​

A disk-based (HDD or SSD) CPU plotter, included with Chia 2.0​

Plot capabilities

  • Type: Uncompressed only in Chia 2.0, Compressed beginning in 2.1
  • Size: k32 only

Requirements

  • OS: Windows, Mac, or Linux OS (64-bit is required); Intel and ARM (Apple Silicon) are both supported.
  • Memory: Minimum 2 GB of available RAM, with lower bucket counts requiring up to 12 GB
  • Temporary Disk: 480 GB in default mode, or 390 GB with -alternate mode enabled; can be HDD or SSD:
    • SSD: fast (NVMe is supported) but consumer-grade SSDs wear out over time, enterprise-grade is recommended
    • HDD: slow, but won’t wear out; can plot directly to the final disk
  • GPU: Not used

More info

  • Designed to be used in embedded or entry-level systems
  • Can only create uncompressed plots (C0, 101.4 GiB) in Chia version 2.0
  • The use of temporary HDD or SSD storage makes it accessible to the majority of farmers
  • Sequential writes can better take advantage of SSD burst performance and reduce SSD wear by reducing write amplification factor
  • Can use DRAM write cache to significantly reduce SSD writes and can take advantage of any extra increments (no minimum required)
  • Takes full advantage of increased disk bandwidth from PCIe 4.0
  • Pipelined performance to max out CPU
  • Can also be installed as a standalone build

CLI documentation

Example command

The following command will create a single uncompressed plot (plot compression is disabled in Chia version 2.0.0). It will use the specified keys and contract address in case the farmer is located on a different machine. It will allocate 32 GB of DRAM cache, and it will use the specified temporary drive (typically an SSD) and destination drive. It will use the defaults for the remaining parameters:

Gigahorse​

A third-party GPU plotter, available as a standalone download​

Plot capabilities

  • Type: Compressed or uncompressed
  • Size: k30-34

More info

  • Designed to work with CUDA-class GPUs
  • Creates compressed plots
  • Closed source, has dev fee depending on compression and whether CPU or GPU is used for farming

madMAx​

A third-party CPU plotter, included with Chia 2.0​

Plot capabilities

  • Type: Uncompressed only
  • Size: Any

Requirements

  • OS: Windows, Mac, or Linux OS (64-bit is required); Intel and ARM (Apple Silicon) are both supported.
  • Memory: Depends on setup, minimum 0.5 GB if single-threaded
  • Temporary Disk: SSD and HDD are supported
    • SSD: fast (NVMe is supported) but consumer-grade SSDs wear out over time, enterprise-grade is recommended
    • HDD: slow, but won’t wear out; can plot directly to the final disk
  • GPU: Not used

More info

  • Designed to be used in embedded or entry-level systems
  • Can only create uncompressed plots (C0, 101.4 GiB)
  • The use of temporary HDD or SSD storage makes it accessible to the majority of farmers
  • Pipelined performance to max out CPU

CLI documentation

Example command

ChiaPoS​

The original Chia CPU-based plotter​

Plot capabilities

  • Type: Uncompressed only
  • Size: Any

Requirements

  • OS: Windows, Mac, or Linux OS (64-bit is required); Intel and ARM (Apple Silicon) are both supported.
  • Memory: 2 GB min
  • Temporary Disk: SSD and HDD are supported
    • SSD: fast (NVMe is supported) but consumer-grade SSDs wear out over time, enterprise-grade is recommended
    • HDD: slow, but won’t wear out; can plot directly to the final disk
  • GPU: Not used

More info

  • Originally released pre mainnet, mostly supplanted by newer plotters
  • Slow, single-threaded plotting; parallel plotting and staggering are recommended
  • Designed to be used in embedded or entry-level systems
  • Can only create uncompressed plots (C0, 101.4 GiB)
  • The use of temporary HDD or SSD storage makes it accessible to the majority of farmers
  • Not optimized for performance

CLI documentation

Example command

BladeBit (standalone)​

While BladeBit does come installed with Chia version 2.0, there are a few reasons you might want to run the standalone version instead:

  • You want to run the simulate command, to determine your farm’s maximum size
  • You want to test features that have not yet been released in the embedded version

Complete instructions for downloading and installing BladeBit are available on the BladeBit GitHub page.

Abbreviated installation instructions:

  1. Install cmake
  2. Clone BladeBit:
  1. Create and enter a build directory:
  1. Generate config files:
  1. Build BladeBit:

After the installation has completed, the bladebit command will be available from the Release directory. For example, to obtain a list of options, run:

Choosing a plotter​

With so many plotters available, the decision of which one to choose may seem daunting. However, your hardware setup will often make the choice for you. If you have:

  • At least 256 GB of RAM and a CUDA-class GPU with at least 8 GB of VRAM
    • You will likely want BladeBit CUDA
    • Gigahorse will also work
    • Other plotters will under-perform
  • At least 16 GB of RAM (experimental) or at least 128 GB, but less than 256 GB and a CUDA-class GPU with at least 8 GB of VRAM, and a 256 GB or larger SSD (ideally enterprise NVMe)
    • You will likely want BladeBit CUDA
    • Gigahorse will also work
    • Other plotters will under-perform
  • 416 GB of RAM but no GPU
    • BladeBit RAM is easily your best choice
  • Less than 16 GB of RAM and don’t mind creating uncompressed plots
    • BladeBit Disk and madMAx are both good options

It is always possible, and indeed recommended, to create a plot with a few different plotters to understand how well your system will perform. Once you have a feel for using the different plotters, you can begin plotting in earnest.

CLI usage​

There is a new chia command for creating plots called plotters. For compatibility, the original command for creating plots chia plots create remains in place, however, this will always use the reference ChiaPoS plotter. In order to use the other plotters, you must use the new chia plotters command. Command line options differ with each plotter, so be sure to check the available options using chia plotters <plotter> -h or by reading the online CLI documentation. Available plotter values include “chiapos”, “bladebit”, and “madmax”.

The UI also has new functionality to support selecting a plotter.

Note: Not all features are available for every plotter.

This post was last modified on Tháng hai 7, 2024 3:56 sáng

Trần Thu Uyên: Họ tên đầy đủ: Xin chào mình là thuuyen (tên thật là Trần Thu Uyên) Trình độ chuyên môn: Tốt nghiệp khoa Báo chí trường Học Viện Báo Chí Và Tuyên Truyền. Hai năm kinh nghiệm sản xuất những nội dung như ẩm thực, giáo dục, phong thủy, sự kiện, thần số học, chiêm tinh... Kinh nghiệm làm việc: Từ nhỏ mình đã có niềm đam mê với viết lách, đặc biệt là chia sẻ góc nhìn của mình về 1 vấn đề cụ thể. Mình luôn luôn nói đến và tìm kiếm liên hệ giữa mình và vũ trụ xung quanh. Mình là một biên tập viên của Vui Học Tiếng Hàn. Mình muốn dùng kiến thức và kinh nghiệm của bản thân qua một thời gian dài làm việc để truyền tải những giá trị tốt đẹp và tích cực tới mọi người. Các bạn có thể kết nối thêm với mình tại: https://twitter.com/thuuyen268 https://www.linkedin.com/in/thuuyen268/ https://www.twitch.tv/thuuyen268/ https://www.flickr.com/people/thuuyen268/ https://dribbble.com/thuuyen268/about https://www.behance.net/thuuyen268 https://github.com/thuuyen268 https://www.scoop.it/u/thuuyen268 https://vi.gravatar.com/thuuyen268 https://www.reddit.com/user/thuuyen268 https://issuu.com/thuuyen268 https://www.goodreads.com/thuuyen268 https://myspace.com/thuuyen268/ https://trello.com/u/thuuyen268 https://www.plurk.com/thuuyen268 https://linkhay.com/u/thuuyen268 https://linktr.ee/thuuyen268 https://www.pinterest.com/thuuyen286/ www.tumblr.com/thuuyen268 https://vimeo.com/thuuyen268 https://soundcloud.com/thuuyen268 https://patreon.com/thuuyen268