Skip to content

Latest commit

 

History

History
207 lines (152 loc) · 7.27 KB

README.md

File metadata and controls

207 lines (152 loc) · 7.27 KB
Raw

Add 3090 Support

  • 500s+ when after all optimization, should be 400s+
  • the branch 3090 works now.
  • need team work to finish the code to save more time.

to do list:

  1. prover.rs: 425 change to run multi fft on one 3090 card.
  2. provers.rs: 500, density_filter change to parallel.
  3. provers.rs:305, synthesize function change to multi cpu parallel.
  4. provers.rs:425 change to run multi gpu( later I will upload a multi gpu branch, change a little can work.)
  5. gpu params opimization: src/gpu/multiexp.rs 52:calc_num_groups 76:calc_best_chunk_size 87:calc_chunk_size
  6. change openCL to CUDA.

C2 in 880s tested.

  • this is special for 2080Ti, may not work with any other GPU cards. because different gpu have different cuda-cores and mem.
  • some calculations with cpus change to parallel.
  • change params special for 2080Ti. because it has 68 SMs and 64 SP per SM.
  • we are going on with more optimizations. May 100s even more can be saved.
  • Team work requests are welcome!

From ZQBC

HOWTO

  • get the code:
cd ../lotus_code_path && git clone https://github.com/jackoelv/bellperson.git && git checkout origin/2080Ti
  • patch the filecoin-ffi submodule
cd ./lotus_code_path && git submodule update --init --recursive
cd ./lotus_code_path/extern/filecoin-ffi/rust/
vi Cargo.toml

in the end of the file add patch code:

[patch.crates-io]
bellperson = { path = "../../../../bellperson" }
  • then update cargo package
cd ./lotus_code_path/extern/filecoin-ffi/rust/
cargo update
cd ./lotus_code_path
RUSTFLAGS="-C target-cpu=native -g" FFI_BUILD_FROM_SOURCE=1 make clean all

enjoy it!

DONATION

Jennifer suggested that I should have a donation wallet.

fils are welcome if you like.

my wallet addr:

f1ki5mgbm4cyz43oamnbvv5bjrqdsvkphuxxs2h4a

FAQ

openCL error

if you see error like this:

Status error code: CL_MEM_OBJECT_ALLOCATION_FAILURE (-4)

because the MEM needs exceeds maxinum gpu mem. I am looking for new solution to caculate the suitable mem needs. a temporary solution is change the variable ：

  • src/gpu/multiexp.rs
324: jack_chunk = (jack_chunk as f64 / 10f64).ceil() as usize;

you can increase 10f64 to 11f64 even bigger to reduce mem needs. I am trying a new solution which change window_size for lower cost of mem. In prover.rs, b_g2_aux caculation size of <::Projective> is doubled, so with same window_size and num_groups, mem needs increase fast.

attention: this code removed from 3090 branch.

bellperson Crates.io

This is a fork of the great bellman library.

bellman is a crate for building zk-SNARK circuits. It provides circuit traits and primitive structures, as well as basic gadget implementations such as booleans and number abstractions.

Backend

There are currently two backends available for the implementation of Bls12 381:

  • paired - pure Rust implementation
  • blstrs - optimized with hand tuned assembly, using blst

They can be selected at compile time with the mutually exclusive features pairing and blst. Specifying one of them is enough for a working library, no additional features need to be set. The default for now is pairing, as the secure and audited choice.

GPU

This fork contains GPU parallel acceleration to the FFT and Multiexponentation algorithms in the groth16 prover codebase under the compilation feature gpu, it can be used in combination with pairing or blst.

Requirements

  • NVIDIA or AMD GPU Graphics Driver
  • OpenCL

( For AMD devices we recommend ROCm )

Environment variables

The gpu extension contains some env vars that may be set externally to this library.

  • BELLMAN_NO_GPU

    Will disable the GPU feature from the library and force usage of the CPU.

    // Example
env::set_var("BELLMAN_NO_GPU", "1");

  • BELLMAN_VERIFIER

    Chooses the device in which the batched verifier is going to run. Can be cpu, gpu or auto.

    Example
env::set_var("BELLMAN_VERIFIER", "gpu");

  • BELLMAN_CUSTOM_GPU

    Will allow for adding a GPU not in the tested list. This requires researching the name of the GPU device and the number of cores in the format ["name:cores"].

    // Example
env::set_var("BELLMAN_CUSTOM_GPU", "GeForce RTX 2080 Ti:4352, GeForce GTX 1060:1280");

  • BELLMAN_CPU_UTILIZATION

    Can be set in the interval [0,1] to designate a proportion of the multiexponenation calculation to be moved to cpu in parallel to the GPU to keep all hardware occupied.

    // Example
env::set_var("BELLMAN_CPU_UTILIZATION", "0.5");

Supported / Tested Cards

Depending on the size of the proof being passed to the gpu for work, certain cards will not be able to allocate enough memory to either the FFT or Multiexp kernel. Below are a list of devices that work for small sets. In the future we will add the cuttoff point at which a given card will not be able to allocate enough memory to utilize the GPU.

Device Name Cores Comments
Quadro RTX 6000 4608
TITAN RTX 4608
Tesla V100 5120
Tesla P100 3584
Tesla T4 2560
Quadro M5000 2048
GeForce RTX 2080 Ti 4352
GeForce RTX 2080 SUPER 3072
GeForce RTX 2080 2944
GeForce RTX 2070 SUPER 2560
GeForce GTX 1080 Ti 3584
GeForce GTX 1080 2560
GeForce GTX 2060 1920
GeForce GTX 1660 Ti 1536
GeForce GTX 1060 1280
GeForce GTX 1650 SUPER 1280
GeForce GTX 1650 896
gfx1010 2560 AMD RX 5700 XT

Running Tests

To run the multiexp_consistency test you can use:

RUST_LOG=info cargo test --features gpu -- --exact multiexp::gpu_multiexp_consistency --nocapture

Considerations

Bellperson uses rust-gpu-tools as its OpenCL backend, therefore you may see a directory named ~/.rust-gpu-tools in your home folder, which contains the compiled binaries of OpenCL kernels used in this repository.

License

Licensed under either of

at your option.

Contribution

Unless you explicitly state otherwise, any contribution intentionally submitted for inclusion in the work by you, as defined in the Apache-2.0 license, shall be dual licensed as above, without any additional terms or conditions.