Skip to content

Latest commit

 

History

History
118 lines (65 loc) · 7.19 KB

README.md

File metadata and controls

118 lines (65 loc) · 7.19 KB

OctoRay

OctoRay is a framework that enables users to scale FPGA accelerated applications on multiple FPGAs.

General architecture:

Overview

The idea is based on the principle of data parallelism. It works by splitting the input data payload into as many chunks as the number of available FPGAs and performing the computation in parallel. The results are then combined into a single output object.

Assuming a Dask cluster is already set up, the steps involved to parallelize any task are:

  1. A Dask client reads the input data (from a file, socket, etc.).

  2. The client detects the number of workers in the cluster. It splits the input data and scatters the chunks to the workers.

  3. Each worker uses a Python Driver (a Pynq Overlay/custom driver) to send the data to the FPGA, and wait for the results. The results are then returned to the client.

  4. The client, after receiving all the results, combines them and emits the final output.

Advantages

  1. This architecture can be used in both setups - one host with multiple attached FPGAs, or multiple (remotely connected) hosts each with their own FPGA. Several Dask workers can be spawned on the same or different machines, and each is associated to one FPGA. Any number of FPGAs may be connected this way, and we will observe speedup as long as the network is not saturated.

  2. Different types of kernels/accelerators can be used. For example, kernels built using Vitis libraries, FINN, or other custom kernels can be used as long as a Python interface to them is available. In this project, we have used a Vitis library kernel and a FINN kernel, and written Python drivers for these. Also, any accelerator built for a PYNQ-supported platform can be used.

  3. Since we use popular Python libraries (Dask, Numpy, etc.), this makes the system hardware agnostic. As long as we can compile an accelerator for the available hardware platform, the system can be deployed on that platform. We have run this setup on various hardware platforms such as Pynq-Z1 boards, AWS F1 instances, Nimbix cloud instances, and our in-house Alveo servers.

Installation

Create a virtual environment

python3 -m venv octoray-env

Source the environment

source octoray-env/bin/activate

Install octoray (this takes several minutes)

pip install -v -e . python3 setup.py install

Getting started

OctoRay is build on two fundamental concepts: the cluster configuration and OctoRay kernels. The cluster configuration consists of a dictionary that is used to specify how the Dask distributed cluster should be deployed. OctoRay kernels are structures that consist of all the specifications necessary to deploy an application on the cluster. The example notebooks provide explanations on how to use the cluster configuration and OctoRay kernels to scale applications.

Cluster management

To scale applications on multiple nodes OctoRay uses Dask Distributed to create a cluster. It is possible to deploy the cluster automatically or manually. Manual deployment is recommended during development as it results in a clearer overview where errors occur. Manual deployment is explained is the section below. The example notebooks show how automatica deployment is setup.

Setting up a Dask cluster manually

This consists of two steps:

  1. Starting a dask scheduler. This is a Python process which is responsible for scheduling tasks on the worker and maintaining the application state.

$ dask-scheduler

This emits an IP address, which can be used to register Dask clients and workers.

  1. Starting one or more dask workers. These are Python processes that perform the actual computation. These can be present on the same machine as the scheduler or remote ones. The number of workers per FPGA dependson the number of instances of the application that are present in the bitstream.

$ dask-worker <IP_OF_SCHEDULER> --n_workers <NUMBER_OF_INSTANCES> --nthreads 1 --memory-limit 0 --no-nanny

Examples

1. gzip_compression (GZIP accleration using Vitis Data Compression Library)

From the man page of gzip (https://linux.die.net/man/1/gzip): Multiple compressed files can be concatenated. In this case, gunzip will extract all members at once.

Hence, the concatenation of several gzip files is also a valid gzip file. We use this principle to split up the input uncompressed file into 2 parts, compress these parts separately, and then concatenate them.

The following plot shows the acceleration achieved on using 2 U50 FPGAs to compress files of various sizes. In both cases, the Dask client and the workers were located on separate machines in the Nimbix cloud. The time shown is the time taken to compress the entire input file. It does NOT include the network I/O time taken to transmit data between the client and the worker(s).

Perf comparision

Using 2 FPGAs in parallel resulted in a speedup of ~2x for all input sizes.

Comparision to software baseline:

This setup is faster than a purely software (SW) implementation. The below table shows the throughput benchmarks (in MBps) between FPGA and software-based compression. For the software benchmarks, we used 2 tools:

  1. Linux's inbuilt gzip tool (lowest compression level - fastest)
  2. pigz tool - A multiprocessor implementation of gzip (lowest compression level - fastest) (https://zlib.net/pigz/)

The machine used is a 2.7 GHz Dual-Core Intel Core i5 processor.

gzip SW throughput pigz SW throughput Dask 1 worker Dask 2 workers
24.3 40.6 348.3 627.8

(Units: MBps)

The 2-FPGA implementation is about 15 times faster than a parallalised software implementation.


2. cnv_w1a1_u50 (CNV acceleration using FINN)

In this example, the FINN library was used to build a binarized convolutional network for the U50 FPGA. The CNV-W1A1 network is used to classify a 32x32 RGB image to one of CIFAR-10 dataset's classes. The bitstream was generated with the help of the test cases here - https://github.com/Xilinx/finn/blob/master/tests/end2end/test_end2end_bnn_pynq.py.

Once the accelerator is built, we can use data parallelism to improve the speed of inference. In our setup, we split the test dataset into 8 parts and used 8 accelerators from ETH's XACC cluster to classify these parts in parallel. As a result, a speedup of 8x was observed in the classification time (inference time + data copying time to/from the FPGA) when compared to using just 1 FPGA.

Perf comparision

Comparision to software baseline:

This setup is also faster than a purely software (SW) implementation. A pretained version of the same network (CNV-W1A1) from Xilinx's Brevitas (https://github.com/Xilinx/brevitas/tree/master/brevitas_examples/bnn_pynq) framework was used as a software baseline.

The machine used is a 2.7 GHz Dual-Core Intel Core i5 processor (no GPU).

The table below shows the throughput values for the networks:

brevitas SW model Dask 1 worker Dask 2 workers
232.56 2944.96 5709.42

(Units: images/second)

The 2-FPGA implementation is almost 24 times faster than a software implementation of the same neural network running on a CPU.