BOINC: A Platform for Volunteer Computing

Abstract

Volunteer computing" is the use of consumer digital devices for high-throughput scientific computing. It can provide large computing capacity at low cost, but presents challenges due to device heterogeneity, unreliability, and churn. BOINC, a widely-used open-source middleware system for volunteer computing, addresses these challenges. We describe its features, architecture, and implementation.Volunteer computing (VC) is the use of consumer digital devices, such as desktop and laptop computers, tablets, and smartphones, for high-throughput scientific computing. Device owners participate in VC by installing a program that downloads and executes jobs from servers operated by science projects. There are currently about 30 VC projects in many scientific areas and at many institutions. The research enabled by VC has resulted in numerous papers in Nature, Science, PNAS, Physical Review, Proteins, PloS Biology, Bioinformatics, J. of Mol. Biol., J. Chem. Phys, and other top journals [1].About 700,000 devices are actively participating in VC projects. These devices have about 4 million CPU cores and 560,000 GPUs, and collectively provide an average throughput of 93 PetaFLOPS. The devices are primarily modern, high-end computers: they average 16.5 CPU GigaFLOPS and 11.4 GB of RAM, and most have a GPU capable of general-purpose computing using OpenCL or CUDA.In terms of power consumption, data-center nodes are more efficient (i.e. have greater FLOPS/watt) than many consumer devices. However, to compare net energy usage, two other factors must be considered. First, data centers use air conditioning to remove the heat generated by hardware; consumer devices generally do not. Second, in cool climates, the heat generated by consumer devices contributes to ambient heating, so the net energy cost of computing may be zero. Thus it's not clear whether VC is globally more or less energy efficient than data-center computing. In any case, this does not affect VC's cost to the scientist, since the energy is paid for by volunteers.VC is best suited to high-throughput computing (HTC): workloads consisting of large groups or streams of jobs where the goal is high rate of job completion, rather than low job turnaround time. VC is less suited to workloads that have extreme memory or storage requirements, or for which the ratio of network communication (i.e. input and output file size) to computing is extremely high.VC differs from other forms of HTC, such as grids and clouds, in several ways: The computers are anonymous and untrusted.  The computers are heterogeneous in many hardware and software dimensions; for example, of desktops and laptops participating in VC, 85% are Windows, 7% are Macs, and 7% are Linux. Job turnaround times are uncertain because of sporadic host availability.  There is significant "device churn", i.e. arrival and disappearance of computers. Creating the resource pool requires volunteer recruitment and retention, and the client software must be easy for consumers to install and run. The scale of VC is larger: ...

JSDoop and TensorFlow.js: Volunteer Distributed Web Browser-Based Neural Network Training

“…In spite of these difficulties, VC has grown to achieve unprecedented processing capacity. One such example is BOINC, that has 27 PetaFLOPS average computing power available, more than 300 thousand active participants and nearly 850 thousand active computers [5]. As a remark, Summit or OLCF-4 (among the fastest supercomputers in the world) achieve 143.5 PetaFLOPS in the LINPACK benchmark (https://www.top500.org/lists/2018/11/).…”

“…First, we need to get access to those billion devices held by users all around the world, and second (but not least), we need to find a way of managing the globally distributed resources transparently. In this sense, several attempts have been made, and amid them, volunteer computing (VC) arises as a prominent approach [5]- [7].…”

“…If the data or target application has a clear public vocation, many users will share their resources for humanitarian purposes, such as studying cancer, water quality in a city, or helping to reduce energy consumption. All these reasons have motivated several scientific projects that accomplish their mission thanks to volunteers, e.g., IBM World Community Grid [8], Rosetta@home [9], SETI@home [10], among others [5].…”

Task Assignment Algorithm Based on Trust in Volunteer Computing Platforms

“…In the past decades, volunteer computing (VC) has provided huge computing power for large-scale scientific research projects by using idle resources over the Internet. A well-known open volunteer computing platform (VCP) is BOINC [1] (Berkeley Open Infrastructure for Network Computing), and some scientific research projects are running in BOINC, such as SETI@home [2] and Einstein@ Home. In addition, there are also some VC projects running in other VC platforms, such as Folding@home [3] and ATLAS@Home [4].…”

“…The task t u continues to be divided up into two task slices, whose sizes are T (u).cost-v.compute and t u .cost-(T (u).cost-v.compute) Step two: the IEFTT algorithm calculates the total computing time of the first time period V 1 .con = 10, and the total computing time of the tasks with task priority 1 is T (1) .cost = 6.…”