A Reconfigurable Fabric for Accelerating Large-Scale Datacenter Services
| Autor: | Phillip Yi Xiao, Gopi Prashanth Gopal, James R. Larus, Sitaram Lanka, Hadi Esmaeilzadeh, Adrian M. Caulfield, Simon Pope, Derek Chiou, Michael Haselman, Doug Burger, Amir Hormati, Aaron L. Smith, Kypros Constantinides, Scott Hauck, Jeremy Fowers, Andrew Putnam, John Demme, Eric C. Peterson, Eric S. Chung, Stephen F. Heil, Jason Thong, Joo-Young Kim, Jan Gray |
|---|---|
| Předmět: |
accelerator
General Computer Science Computer science 02 engineering and technology data center computer.software_genre Software Factor (programming language) 020204 information systems Server 0202 electrical engineering electronic engineering information engineering Electrical and Electronic Engineering Resilience (network) Field-programmable gate array Throughput (business) FPGA computer.programming_language PCI Express Flexibility (engineering) business.industry General Medicine Reconfigurable computing 020202 computer hardware & architecture Computer architecture Hardware and Architecture Software deployment Embedded system reconfigurable computing Web search engine search enginee Data center Web service business computer |
| Zdroj: | ISCA |
| Popis: | Datacenter workloads demand high computational capabilities, flexibility, power efficiency, and low cost. It is challenging to improve all of these factors simultaneously. To advance datacenter capabilities beyond what commodity server designs can provide, we have designed and built a composable, reconfigurablefabric to accelerate portions of large-scale software services. Each instantiation of the fabric consists of a 6x8 2-D torus of high-end Stratix V FPGAs embedded into a half-rack of 48 machines. One FPGA is placed into each server, accessible through PCIe, and wired directly to other FPGAs with pairs of 10 Gb SAS cables In this paper, we describe a medium-scale deployment of this fabric on a bed of 1,632 servers, and measure its efficacy in accelerating the Bing web search engine. We describe the requirements and architecture of the system, detail the critical engineering challenges and solutions needed to make the system robust in the presence of failures, and measure the performance, power, and resilience of the system when ranking candidate documents. Under high load, the largescale reconfigurable fabric improves the ranking throughput of each server by a factor of 95% for a fixed latency distribution--- or, while maintaining equivalent throughput, reduces the tail latency by 29% |
| Databáze: | OpenAIRE |
| Externí odkaz: |
|