Can Traditional Programming Bridge the Ninja Performance Gap for Parallel Computing Applications?
Current processor trends of integrating more cores with wider Single-instruction multiple-data (SIMD) units, along with a deeper and complex memory hierarchy, have made it increasingly more challenging to extract performance from applications. It is believed by some that traditional approaches to pr...
| Published in: | Communications of the ACM Vol. 58; no. 5; pp. 77 - 87 |
|---|---|
| Main Authors: | , , , , , , , |
| Format: | Article |
| Published: |
Association for Computing Machinery
May2015
|
| Subjects: | |
| Online Access: | View this record in EBSCOhost |
| fields | @attributes: recordID: 1 pdfLink: plink: https://search.ebscohost.com/login.aspx?direct=true&db=hlh&AN=102403631&site=ehost-live header: @attributes: shortDbName: hlh uiTerm: 102403631 longDbName: Humanities International Complete uiTag: AN controlInfo: bkinfo: jinfo: jid: 00010782 ACM jtl: Communications of the ACM issn: 00010782 maglogo: N pubinfo: dt: May2015 vid: 58 iid: 5 pid: 68 pub: Association for Computing Machinery artinfo: ui: 102403631 10.1145/2742910 ppf: 77 ppct: 10 formats: tig: atl: Can Traditional Programming Bridge the Ninja Performance Gap for Parallel Computing Applications? aug: au: Satish, Nadathur Kim, Changkyu Chhugani, Jatin Saito, Hideki Krishnaiyer, Rakesh Smelyanskiy, Mikhail Girkar, Milind Dubey, Pradeep affil: Parallel Computing Lab, Intel Corp. Google Inc. Ebay Inc. Intel Compiler Lab, Intel Corp. su: Performance Parallel programs (Computer programs) Parallel programming C++ Multicore processors SIMD (Computer architecture) sug: subj: Performance Parallel programs (Computer programs) Parallel programming C++ Multicore processors SIMD (Computer architecture) ab: Current processor trends of integrating more cores with wider Single-instruction multiple-data (SIMD) units, along with a deeper and complex memory hierarchy, have made it increasingly more challenging to extract performance from applications. It is believed by some that traditional approaches to programming do not apply to these modern processors and hence radical new languages must be designed. In this paper, we question this thinking and offer evidence in support of traditional programming methods and the performance-versus-programming effort effectiveness of multi-core processors and upcoming many-core architectures in delivering significant speedup, and close-to-optimal performance for commonly used parallel computing workloads. We first quantify the extent of the “Ninja gap,” which is the performance gap between naively written C/C++ code that is parallelism unaware (often serial) and best-optimized code on modern multi-/many-core processors. Using a set of representative throughput computing benchmarks, we show that there is an average Ninja gap of 24X (up to 53X) for a 6-core Intel® Core™ i7 X980 Westmere CPU, and that this gap if left unaddressed will inevitably increase. We show how a set of well-known algorithmic changes coupled with advancements in modern compiler technology can bring down the Ninja gap to an average of just 1.3X. These changes typically require low programming effort, as compared to the very high effort in producing Ninja code. We show equally encouraging results for the upcoming Intel® Xeon Phi™ architecture which has more cores and wider SIMD. We thus demonstrate that we can contain the otherwise uncontrolled growth of the Ninja gap and offer a more stable and predictable performance growth over future architectures, offering strong evidence that radical language changes are not required. pubtype: Periodical doctype: Article src: R language: English refInfo: copyright: @attributes: flag: Y dt: @attributes: year: 2015 holdings: @attributes: islocal: N |
|---|