Latency lags bandwidth
Abstract
Summary form only given. As I review performance trends, I am struck by a consistent theme across many technologies over many years: bandwidth improves much more quickly than latency for four different technologies: disks, networks, memories and processors. A rule of thumb to quantify the imbalance is: bandwidth improves by more than the square of the improvement in latency. This paper lists a half-dozen performance milestones to document this observation, many reasons why it happens, a few ways to cope with it, and two small examples of how you might design systems differently if you kept this simple rule of thumb in mind.
Integrity signals
Registry
Cites
Statistics
Claims the paper makes
Claims other papers make about this one
Each is checked against this paper’s own text: does it say what the citing paper says it does?
For the recent past and foreseeable future, off-chip memory bandwidth will often be the constraining resource for floating-point computations.
“We believe that for the recent past and foreseeable future, off-chip memory bandwidth will often be the constraining resource [23].”
From Roofline: An Insightful Visual Performance Model for Floating-Point Programs and Multicore Architectures (2009) · cites this paper as [23]
Cited by
Sorted by citations
Williams, S., Waterman, A., & Patterson, D. A. (2009) · Communications of the ACM · doi:10.1145/1498765.1498785 · Cited by 2,484 · see citing works
Self-citationWilliams, S., Waterman, A., & Patterson, D. (2009) · doi:10.2172/1407078 · Cited by 110 · see citing works
Self-citation
- Authors
- Patterson, D.
- Affiliations
- University of California, Berkeley (US)
- Corresponding author
- Patterson, D.
- ORCID
- Patterson, D.
- Venue
- 2005 International Conference on Computer Design
- Publisher
- IEEE Comput. Soc
- Published
- 2006-08-15
- Pages
- 3
- ISSN
- 1063-6404 (ISSN-L)
- OpenAlex
- W1484897302
- Cited by
- 20
- References
- 0
- Citations by year
- 2019: 1 · 2018: 1 · 2017: 1 · 2015: 1 · 2014: 4 · 2012: 2
- Type
- conference abstract
- Language
- en
- Open access
- closed
- Locations
- IEEE International Conference on Computer Design: VLSI in Computers and Processors (ICCD) (publishedVersion)
- Topics
- Physical Sciences › Computer Science › Hardware and Architecture › Parallel Computing and Optimization Techniques
- Physical Sciences › Computer Science › Computer Networks and Communications › Software System Performance and Reliability
- Physical Sciences › Computer Science › Computer Networks and Communications › Interconnection Networks and Systems
- Keywords
- latency, bandwidth, performance trends, disk performance, network performance, memory performance, processor performance, system design
About these checks
Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.
Statistics and arithmetic are recomputed, not judged.
Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.
What counts as a claim ›How we measure confidence ›How we check this ›
Are you an author, or do you see a problem with this page? Request a review
