DOCSLIB.ORG

I/O Speculation for the Microsecond

Total Page：16

File Type：pdf, Size：1020Kb

Download full-text PDF Read full-text
I/O Speculation for the Microsecond I/O Speculation for the Microsecond Era Michael Wei, University of California, San Diego; Matias Bjørling and Philippe Bonnet, IT University of Copenhagen; Steven Swanson, University of California, San Diego https://www.usenix.org/conference/atc14/technical-sessions/presentation/wei This paper is included in the Proceedings of USENIX ATC ’14: 2014 USENIX Annual Technical Conference. June 19–20, 2014 • Philadelphia, PA 978-1-931971-10-2 Open access to the Proceedings of USENIX ATC ’14: 2014 USENIX Annual Technical Conference is sponsored by USENIX. I/O Speculation for the Microsecond Era Michael Wei†, Matias Bjørling‡, Philippe Bonnet‡, Steven Swanson† †University of California, San Diego ‡IT University of Copenhagen Device Read Write Millisecond Scale Abstract 10G Intercontinental RPC 100 ms 100 ms Microsecond latencies and access times will soon domi- 10G Intracontinental RPC 20 ms 20 ms nate most datacenter I/O workloads, thanks to improve- Hard Disk 10 ms 10 ms ments in both storage and networking technologies. Cur- 10G Interregional RPC 1 ms 1 ms rent techniques for dealing with I/O latency are targeted Microsecond Scale for either very fast (nanosecond) or slow (millisecond) 10G Intraregional RPC 300 µs 300 µs devices. These techniques are suboptimal for microsec- SATA NAND SSD 200 µs 50 µs ond devices - they either block the processor for tens PCIe/NVMe NAND SSD 60 µs 15 µs of microseconds or yield the processor only to be ready 10Ge Inter-Datacenter RPC 10 µs 10 µs again microseconds later. Speculation is an alternative 40Ge Inter-Datacenter RPC 5 µs 5 µs technique that resolves the issues of yielding and block- PCM SSD 5 µs 5 µs ing by enabling an application to continue running un- Nanosecond Scale til the application produces an externally visible side ef- 40 Gb Intra-Rack RPC 100 ns 100 ns fect. State-of-the-art techniques for speculating on I/O DRAM 10 ns 10 ns requests involve checkpointing, which can take up to a STT-RAM <10 ns <10 ns millisecond, squandering any of the performance bene- fits microsecond scale devices have to offer. In this paper, Table 1: I/O device latencies. Typical random read and we survey how speculation can address the challenges write latencies for a variety of I/O devices. The major- that microsecond scale devices will bring. We mea- ity of I/Os in the datacenter will be in the microsecond sure applications for the potential benefit to be gained range. from speculation and examine several classes of specu- lation techniques. In addition, we propose two new tech- opments is that the datacenter will soon be dominated by niques, hardware checkpoint and checkpoint-free spec- microsecond-scale I/O requests. ulation. Our exploration suggests that speculation will Today, an operating system uses one of two options enable systems to extract the maximum performance of when an application makes an I/O request: either it can I/O devices in the microsecond era. block and poll for the I/O to complete, or it can com- plete the I/O asynchronously by placing the request in 1 Introduction a queue and yielding the processor to another thread or application until the I/O completes. Polling is an We are at the dawn of the microsecond era: current effective strategy for devices with submicrosecond la- state-of-the-art NAND-based Solid State Disks (SSDs) tency [2, 20], while programmers have used yielding offer latencies in the sub-100µs range at reasonable and asynchronous I/O completion for decades on devices cost [16, 14]. At the same time, improvements in net- with millisecond latencies, such as disk. Neither of these work software and hardware have brought network laten- strategies, however, is a perfect fit for microsecond-scale cies closer to their physical limits, enabling sub-100µs I/O requests: blocking will prevent the processor from communication latencies. The net result of these devel- doing work for tens of microseconds, while yielding may USENIX Association 2014 USENIX Annual Technical Conference 475 reduce performance by increasing the overhead of each 2.1 Interfaces versus Strategies I/O operation. A third option exists as a solution for dispatching I/O When an application issues a request for an I/O, it uses an requests, speculation. Under the speculation strategy, the interface to make that request. A common example is the operating system completes I/O operations speculatively, POSIX write/read interface, where applications make returning control to the application without yielding. The I/O requests by issuing blocking calls. Another example operating system monitors the application: in the case of is the POSIX asynchronous I/O interface, in which ap- a write operation, the operating system blocks the appli- plications enqueue requests to complete asynchronously cation if it makes a side-effect, and in the case of a read and retrieve the status of their completion at some later operation, the operating system blocks the application if time. it attempts to use data that the OS has not read yet. In Contrast interfaces with strategies, which refers to addition, the operating system may have a mechanism to how the operating system actually handles I/O requests. rollback if the I/O operation does not complete success- For example, even though the write interface is block- fully. By speculating, an application can continue to do ing, the operating system may choose to handle the I/O useful work even if the I/O has not completed. In the asynchronously, yielding the processor to some other context of microsecond-scale I/O, speculation can be ex- thread. tremely valuable since, as we discuss in the next section, In this work, we primarily discuss operating system there is often enough work available to hide microsecond strategies for handling I/O requests, and assume that ap- latencies. We expect that storage class memories, such plication developers are free to choose interfaces. as phase-change memory (PCM), will especially benefit from speculation since their access latencies are unpre- 2.2 Asynchronous I/O - Yielding dictable and variable [12]. Any performance benefit to be gained from specu- Yielding, or the asynchronous I/O strategy, follows the lation is dependent upon the performance overhead of traditional pattern for handling I/O requests within the speculating. Previous work in I/O speculation [9, 10] operating system: when a userspace thread issues an I/O has relied on checkpointing to enable rollback in case request, the I/O subsystem issues the request and the of write failure. Even lightweight checkpointing, which scheduler places the thread in an I/O wait state. Once utilizes copy-on-write techniques, has a significant over- the I/O device completes the request, it informs the oper- head which can exceed the access latency of microsec- ating system, usually by means of a hardware interrupt, ond devices. and the operating system then places the thread into a ready state, which enables the thread to resume when it In this paper, we survey speculation in the context is rescheduled. of microsecond-scale I/O devices, and attempt to quan- Yielding has the advantage of allowing other tasks to tify the performance gains that speculation has to of- utilize the CPU while the I/O is being processed. How- fer. We then explore several techniques for speculation, ever, yielding introduces significant overhead, which is which includes exploring existing software-based check- particularly relevant for fast I/O devices [2, 20]. For ex- pointing techniques. We also propose new techniques ample, yielding introduces contexts switches, cache and which exploit the semantics of the traditional I/O inter- TLB pollution as well as interrupt overhead that may ex- face. We find that while speculation could allow us to ceed the cost of doing the I/O itself. These overheads maximize the performance of microsecond scale devices, are typically in the microsecond range, which makes the current techniques for speculation cannot deliver the per- cost of yielding minimal when dealing with millisecond formance which microsecond scale devices require. latencies as with disks and slow WANs, but high when dealing with nanosecond devices, such as fast NVMs. 2 Background 2.3 Synchronous I/O - Blocking Past research has shown that current systems have built- in the assumption that I/O is dominated by millisecond Blocking, or the synchronous I/O strategy, is a solution scale requests [17, 2]. These assumptions have impacted for dealing with devices like fast NVMs. Instead of the core design of the applications and operating systems yielding the CPU in order for I/O to complete, block- we use today, and may not be valid in a world where ing prevents unnecessary context switches by having the I/O is an order of magnitude faster. In this Section, we application poll for I/O completions, keeping the entire discuss the two major strategies for handling I/O and context of execution within the executing thread. Typi- show that they do not adequately address the needs of cally, the application stays in a spin-wait loop until the microsecond-scale devices, and we give an overview of I/O completes, and resumes execution once the device I/O speculation. flags the I/O as complete. 476 2014 USENIX Annual Technical Conference USENIX Association Blocking prevents the CPU from incurring the cost of t Yielding app2 context switches, cache and TLB pollution as well as in- t t terrupt overhead that the yielding strategy incurs. How- sched sched t t ever, the CPU is stalled for the amount of time the I/O io_start io_comp takes to complete. If the I/O is fast, then this strategy is optimal since the amount of time spent waiting is much shorter than the amount of CPU time lost due to soft- tdevice ware overheads.
Load more

Recommended publications
  • Time and Frequency Users' Manual
    ,>'.)*• r>rJfl HKra mitt* >\ « i If I * I IT I . Ip I * .aference nbs Publi- cations / % ^m \ NBS TECHNICAL NOTE 695 U.S. DEPARTMENT OF COMMERCE/National Bureau of Standards Time and Frequency Users' Manual 100 .U5753 No. 695 1977 NATIONAL BUREAU OF STANDARDS 1 The National Bureau of Standards was established by an act of Congress March 3, 1901. The Bureau's overall goal is to strengthen and advance the Nation's science and technology and facilitate their effective application for public benefit To this end, the Bureau conducts research and provides: (1) a basis for the Nation's physical measurement system, (2) scientific and technological services for industry and government, a technical (3) basis for equity in trade, and (4) technical services to pro- mote public safety. The Bureau consists of the Institute for Basic Standards, the Institute for Materials Research the Institute for Applied Technology, the Institute for Computer Sciences and Technology, the Office for Information Programs, and the Office of Experimental Technology Incentives Program. THE INSTITUTE FOR BASIC STANDARDS provides the central basis within the United States of a complete and consist- ent system of physical measurement; coordinates that system with measurement systems of other nations; and furnishes essen- tial services leading to accurate and uniform physical measurements throughout the Nation's scientific community, industry, and commerce. The Institute consists of the Office of Measurement Services, and the following center and divisions: Applied Mathematics
    [Show full text]
  • HOW FAST IS RF? by RON HRANAC
    Originally appeared in the September 2008 issue of Communications Technology. HOW FAST IS RF? By RON HRANAC Maybe a better question is, "What's the speed of an electromagnetic signal in a vacuum?" After all, what we call RF refers to signals in the lower frequency or longer wavelength part of the electromagnetic spectrum. Light is an electromagnetic signal, too, existing in the higher frequency or shorter wavelength part of the electromagnetic spectrum. One way to look at it is to assume that RF is nothing more than really, really low frequency light, or that light is really, really high frequency RF. As you might imagine, the numbers involved in describing RF and light can be very large or very small. For more on this, see my June 2007 CT column, "Big Numbers, Small Numbers" (www.cable360.net/ct/operations/techtalk/23783.html). According to the National Institute of Standards and Technology, the speed of light in a vacuum, c0, is 299,792,458 meters per second (http://physics.nist.gov/cgi-bin/cuu/Value?c). The designation c0 is one that NIST uses to reference the speed of light in a vacuum, while c is occasionally used to reference the speed of light in some other medium. Most of the time the subscript zero is dropped, with c being a generic designation for the speed of light. Let's put c0 in terms that may be more familiar. You probably recall from junior high or high school science class that the speed of light is 186,000 miles per second.
    [Show full text]
  • The Impact of the Speed of Light on Financial Markets and Their Regulation
    When Finance Meets Physics: The Impact of the Speed of Light on Financial Markets and their Regulation by James J. Angel, Ph.D., CFA Associate Professor of Finance McDonough School of Business Georgetown University 509 Hariri Building Washington DC 20057 USA 1.202.687.3765 [email protected] The Financial Review, Forthcoming, May 2014 · Volume 49 · No. 2 Abstract: Modern physics has demonstrated that matter behaves very differently as it approaches the speed of light. This paper explores the implications of modern physics to the operation and regulation of financial markets. Information cannot move faster than the speed of light. The geographic separation of market centers means that relativistic considerations need to be taken into account in the regulation of markets. Observers in different locations may simultaneously observe different “best” prices. Regulators may not be able to determine which transactions occurred first, leading to problems with best execution and trade- through rules. Catastrophic software glitches can quantum tunnel through seemingly impregnable quality control procedures. Keywords: Relativity, Financial Markets, Regulation, High frequency trading, Latency, Best execution JEL Classification: G180 The author is also on the board of directors of the Direct Edge stock exchanges (EDGX and EDGA). I wish to thank the editor and referee for extremely helpful comments and suggestions. I also wish to thank the U.K. Foresight Project, for providing financial support for an earlier version of this paper. All opinions are strictly my own and do not necessarily represent those of Georgetown University, Direct Edge, the U.K. Foresight Project, The Financial Review, or anyone else for that matter.
    [Show full text]
  • Using Microsecond Single-Molecule FRET to Determine the Assembly
    Using microsecond single-molecule FRET to determine PNAS PLUS the assembly pathways of T4 ssDNA binding protein onto model DNA replication forks Carey Phelpsa,b,1, Brett Israelsa,b, Davis Josea, Morgan C. Marsha,b, Peter H. von Hippela,2, and Andrew H. Marcusa,b,2 aDepartment of Chemistry and Biochemistry, Institute of Molecular Biology, University of Oregon, Eugene, OR 97403; and bDepartment of Chemistry and Biochemistry, Oregon Center for Optical, Molecular and Quantum Science, University of Oregon, Eugene, OR 97403 Edited by Stephen C. Kowalczykowski, University of California, Davis, CA, and approved March 20, 2017 (received for review December 2, 2016) DNA replication is a core biological process that occurs in pro- complete coverage of the exposed ssDNA templates at the precisely karyotic cells at high speeds (∼1 nucleotide residue added per regulated concentration of protein that needs to be maintained in millisecond) and with high fidelity (fewer than one misincorpora- the infected Escherichia coli cell (8, 9). The gp32 protein has an tion event per 107 nucleotide additions). The ssDNA binding pro- N-terminal domain, a C-terminal domain, and a core domain. tein [gene product 32 (gp32)] of the T4 bacteriophage is a central The N-terminal domain is necessary for the cooperative binding integrating component of the replication complex that must con- of the gp32 protein through its interactions with the core domain of tinuously bind to and unbind from transiently exposed template an adjacent gp32 protein. To bind to ssDNA, the C-terminal domain strands during DNA synthesis. We here report microsecond single- of the gp32 protein must undergo a conformational change that molecule FRET (smFRET) measurements on Cy3/Cy5-labeled primer- exposes the positively charged region of its core domain, which in template (p/t) DNA constructs in the presence of gp32.
    [Show full text]
  • Aquatic Primary Productivity Field Protocols for Satellite Validation and Model Synthesis (DRAFT)
    Ocean Optics & Biogeochemistry Protocols for Satellite Ocean Colour Sensor Validation IOCCG Protocol Series Volume 7.0, 2021 Aquatic Primary Productivity Field Protocols for Satellite Validation and Model Synthesis (DRAFT) Report of a NASA-sponsored workshop with contributions (alphabetical) from: William M. Balch Bigelow Laboratory for Ocean Sciences, Maine, USA Magdalena M. Carranza Monterey Bay Aquarium Research Institute, California, USA Ivona Cetinic University Space Research Association, NASA Goddard Space Flight Center, Maryland, USA Joaquín E. Chaves Science Systems and Applications, Inc., NASA Goddard Space Flight Center, Maryland, USA Solange Duhamel University of Arizona, Arizona, USA Zachary K. Erickson University Space Research Association, NASA Goddard Space Flight Center, Maryland, USA Andrea J. Fassbender NOAA Pacific Marine Environmental Laboratory, Washington, USA Ana Fernández-Carrera Leibniz Institute for Baltic Sea Research Warnemünde, Rostock, Germany Sara Ferrón University of Hawaii at Manoa, Hawaii, USA E. Elena García-Martín National Oceanography Centre, Southampton, UK Joaquim Goes Lamont Doherty Earth Observatory at Columbia University, New York, USA Helga do Rosario Gomes Lamont Doherty Earth Observatory at Columbia University, New York, USA Maxim Y. Gorbunov Department of Marine and Coastal Sciences, Rutgers University, New Jersey, USA Kjell Gundersen Plankton Research Group, Institute of Marine Research, Bergen, Norway Kimberly Halsey Department of Microbiology, Oregon State University, Oregon, USA Toru Hirawake
    [Show full text]
  • Nanosecond X-Ray Photon Correlation Spectroscopy Using Pulse Time
    research papers Nanosecond X-ray photon correlation spectroscopy IUCrJ using pulse time structure of a storage-ring source ISSN 2052-2525 NEUTRONjSYNCHROTRON Wonhyuk Jo,a Fabian Westermeier,a Rustam Rysov,a Olaf Leupold,a Florian Schulz,b,c Steffen Tober,b‡ Verena Markmann,a Michael Sprung,a Allesandro Ricci,a Torsten Laurus,a Allahgholi Aschkan,a Alexander Klyuev,a Ulrich Trunk,a Heinz Graafsma,a Gerhard Gru¨bela,c and Wojciech Rosekera* Received 25 September 2020 Accepted 2 December 2020 aDeutsches Elektronen-Synchrotron (DESY), Notkestr. 85, 22607 Hamburg, Germany, bInstitute of Physical Chemistry, University of Hamburg, Grindelallee 117, 20146 Hamburg, Germany, and cThe Hamburg Centre for Ultrafast Imaging, Luruper Chaussee 149, 22761 Hamburg, Germany. *Correspondence e-mail: [email protected] Edited by Dr T. Ishikawa, Coherent X-ray Optics Laboratory, Harima Institute, RIKEN, Japan X-ray photon correlation spectroscopy (XPCS) is a routine technique to study slow dynamics in complex systems at storage-ring sources. Achieving ‡ Current address: Deutsches Elektronen- Synchrotron (DESY), Notkestr. 85, 22607 nanosecond time resolution with the conventional XPCS technique is, however, Hamburg, Germany. still an experimentally challenging task requiring fast detectors and sufficient photon flux. Here, the result of a nanosecond XPCS study of fast colloidal Keywords: materials science; nanoscience; dynamics is shown by employing an adaptive gain integrating pixel detector SAXS; dynamical studies; time-resolved studies; (AGIPD) operated at frame rates of the intrinsic pulse structure of the storage X-ray photon correlation spectroscopy; adaptive ring. Correlation functions from single-pulse speckle patterns with the shortest gain integrating pixel detectors; storage rings; pulse structures.
    [Show full text]
  • Low Latency – How Low Can You Go?
    WHITE PAPER Low Latency – How Low Can You Go? Low latency has always been an important consideration in telecom networks for voice, video, and data, but recent changes in applications within many industry sectors have brought low latency right to the forefront of the industry. The finance industry and algorithmic trading in particular, or algo-trading as it is known, is a commonly quoted example. Here latency is critical, and to quote Information Week magazine, “A 1-millisecond advantage in trading applications can be worth $100 million a year to a major brokerage firm.” This drives a huge focus on all aspects of latency, including the communications systems between the brokerage firm and the exchange. However, while the finance industry is spending a lot of money on low- latency services between key locations such as New York and Chicago or London and Frankfurt, this is actually only a small part of the wider telecom industry. Many other industries are also now driving lower and lower latency in their networks, such as for cloud computing and video services. Also, as mobile operators start to roll out 5G services, latency in the xHaul mobile transport network, especially the demanding fronthaul domain, becomes more and more important in order to reach the stringent 5G requirements required for the new class of ultra-reliable low-latency services. This white paper will address the drivers behind the recent rush to low- latency solutions and networks and will consider how network operators can remove as much latency as possible from their networks as they also race to zero latency.
    [Show full text]
  • Timestamp Precision
    Timestamp Precision Kevin Bross 15 April 2016 15 April 2016 IEEE 1904 Access Networks Working Group, San Jose, CA USA 1 Timestamp Packets The orderInfo field can be used as a 32-bit timestamp, down to ¼ ns granularity: 1 s 30 bits of nanoseconds 2 ¼ ms µs ns ns Two main uses of timestamp: – Indicating start or end time of flow – Indicating presentation time of packets for flows with non-constant data rates s = second ms = millisecond µs = microsecond ns = nanosecond 15 April 2016 IEEE 1904 Access Networks Working Group, San Jose, CA USA 2 Presentation Time To reduce bandwidth during idle periods, some protocols will have variable rates – Fronthaul may be variable, even if rate to radio unit itself is a constant rate Presentation times allows RoE to handle variable data rates – Data may experience jitter in network – Egress buffer compensates for network jitter – Presentation time is when the data is to exit the RoE node • Jitter cleaners ensure data comes out cleanly, and on the right bit period 15 April 2016 IEEE 1904 Access Networks Working Group, San Jose, CA USA 3 Jitter vs. Synchronization Synchronization requirements for LTE are only down to ~±65 ns accuracy – Each RoE node may be off from TAI by up to 65 ns (or more in some circumstances) – Starting and ending a stream may be off by this amount …but jitter from packet to packet must be much tighter – RoE nodes should be able to output data at precise relative times if timestamp is used for a given packet – Relative bit time within a flow is important 15 April 2016 IEEE 1904 Access
    [Show full text]
  • Simulating Nucleic Acids from Nanoseconds to Microseconds
    UC Irvine UC Irvine Electronic Theses and Dissertations Title Simulating Nucleic Acids from Nanoseconds to Microseconds Permalink https://escholarship.org/uc/item/6cj4n691 Author Bascom, Gavin Dennis Publication Date 2014 Peer reviewed|Thesis/dissertation eScholarship.org Powered by the California Digital Library University of California UNIVERSITY OF CALIFORNIA, IRVINE Simulating Nucleic Acids from Nanoseconds to Microseconds DISSERTATION submitted in partial satisfaction of the requirements for the degree of DOCTOR OF PHILOSOPHY in Chemistry, with a specialization in Theoretical Chemistry by Gavin Dennis Bascom Dissertation Committee: Professor Ioan Andricioaei, Chair Professor Douglas Tobias Professor Craig Martens 2014 Appendix A c 2012 American Chemical Society All other materials c 2014 Gavin Dennis Bascom DEDICATION To my parents, my siblings, and to my love, Lauren. ii TABLE OF CONTENTS Page LIST OF FIGURES vi LIST OF TABLES x ACKNOWLEDGMENTS xi CURRICULUM VITAE xii ABSTRACT OF THE DISSERTATION xiv 1 Introduction 1 1.1 Nucleic Acids in a Larger Context . 1 1.2 Nucleic Acid Structure . 5 1.2.1 DNA Structure/Motion Basics . 5 1.2.2 RNA Structure/Motion Basics . 8 1.2.3 Experimental Techniques for Nucleic Acid Structure Elucidation . 9 1.3 Simulating Trajectories by Molecular Dynamics . 11 1.3.1 Integrating Newtonian Equations of Motion and Force Fields . 12 1.3.2 Treating Non-bonded Interactions . 15 1.4 Defining Our Scope . 16 2 The Nanosecond 28 2.1 Introduction . 28 2.1.1 Biological Processes of Nucleic Acids at the Nanosecond Timescale . 29 2.1.2 DNA Motions on the Nanosecond Timescale . 32 2.1.3 RNA Motions on the Nanosecond Timescale .
    [Show full text]
  • Discovery of Frequency-Sweeping Millisecond Solar Radio Bursts Near 1400 Mhz with a Small Interferometer Physics Department Senior Honors Thesis 2006 Eric R
    Discovery of frequency-sweeping millisecond solar radio bursts near 1400 MHz with a small interferometer Physics Department Senior Honors Thesis 2006 Eric R. Evarts Brandeis University Abstract While testing the Small Radio Telescope Interferometer at Haystack Observatory, we noticed occasional strong fringes on the Sun. Upon closer study of this activity, I find that each significant burst corresponds to an X-ray solar flare and corresponds to data observed by RSTN. At very short integrations (512 µs), most bursts disappear in the noise, but a few still show very strong fringes. Studying these <25 events at 512 µs integration reveals 2 types of burst: one with frequency structure and one without. Bursts with frequency structure have a brightness temperature ~1013K, while bursts without frequency structure have a brightness temperature ~108K. These bursts are studied at higher frequency resolution than any previous solar study revealing a very distinct structure that is not currently understood. 1 Table of Contents Abstract ..........................................................................................................................1 Table of Contents...........................................................................................................3 Introduction ...................................................................................................................5 Interferometry Crash Course ........................................................................................5 The Small Radio Telescope
    [Show full text]
  • Microsecond and Millisecond Dynamics in the Photosynthetic
    Microsecond and millisecond dynamics in the photosynthetic protein LHCSR1 observed by single-molecule correlation spectroscopy Toru Kondoa,1,2, Jesse B. Gordona, Alberta Pinnolab,c, Luca Dall’Ostob, Roberto Bassib, and Gabriela S. Schlau-Cohena,1 aDepartment of Chemistry, Massachusetts Institute of Technology, Cambridge, MA 02139; bDepartment of Biotechnology, University of Verona, 37134 Verona, Italy; and cDepartment of Biology and Biotechnology, University of Pavia, 27100 Pavia, Italy Edited by Catherine J. Murphy, University of Illinois at Urbana–Champaign, Urbana, IL, and approved April 11, 2019 (received for review December 13, 2018) Biological systems are subjected to continuous environmental (5–9) or intensity correlation function analysis (10). CPF analysis fluctuations, and therefore, flexibility in the structure and func- bins the photon data, obscuring fast dynamics. In contrast, inten- tion of their protein building blocks is essential for survival. sity correlation function analysis characterizes fluctuations in the Protein dynamics are often local conformational changes, which photon arrival rate, accessing dynamics down to microseconds. allows multiple dynamical processes to occur simultaneously and However, the fluorescence lifetime is a powerful indicator of rapidly in individual proteins. Experiments often average over conformation for chromoproteins and for lifetime-based FRET these dynamics and their multiplicity, preventing identification measurements, yet it is ignored in intensity-based analyses. of the molecular origin and impact on biological function. Green 2D fluorescence lifetime correlation (2D-FLC) analysis was plants survive under high light by quenching excess energy, recently introduced as a method to both resolve fast dynam- and Light-Harvesting Complex Stress Related 1 (LHCSR1) is the ics and use fluorescence lifetime information (11, 12).
    [Show full text]
  • Challenges Using Linux As a Real-Time Operating System
    https://ntrs.nasa.gov/search.jsp?R=20200002390 2020-05-24T04:26:54+00:00Z View metadata, citation and similar papers at core.ac.uk brought to you by CORE provided by NASA Technical Reports Server Challenges Using Linux as a Real-Time Operating System Michael M. Madden* NASA Langley Research Center, Hampton, VA, 23681 Human-in-the-loop (HITL) simulation groups at NASA and the Air Force Research Lab have been using Linux as a real-time operating system (RTOS) for over a decade. More recently, SpaceX has revealed that it is using Linux as an RTOS for its Falcon launch vehicles and Dragon capsules. As Linux makes its way from ground facilities to flight critical systems, it is necessary to recognize that the real-time capabilities in Linux are cobbled onto a kernel architecture designed for general purpose computing. The Linux kernel contain numerous design decisions that favor throughput over determinism and latency. These decisions often require workarounds in the application or customization of the kernel to restore a high probability that Linux will achieve deadlines. I. Introduction he human-in-the-loop (HITL) simulation groups at the NASA Langley Research Center and the Air Force TResearch Lab have used Linux as a real-time operating system (RTOS) for over a decade [1, 2]. More recently, SpaceX has revealed that it is using Linux as an RTOS for its Falcon launch vehicles and Dragon capsules [3]. Reference 2 examined an early version of the Linux Kernel for real-time applications and, using black box testing of jitter, found it suitable when paired with an external hardware interval timer.
    [Show full text]