Add to Watchlist

Timely Dataflow in Rust

7 views

Citation of segment
Embed Code
Purchasing a DVD Cite video

Automated Media Analysis

Beta
Recognized Entities
Speech transcript
and so the all the way on the hand of less occur along the so far we've you know that kind of more traditional by allowing the size name leading rocks and the only thing that I would have had a list of all right out of the usage of the where this is known as the thing I through with that and the use of the of the of the of the role of resources research was sometimes the problems the of the world that the problem rather than the effect of that is the only 1 who has the you're probably right but I think there is interesting question of of the the the in the this is a product of working throughout the year the build of working on where the notice is that the sum of the goods parts of the data the system from the probability of that he that and approach doing most of the cells without the performance of the solar system 1 of them will show you what it was not in the world the line it's a little bit of the monsters so you see the word you have the start of the the number is that the verdict of the when evaluating the systems that most of the time of the conference of the American players in the view of the graph of the system and at the core of the system on the rest of you in right and they're they're really this is going to be the sum of the number of the rights of people in all those little red spot and so forth over the not all these things in the limit is the gross continued with the end of the year your and it turns out that you know what you write about the minds of the the performance that outperforming the process the the way the website of you the program that you but the system on the top of our research is all over the structure of the trace of the you might say well look we get all the Russian scientists are incredibly also has a single headed of so this there are 2 that then the rest of the graph of the this is more of the of the of the medical so you reality the memory but not play around with the range of the the beginning of the of the vertices of 20 times groups of nodes in the race in all the edges and increments and with the process that's because all over the world know what department of the sort of thing so I forget make inferences about his later was get the 2nd moment later or a little of the of the worst my left up because you're using computers not well and that makes this a voting instead of having it was the core of the capabilities of doing things that would so the project and it is something I through each of these words means of whether it's a year and I usually non-technical ways of thinking about the fact that the things that you agree that you set rest is the great new language what was the job of the and here we go a little and it has to do with the amount of experience of the writing of the performance of the people of the but also having to but when they do that give up some of the final abstractions that relational so this turning assistance through language and they think that means the performance of the but there you go on the 2 of them that I was going to monitor based resources and is but in contrast with the the people you explicit minute here in sort of the thinking of the the of the of the agreement with 1 of the current collection of data for you but you don't really help me but warehouse this In order to this problem and this was like 1 of the things I have the memory and the I go away the memory of the way it it is likely that the people in the brain was you get the look-and-feel of garbage-collected languages that and really work for performance actually distributed systems where the prediction of what the the they get holding the to the idea of the of the rules of the the of but these are highly sensitive useful on the iterators closes the little amount of functions use they all have the property of basically the good life and you manually and from the renewal of the the everything works of was in most of the people so in my experience it's been known as the people of the language with the performance of the image which it the universe but this to the thing with this the use of of the but usually put together performance the system players of something like that the get some data for so many was married a lot of a lot of work still included in the back the model of model based on directed graphs where you have a vertices would respond computation of the vertices and edges and preventing the use of all the edges in the beginning we have the data for like the rain announcing 1 of my yeah and they move around between the program as the land at the end of the computation of the the the the use of the logic of knowledge of the communications I want to look at the edges of the node may be scaled by the and I just the result of all of you to sort this and they have a little further on we have segregation of those linear linear things up and understanding is known as the place the based because of what and so have you all the all the of so this this is that while the properties of the view I don't want to differently right parameters in the writing of the report of the this will be but when you start your application and data-driven model you let go of it is the role of the data moving around what gets the running and this is the way that I was in the will of the 1st year of his 1 of the workers machines the competition do you think the I right of is of those really nicely roughly 1 that moving data around the rest of this is very very confident in using the old
and the passage of time the model is that
countries in the world of you know things around because of the preceding and and or operators we look at what you know what you have like to you or and is that this is a little the the the family all of this is the 1st thing I you want in real time from a from a stream of edges and you look at the reason that we know of that also salutes going on is viewed as a means of a times the actual what kinds of reference within a year of what was going on around at the end of the world and that the best students of the constraint on the net and there was a lot of iterations of each of the conditions of the of the responses so what is the meaning of and the validity of the law of the universe all this over here where each of these stages is good for all of this and that we do not know the answer to the here this is known the goal with with the to the whole reason for this is where I work on it and it was sort of progress in the end of the of the of the of the of the of the of the of the of the of the of the world of the work and the latency of the lines we don't want to be a consequence of the order in which these people in the world in the scalability of the other the is the theory of human the end of day the originally progress was essentially where you will be in the hands of a total of 100 and this is the coolest thing in the news so so that the idea of it and so on and what sort of the beginning of a word the of the data vectors of the of the sort of thing that would you use I you work you will where you have to think of really what do you think of what that the the work of the rest of and it is important that the general at the the time for this and you have a 2nd or 2 of the report what they're doing this and the fact that the presence or status of also advertising in of the year the yeah but you're right so it has a lot of talk about it of the of the of the incremental computation and that's the general that use of the actually really really really begins on the basis of the of the and the goal of this article of their birth of where forms do you want to be distributing the loss of life of the of the of the of the of of this really cool the start of the article and the roads were done with of 1 that we will be on I think we should have a lot more than the rest of the way optimizes the statement to binary and so on and so physical well suppose that all of them can be together and it you to learn about and also you select your the good of the of the of the graph of library in England but it is usually done by the most of the theory the on the most of the people who but I would say that you can think of a form of of of of the future versions of the solid and the the last that we read about this but I don't know this flexible and what you want to get out the yesterday well defined in and of themselves for the last and so you can also have a very small family metal itself and the regret to the Government beginning the so show you that in that so you have also of course something be so I parameters here is the single thing and we have the computational social you and and I don't know what we're going for a given universe so I the different go so it is not the same we have conversations about where the gap is there a you know the position to the end of the people
in the use of the phrase so get and this is the same thinking on graphics family running and utility of the principles of this because you didn't get the 2nd half of the the entire remind the variables if the the on this 61 of anyone 1 so these these machines using our in getting up to about 70 % of variables in order the value of the projection of the vector at the end of the use
of information about this the year that it was sort of acquiring language you know some about time I don't think when you get to the end of the after the war and here is that the 1 on the which is included in the years and use and the 1st thing things to do it the wrong way is using it for a given write a vertex that those those thinking usually get what I want to do and we can use it right for the the what the world of the the the data on the the and my or a prize that not you have to have some of my actually the the data and the location of the I will have a lot of the memory access to the knowledge of the theories that we use to prevent this sort of thing you do so if you're put the writing can actually look be then you should to the that the you know the size of the data throughout but it's all basically near-zero 0 copy they get down to the network status and the length of the materials in this this and you know of the work that you have to read the program and the In this kind of where you want to do what what said this and by the beauty of the union of the this example of the really want to give us the numbers for this example of things on the edges of the and what do they where and I you get the numbers because the optimization has to get by the that's interesting obligated to do that which is that the hearing of the because data about the issue of the receptor and the quality of the work is the problem is to of so I can relate with to put
together the performance of of the remote and is an open question was visible or would you use for this is the last slide to remind everyone that you put in the grounds of the data and and you know what we're going to vote is that of the of the of the of the of the the thank thank actions well you the viruses and held the view from the In the question was about as it was using the powerful and there is an absolutely yes by the leaders of the origin of the properties of the rest of the world this is the data from the Supernova by a lot when you read programming think who it was going to show you data that you I don't think this is 1 way to do it now and I only allowed to look at 1 1 really little with the realization that most of the original start by the year accuracy of the best you can trust the reference of the people and you know you and that the full the use of the data itself you wanted to know what you think about the direction of the all of the binary with some pointers we your article on on this in the era of of the of the although the distinguishes between the performance of the unless you so I irrespective of appropriate analytical because it remains unclear whether the people want a lot of the people are violence and very early and the rest and that's all for you in the solution of the elderly that's a and which is not so a 3 answers what rate of 1 of the things we know about the best is to it is a regret of the new there are always going to be about structure but I think it would go up because of a good number of you might think that this was the availability of things I don't actually know the mean of the values that people in the world around lately has been that makes with the of the we will use of the and and you have the memory because of the use of force power showing a lot of data from various collections of the hormones and so on these are all very very good so the things from my point of view how did you get another of this and of this is really this is the gradient your bank and you know you want something that actually work all the time so it's not watching this as the variances get rid of the so much else the the another way of looking really don't which is the regular version of the top is well it has much more theory of deterministic semantics about what are what some of the things that we're going to have a loss the here and it was particularly in light of the data around I may not be the right thing to do is I don't so these are the questions in of the going and make so there is no sort of most of in don't the the the what do the other thing is you need to of the the the yeah so this is about so in the service of the data and the data were produced by the time of the year and so the 3 reviewers of all around but I would ask you in the network be beginning and I think this is a great thing to think about what the the the the the definition is that was the the the minimum and without
Manufacturing execution system
Graph (mathematics)
Scientific modelling
Multiplication sign
View (database)
Range (statistics)
Compiler
Parameter (computer programming)
Mereology
Inference
Formal language
Data model
Medical imaging
Video game
Document management system
Algebraic closure
Memory management
Core dump
Monster group
Physical system
Metropolitan area network
Process (computing)
Product (category theory)
Moment (mathematics)
Electronic mailing list
Sound effect
Bit
Prediction
Functional (mathematics)
Category of being
Process (computing)
Telecommunication
Order (biology)
System programming
Website
Right angle
Energy level
Quicksort
Resultant
Writing
Laptop
Ocean current
Computer programming
Read-only memory
Cellular automaton
Virtual machine
Graph (mathematics)
Discrete element method
Rule of inference
Number
Goodness of fit
Degree (graph theory)
Ranking
Data structure
Contrast (vision)
Traffic reporting
Renewal theory
Graph (mathematics)
Dataflow
Computer
Cellular automaton
Projective plane
Code
Line (geometry)
Cartesian coordinate system
Limit (category theory)
Computer programming
Local Group
Supercomputer
Summation
Word
Keilförmige Anordnung
Voting
Computer animation
Personal digital assistant
Logic
Universe (mathematics)
Vertex (graph theory)
Iteration
Abstraction
Algorithm
Graph (mathematics)
Scientific modelling
Multiplication sign
Insertion loss
Real-time operating system
Parameter (computer programming)
Streaming media
Student's t-test
Weight
Discrete element method
Scalability
Theory
Timestamp
Revision control
Goodness of fit
Video game
Operator (mathematics)
Data conversion
Traffic reporting
Position operator
Form (programming)
Condition number
Constraint (mathematics)
Graph (mathematics)
Dataflow
Validity (statistics)
Computer
Concurrency (computer science)
Physical law
Basis (linear algebra)
Total S.A.
Line (geometry)
Control flow
Word
Arithmetic mean
Vector space
Order (biology)
Universe (mathematics)
Statement (computer science)
Dependent and independent variables
Right angle
Iteration
Ideal (ethics)
Quicksort
Family
Arithmetic progression
Library (computing)
Computer programming
Read-only memory
Length
Multiplication sign
Virtual machine
Materialization (paranormal)
Theory
Storage area network
Number
Formal language
Utility software
Dataflow
Information
Projective plane
Variable (mathematics)
Inclusion map
Uniform resource locator
Vector space
Order (biology)
Computer network
Vertex (graph theory)
Hill differential equation
Quicksort
Family
Mathematical optimization
Writing
Point (geometry)
Computer virus
Computer programming
Slide rule
Group action
Service (economics)
Direction (geometry)
Multiplication sign
View (database)
Determinism
Control engineering
Insertion loss
Discrete element method
Semantics (computer science)
Theory
Power (physics)
Number
Revision control
Bit rate
Data structure
Dataflow
Forcing (mathematics)
Gradient
Variance
Core dump
Supercomputer
Category of being
Arithmetic mean
Pointer (computer programming)
Computer animation
Computer network
Quicksort
Identical particles

Metadata

Formal Metadata

Title Timely Dataflow in Rust
Subtitle Big Data programmability and HPC performance
Title of Series FOSDEM 2016
Part Number 39
Number of Parts 110
Author McSherry, Frank
License CC Attribution 2.0 Belgium:
You are free to use, adapt and copy, distribute and transmit the work or content in adapted or unchanged form for any legal purpose as long as the work is attributed to the author in the manner specified by the author or licensor.
DOI 10.5446/30950
Publisher FOSDEM VZW
Release Date 2016
Language English

Content Metadata

Subject Area Information technology

Recommendations

Loading...
Feedback
AV-Portal 3.5.0 (cb7a58240982536f976b3fae0db2d7d34ae7e46b)

Timings

  620 ms - page object