- firedrake - imperial.ac.uk

12mpi wave_explicit breakdown plots
by Loriant, Nicolas 20 Feb '14

20 Feb '14

2 run, same program each plot is the breakdown of a kernel, kernel 0 is an init par loop (executed only once) kernel 2 and 4 are indirect loops the rest are direct loops in each plot, the steps in executing the par loops is break down (yellow are halo send/recv) blue is execution on core element, green are owned ones, orange are halos

1 0

Re: [firedrake] Regarding using UFL in meshfree code
by Florian Rathgeber 17 Feb '14

17 Feb '14

Dear Sarosh, Apologies for the slow response. Find some answers inline. On 12/02/14 10:44, quraishi(a)math.tu-berlin.de wrote: > Dear Florian > I saw your PyOP2 library and so I got motivated to interface my meshfree > code with high level interfaces like UFL. Glad to hear PyOP2 motivated you to look at UFL and high-level interfaces! I should add that PyOP2 is not "my" library but the effort of our entire research group. > Currently my code is in Matlab, and everything is hardcoded. I wish that > by using UFL I could also generate variational forms using highlevel code. By "everything" I'm assuming you mean the finite-element basis functions, derivatives etc.? You're absolutely right that UFL allows you to generate implementations of variational forms via a form compiler such as the FEniCS form compiler FFC. > Please suggest me how could I use UFL in this context? Is it possible to > generate python code (in addition to C++ code) by using UFL as well? FFC by default generates code conforming to the UFC interface specification. I'm not very familiar with MATLAB and even less with its interfacing capabilities to 3rd party code, but in theory you might be able to compile FFC generated forms into a library and call them from MATLAB. Depending on how invested you are into your MATLAB code base you might find it easier to port it over to Python and use FFC's ability to generate and compile code just-in-time. I'm not sure what exactly you mean by "generate python code": the implementation of the variational form? FFC currently only has a C++ backend and I don't think you would want tight loops and heavy arithmetic in Python for performance reasons. However since UFL is an embedded DSL in Python you can do pretty much whatever you want with the UFL form. UFL comes with a number of transformation algorithms. I'm copying in our mailing list, so maybe others have things to add. Best, Florian > Thanks > Best Regards > Sarosh Quraishi

1 0

hardware prefetching
by Kelly, Paul H J 16 Feb '14

16 Feb '14

We discussed prefetching in the MAPDES meeting the other day. I would love to understand hardware prefetching better. I am, after all, supposed to teach it. Here's one interesting comment: http://www.tomshardware.com/reviews/Intel-i7-nehalem-cpu,2041-12.html (scroll down a page) Basically: prefetching can help a lot and sometimes hurt a bit, but Intel are not saying much about how they fix it. One thought: even if prefetching cannot be improved using explicit instructions, it might be possible to do something usefull with non-temporal stores? Eg see http://blogs.fau.de/hager/archives/2103 But here is what Intel *does* say, about Sandy Bridge but I imagine other processors are similar: Data Prefetch to L1 Data Cache Data prefetching is triggered by load operations when the following conditions are met: • Load is from writeback memory type. [[ie not on direct-mapped I/O pages, and perhaps pinned]] • The prefetched data is within the same 4K byte page as the load instruction that triggered it. • No fence is in progress in the pipeline. • Not many other load misses are in progress. • There is not a continuous stream of stores. Two hardware prefetchers load data to the L1 DCache: • Data cache unit (DCU) prefetcher. This prefetcher, also known as the streaming prefetcher, is triggered by an ascending access to very recently loaded data. The processor assumes that this access is part of a streaming algorithm and automatically fetches the next line. • Instruction pointer (IP)-based stride prefetcher. This prefetcher keeps track of individual load instructions. If a load instruction is detected to have a regular stride, then a prefetch is sent to the next address which is the sum of the current address and the stride. This prefetcher can prefetch forward or backward and can detect strides of up to 2K bytes. Data Prefetch to the L2 and Last Level Cache The following two hardware prefetchers fetched data from memory to the L2 cache and last level cache: Spatial Prefetcher: This prefetcher strives to complete every cache line fetched to the L2 cache with the pair line that completes it to a 128-byte aligned chunk. Streamer: This prefetcher monitors read requests from the L1 cache for ascending and descending sequences of addresses. Monitored read requests include L1 DCache requests initiated by load and store operations and by the hardware prefetchers, and L1 ICache requests for code fetch. When a forward or backward stream of requests is detected, the anticipated cache lines are prefetched. Prefetched cache lines must be in the same 4K page. The streamer and spatial prefetcher prefetch the data to the last level cache. Typically data is brought also to the L2 unless the L2 cache is heavily loaded with missing demand requests. Enhancement to the streamer includes the following features: • The streamer may issue two prefetch requests on every L2 lookup. The streamer can run up to 20 lines ahead of the load request. • Adjusts dynamically to the number of outstanding requests per core. If there are not many outstanding requests, the streamer prefetches further ahead. If there are many outstanding requests it prefetches to the LLC only and less far ahead. • When cache lines are far ahead, it prefetches to the last level cache only and not to the L2. This method avoids replacement of useful cache lines in the L2 cache. • Detects and maintains up to 32 streams of data accesses. For each 4K byte page, you can maintain one forward and one backward stream can be maintained. Furthermore, for Ivy Bridge there is this enhancement (§2.2.7): Hardware prefetch enhancement: A next-page prefetcher (NPP) is added in Intel microarchitecture code name Ivy Bridge. The NPP is triggered by sequential accesses to cache lines approaching the page boundary, either upwards or downwards. (Source: http://www.intel.com/content/dam/www/public/us/en/documents/manuals/64-ia-3… §2.2.5.4).

1 0

High order domains
by Martin Sandve Alnæs 14 Feb '14

14 Feb '14

The input syntax is in place in ufl, and the basic representation types are also mostly in place. What's left in ufl is implementing is_cellwise_constant where necessary and writing the Grad(f) to JacobianInverse(f.domain())*LocalGrad(f) transformation algorutm. Here's the issue https://bitbucket.org/f<https://bitbucket.org/fenics-project/ufl/issue/30/defining-domains-with-hig…> enics-project/ufl/issue/30/defining-domains-with-high-order<https://bitbucket.org/fenics-project/ufl/issue/30/defining-domains-with-hig…> Martin

1 0

PP14 slides
by Lawrence Mitchell 14 Feb '14

14 Feb '14

Hi folks, I've attached the current PP14 slides, they're also at bitbucket.org/mapdes/pp14. The author list is the one for the mini symposium, please shout if you feel left out. All suggestions gratefully received. Presentation is not til Friday next week so I will probably make further alterations. Lawrence

2 2

Fwd: Tomorrow's Professor eNewsletter: 1306. What's All This About Philosophy?
by Kelly, Paul H J 14 Feb '14

14 Feb '14

Something for the weekend.... Begin forwarded message: From: Rick Reis <reis(a)stanford.edu<mailto:reis@stanford.edu>> Date: 13 February 2014 23:44:11 GMT To: <tomorrows-professor(a)lists.stanford.edu<mailto:tomorrows-professor@lists.stanford.edu>> Subject: Tomorrow's Professor eNewsletter: 1306. What's All This About Philosophy? Please click here if e-mail below is not displayed correctly.<http://cgi.stanford.edu/~dept-ctl/cgi-bin/tomprof/enewsletter.php?msgno=1306> [Tomorrow's Professor eNewsletter, Sponsored by Stanford Center for Teaching and Learning] Being aware of your philosophical or theoretical standpoint is an important aspect of your research. The assumptions made at the heart of your investigation have a large bearing on how you approach your subject and investigate the aspects of interest. The opposite poles of approach are positivism - the belief that there is an inherent order in the universe which can be discovered if we possess enough information; and relativism - that as human beings within a culture, we can only understand the world from our own perspective, which could be different from that of other humans in other cultures. 1306. What's All This About Philosophy? Folks: [Rick Reis] The posting below looks at the importance of knowing and articulating your philosophical perspective when writing your thesis or dissertation. It is from Chapter 5, What's All This About Philosophy?, in the book Your Undergraduate Dissertation: The Essential Guide for Success. It is part of the Sage Study Skills series and is by Nicholas Walliman and published by Sage Publications, Inc. 2455 Teller Road Thousand Oaks, California 91320. [www.sagepub.com/‎] Copyright © 2014 Nicholas Walliman. All rights reserved. Reprinted with permission. Regards, Rick Reis reis(a)stanford.edu<mailto:reis@stanford.edu> [cid:0888A9C8-E870-447C-B55B-33B261BA5E0A@Stanford.EDU] UP NEXT: Confirming the MOOC Myth Tomorrow's Research --------- 2,135 words ---------- What's All This About Philosophy? 5.1 Can I believe what I see? Do I see what I believe? You might well ask, 'Why a chapter about philosophy? After all, I'm doing a degree in education, nursing, social science, geography, etc.?' The simple answer is, that your whole life is determined by your philosophical approach, whether you realize it or not. Everyone is a philosopher - everyone has his or her own concept of the world. The alternative to philosophy is not no philosophy but bad philosophy. The 'un-philosophical' person has an unconscious philosophy, which they apply to their practice - whether of science, politics or daily life. (Collier, 1994: 16) The research work involved in doing your dissertation requires you to take a conscious stance with regard to the nature of knowledge, its acquisition and analysis, and the quality and certainty of the conclusions that can be reached from it. You cannot assume that your position will necessarily be shared by your reader, so you will have to make clear to the reader what your philosophical approach is. So, what is it that makes up this philosophical approach, and how can one recognize its nature? The best way is to look at some of the debate about different approaches to knowledge and enquiry - the sorts of issues that are inherent in the process of producing a dissertation. There is a wide spectrum of attitudes in the debate about research, with two opposing camps, the positivists and the relativists, at the extremes, and a range of intermediate stances. I will briefly compare these divergent attitudes. 5.2 Positivism On the one hand, the positivists maintain that in order to know something it should be observable and measurable. The observer must stand apart and take a detached and neutral view of the phenomenon. For example, an engineer testing the strength of a beam will make careful measurements of loading and deflection, and careful observations of cracks and signs of failure. It should not make any difference if the engineer's dog had died that morning and he or she was feeling really low. Any other engineer doing the same test elsewhere would come up with identical results and knowledge about that beam. This is simple enough to understand when dealing with controllable inanimate material and forces. But extreme positivists would go further. They would say that any observable phenomenon can be understood and explained in a logical way, if only enough was known about all the complexities of the situation. Hence, for example, the development of lifeon earth will be fully explained when enough information is gained and enough experiments are carried out to successfully test the theories about the process. Inherent to this way of thinking is a set of assumptions. It is these that underpin the positivist approach and form the basis of scientific method that has brought us so many advances in science and technology. The main assumptions are as follows. Order - There is a conviction that the universe has some kind of order, and because of this, it is possible that we should be able to achieve some kind of understanding of it. Consequently, we can find out the links between events and their causes and thus understand 'the rules of the game'. This then allows us to make predictions. Admittedly, some phenomena are so complicated that it is very difficult to possess enough information and understanding to make reliable predictions (e.g. long-range weather forecasting). External reality - This maintains that everyone shares the same reality and that we do not all live in different worlds that follow different rules. Although there is much philosophical debate about the nature of reality, the positivists rely on the assumption that knowledge is shareable and verifiable: that is, you see the same as I do when, say, looking down a microscope. A theory built upon observations can therefore be tested by any observer to see if it is reliable, in order that the theory can achieve general acceptance. Reliability - Human intellect and perceptions are reliable. You can depend on your senses and methods of thinking. Despite the dangers of deception and muddled thinking, careful observation and logical thought can be depended upon. The accuracy of memory is also an important feature in this assumption. Parsimony - This maintains that the simplest possible explanation is the best. Needless complexity should be avoided. Einstein's formula E = mc2, that sums up his momentous theory of relativity, is a good example of this. Generality - It is no good if the results of one experiment are only relevant to that one case, at that particular time, in that particular place. It must be possible to generalize from particular instances to others, e.g. from the performance of one tested beam to the predicted performance of other similar beams. It is impossible to see every instance of a phenomenon, e.g. water boiling when it is heated, but it is possible to maintain that any water will boil if heated sufficiently. When society is organized in a 'scientific' way, logical methods can be applied to all aspects of life, so as to share all the increasing benefits equally and to ensure that people themselves act rationally. 5.3 Relativism On the other hand, the relativists maintain that we humans are inextricably bound up with the events of the world, and that it is impossible for anyone to stand aside and observe it impartially, as it were, 'from on high'. We are all encumbered by our own experiences and viewpoints, and are enmeshed in our society. However well-established they are, facts are human interpretations of reality, and may well change with time or be understood quite differently by different cultures. This approach is particularly relevant when studying anything to do with human society. Scientific method is poorly equipped to track the inconsistencies, conflicts, and subtleties of beliefs, ideals, and feelings that form such an important part of human life. But even in what is regarded as pure scientific research, for example astronomy, physics and biology, the mindset of society, referred to as the current paradigm, is an enormously powerful force that distorts thinking away from idealistic detachment and channels it into socially (and sometimes religiously) 'acceptable' routes. A striking example of this inthe past was the complicated solutions devised to explain the movement of the heavenly bodies to confirm the belief that the earth was the centre of the universe. Commercial pressure can also distort scientific efforts, for example, in the development of particular drug treatments where heavy investment almost 'requires' that it be shown that the treatment works. In order to understand the basically different approach to understanding reality it is revealing to compare the relativists' attitudes to the previously listed assumptions that underlie the positivists' position. Order - The creation of some kind of order in our understanding of the world is based on our own human perceptions of the world. As time passes, our ordering of the world changes, not because the world has changed but because our attitudes to life, society and beliefs have changed. However much knowledge is gained, we will never reach a definitive understanding of the world order. The 'rules of the game' are constructed by our intellect which is irrevocably bound up in our society and individually. External reality - Our perceptions of the world are uniquely individual. The world we actually perceive does not consist of a series of stimuli that we interpret through our senses and make sense of logically in a void. Rather, we already have a picture of the world, and what we perceive is interpreted in relation to our feelings and understanding - our reality. Admittedly, we might share reality by using the same meanings of words in language (even that is debatable), but this sharing is only a tiny part of our individual experience. We look on the world from within it, and from within ourselves. Reliability - Can we believe our senses? Does our memory always fool us? We will answer 'not always' to the first and 'quite often' to the second, if we believe that human nature is inevitably bound up in its culture and past experiences. These lead us to a personal interpretation of our perception and memorizing of events in our surroundings. Our senses can be tricked in many ways, and our memory is far from perfect, so researchers cannot rely on these to give a definitive record and measurement of the work. However, our skills of reasoning must be taken as a reliable method of organizing data and ideas, even though there may be several ways of interpreting data. Parsimony - Life and society are not so simple and uniform that a simple explanation is possible. Hence simplification usually implies oversimplification. Although needless complexity should be avoided, it is rarely possible to sum any situation up in the form of a neat formula. Generality - Relativists tend to reject the importance, or even possibility, of categorizing individuals and events into classes. Owing to the uniqueness of each person and the uniqueness of each event it is very difficult to predict what may happen in the future under similar conditions; it is dangerous to generalize from studied cases. The function of language becomes an important issue in this debate. Actually, this issue is widened into the subject of communication, which goes beyond just the spoken and written message. We communicate by all sorts of gestures; we assume roles and follow conventions. Consider how difficult it is to gauge people's meanings and feelings when you are in astrange country where different social rules apply, even if you understand the language. This subject of communication is called 'discourse'. How things are communicated is often as important as what is communicated - remember Marshall McLuhan's (1976) phrase, 'the medium is the message'? (McLuhan and Fiore, 1976). Discourse analysis recognizes these important factors and stresses that there is no 'neutral' way of communication. 5.4 How do these attitudes affect your dissertation? You will need to think about this. Your own philosophical approach to how we can see and understand the world around us will be a fundamental factor in your investigation. There are many ways in which any situation can be analysed. Each approach will have a tendency to be based on a particular philosophical line that influences what you look at, the data you collect and the types of conclusions that you aim at. This is best explained using an example. Suppose that you have decided to carry out some research into children's playgrounds in cities. You could base your research on official statistics about how many playgrounds there are in relation to population figures, their sizes, facilities and locations, and records of vandalism and child crime. Or you could arrange interviews with children and parents to find out what they felt about different playgrounds. You could also observe the playgrounds from above and plot children's playing patterns in the form of geometrical shapes; or you could observe from nearby and record how the children used each piece of equipment. You could also make measurements of how much force is exerted by children playing on the equipment, test its strength and measure rates of corrosion, in order to ensure that it is safely built. Each of these approaches involves basic theoretical as well as methodological decisions. Accordingly, you will have to decide which philosophical standpoint(s) to adopt when carrying out your research. This is not to say that any approach is better or more true than any other. Your research approach will depend on the characteristics of your research problem and your own convictions about the nature of research. It is even possible to take different approaches to different aspects of your research topic. Your decisions will help you to determine the nature of your enquiry, the choice of appropriate research methods, and the characteristics of the outcomes that you can expect. 5.5 Chapter Summary Being aware of your philosophical or theoretical standpoint is an important aspect of your research. The assumptions made at the heart of your investigation have a large bearing on how you approach your subject and investigate the aspects of interest. The opposite poles of approach are positivism - the belief that there is an inherent order in the universe which can be discovered if we possess enough information; and relativism - that as human beings within a culture, we can only understand the world from our own perspective, which could be different from that of other humans in other cultures. The concepts of order, external reality, reliability, parsimony and generality are all understood differently in positivism and relativism. In positivism, facts can be established that are immutable so certainties can be achieved, while in relativism, it is understanding that is aimed at, that might be 'true' only in certain situations and cases. References Collier, A. (1964) Critical Realism: An Introduction to Roy Bhaskhar's Philosophy. London: Verso. McLuhan, M. and Fiore, Q. (1976) The Medium is the Message. Harmondsworth: Penguin. [http://www.stanford.edu/dept/CTL/Tomprof/images/newsletter/images/bottom.png] "Desktop faculty development 100 times per year." Over 50,000 subscribers at over 850 institutions in more than 100 countries TOMORROW'S PROFESSORSM eMAIL NEWSLETTER<http://cgi.stanford.edu/%7Edept-ctl/cgi-bin/tomprof/postings.php> Archives of all past postings can be found here<http://cgi.stanford.edu/%7Edept-ctl/cgi-bin/tomprof/postings.php>: Sponsored by Stanford Center for Teaching and Learning<http://ctl.stanford.edu/> Check out the Tomorrow's Professor Blog<http://derekbruff.org/blogs/tomprof/> NOTE: To SUBSCRIBE or UNSUBSCRIBE to the Tomorrows-Professor List click HERE<https://mailman.stanford.edu/mailman/listinfo/tomorrows-professor> [Tomorrow's Professor Facebook Page]<http://www.facebook.com/TomorrowsProfessor> [Tomorrow's Professor Twitter] <http://twitter.com/tomorrowsprof> --++**==--++**==--++**==--++**==--++**==--++**==--++**== tomorrows-professor mailing list tomorrows-professor(a)lists.stanford.edu<mailto:tomorrows-professor@lists.stanford.edu> https://mailman.stanford.edu/mailman/listinfo/tomorrows-professor

1 0

Periodic Table of the Finite Elements poster
by Florian Rathgeber 13 Feb '14

13 Feb '14

This may be of interest to some: Anders, Douglas Arnold and Mattias Schläger are looking for feedback on a poster of a Periodic Table of the Finite Elements they have been working on before it goes to print: https://bitbucket.org/fenics-project/dolfin/wiki/Periodic%20Table%20of%20th… Florian

1 0

This week's FEM lecture is Friday 1030-1200
by David Ham 11 Feb '14

11 Feb '14

Greetings all, with a little more notice than last week, this week's lecture will be 1030-1200 in WPL 3.01. We'll be looking at techniques for nonlinear problems, and working up towards boundary condition implementation. Regards, David -- Dr David Ham Departments of Mathematics and Computing Imperial College London http://www.imperial.ac.uk/people/david.ham

1 0

talk today at 12 on multithreading in games
by Kelly, Paul H J 11 Feb '14

11 Feb '14

Thought you might find this interesting: http://www.doc.ic.ac.uk/cpp/aci_talks/?event=23

1 0

Another snazzy red chair
by Kelly, Paul H J 07 Feb '14

07 Feb '14

We are invited to grab the remaining snazzy red chair from wpl212. It might get thrown out otherwise. It is a two person job - go for it! Paul

1 0