Related
I am currently building a Monte Carlo application in C++ and I have a question regarding parallelization with MPI.
The process I want to parallelize is the MC generation of data. To have good precision in my final results, I specify the goal number of data points. Each data point is generated independently, but might require vastly differing amounts of time.
How do I organize the parallelization and workload distribution of the data generation most efficiently?
What I have done so far
So far I have come up with three possible ways of organizing the MPI part of the code:
The simplest way, but most likely inefficient way: I divide the desired sample size by the number of workers and let every worker generate that amount of data in isolation. However, when the slowest worker finishes, all other workers have been idling for a potentially long time. They could have been "supporting" the slowest worker by sharing its workload.
Use a master: A master communicates with the workers who work continuously until the master process registers that we have enough data and tells everybody to stop what they are doing. The disadvantage I see is that the master process might not be necessary and could be generating data instead (especially when I don't have a lot of workers).
A "ring communication" algorithm I came up with myself: A message is continuously sent and updated in a circle (1->2, 2->3, ... , N ->1). This message contains the global number of generated data point. Once the desired goal is met, the message is tagged, circles one more time and thereby tells everybody to stop working. Important here is I use non-blocking communication (with MPI_Iprobe before receiving via MPI_Recv, and sending via MPI_Isend). This way, everybody works, and no one ever idles.
No matter, which solution is chosen, in the end I reduce all data sets to one big set and continue to process the data.
The concrete questions:
Is there an "optimal" way of parallelizing such a fairly simple process? Would you prefer any of the proposed solutions for some reason?
What do you think of this "ring communication" solution?
I'm sure I'm not the first one to come up with e.g. the ring communication algorithm. I have tried to google this problem, but it seems to me that I do not know the right terminology in this context. I'm sure there must be a lot of material and literature on such simple algorithms, but I never had a formal course on MPI/parallelization. What are the "keywords" to look for?
Any advice and tips are much appreciated.
I have many large Fortran programs to run at work. I have access to several desktop computers and the Fortran code runs over takes several consecutive days. It's essentially running the same master module many times (lets say N times) with different parameters, something akin to Monte Carlo on steroids. In that sense the code is parallelizable, however I don't have access to a cluster.
With the scientific computing community, what practices and strategies are used to minimise hardware damaged from heat? The machines of course have their own cooling system (fans and heat sinks), but even so running intense calculations non stop for half a week cannot be healthy for the life of the machines? Though maybe I'm over-thinking this?
I'm not aware of any intrinsic functions in Fortran that can pause the code to give components a break? Current I've written a small module that keeps an eye on system clock, with a do while loop that "wastes time" in between consecutive runs of the master module in order to discharge heat. Is this an acceptable way of doing this? The processor is, after all, still running a while loop.
Another way would be to use a shell scripts or a python code to import Fortran? Alternatively are there any intrinsic routines in the compile (gfortran) that could achieve this? What are the standard, effective and accepted practices for dealing with this?
Edit: I should mention that all machines run on Linux, specifically Ubuntu 12.04.
For MS-DOS application I would consider the following:
Reduce as much as possible I/O operations withHDD, that is, keep data in memory as much as you can,
or keep data on a RamDisk.A RamDisk driver is available on Microsoft's website.
Let me know if you won't be able to find and I look at my CD archives
-Try to use Extended Memory by using aDPMI driver
DPMI - DOS Protected Mode Interface
-Set CPU affinity for a second CPU
Boost a priority to High, butI wouldn't recommend toboost toReal-Time
I think you need a hardware solution here, not a software solution. You need to increase the rate of heat exchange in the computers (new fans, water cooling, etc) and in the room (turn the thermostat way down, get some fans running, etc).
To answer the post more directly, you can use the fortran SLEEP command to pause a computation for a given number of seconds. You could use some system calls in Fortran to set the argument on the fly. But I wouldn't recommend it - you might as well just run your simulations on fewer computers.
To keep the advantages of the multiple computers, you need better heat exchange.
As long as the hardware is adequately dissipating heat and components are not operating at or beyond their "safe" temperature limits, they * should be fine.
*Some video cards were known to run very hot; i.e. 65-105°C. Typically, electronic components have a maximum temperature rating of exactly this. Beyond it, reliability degrades very quickly. Even though the manufacturer made these cards this way, they ended up with a reputation for failing (i.e. older nVidia FX, Quadro series.)
*Ubuntu likely has a "Critical temperature reached" feature where the entire system will power off if it overheats, as explained here. Windows is "blissfully ignorant." :)
*Thermal stress (large, repeated temperature variations) may contribute to component failure of IC's, capacitors, and hard disks. Over three decades of computing has taught that adequate cooling and leaving the PC on 24/7 actually may save wear-and-tear in my experience. (A typical PC will cost around $200 USD/year in electricity, so it's more like a trade-off in terms of cost.)
*PC's must be cleaned twice a year (depending on airborne particulate constituency and concentration.) Compressed air is nice for removing dust. Dust traps heat and causes failures. Operate a shop-vac while "dusting" to prevent the dust from going everywhere. Wanna see a really dusty computer?
*The CPU should be "ok" with it's stock cooler. Check it's temperature at cold system boot-up, then again after running code for an hour or so. The fan is speed-controlled to limit temperature rise. CPU temperature rise shouldn't be much warmer than about 40°C and less would be better. But an aftermarket, better-performing CPU cooler never hurts, such as these. CPU's rarely fail unless there is a manufacturing flaw or they operate near or beyond their rated temperatures for too long, so as long as they stay cool, long calculations are fine. Typically, they stop functioning and/or reset the PC if too hot.
*Capacitors tend to fail very rapidly when overheated. It is a known issue that some cap vendors are "junk" and will fail prematurely, regardless of other factors. "Re-capping" is the art of fixing these components. For a full run-down on this topic, see badcaps.net. It used to be possible to re-cap a motherboard, but today's 12+ layer and ROHS (no lead) motherboards make it very difficult without specialty hot-air tools.
I was wondering if you could theoretically utilise more than one core of of a processor to complete tasks quicker by using inter process communication instead of multithreading.
Say for instance a game engine. You have one executable processing ai and physics, then another handling the sound and rendering.
Maybe there could be shared memory that the physics and ai results get written to, that the renderer then could use to output the graphics.
What do you think? Ridiculous idea or feasible?
Thank you for your time.
Edit: the engine does not exist, it's just an example. Basically I'm asking if two or more programs can work together if any tasks can be parallelized.
Basically I'm asking if two or more programs can work together if any tasks can be parallelized.
Yes, the idea isn't flawed in principle.
That being said, creating a thread and creating a process vary on different OSes in terms of overhead; Windows processes are very expensive to create, while Linux ones are pretty much free.
One advantage of separating the processes is that when one of them crashes, the others aren't impacted and it can be restarted. You can also more easily push the calculations to another machine across the network. However, synchronizing threads will probably be easier and more performant in most cases.
It was hard for me to come up with a real-world example for a concurrency:
Imagine the above situation, where
there are many lanes, many junctions
and a great amount of cars. Besides,
there is a human factor.
The problem is a hard research area for traffic engineers. When I investigated it a some time ago, I noticed that many models failed on it. When people are talking about functional programming, the above problem tends to pop up to my mind.
Can you simulate it in Haskell? Is Haskell really so concurrent? What are the limits to parallelise such concurrent events in Haskell?
I'm not sure what the question is exactly. Haskell 98 doesn't specify anything for concurrency. Specific implementations, like GHC, provide extensions that implement parallelism and concurrency.
To simulate traffic, it would depend on what you needed out of the simulation, e.g. if you wanted to track individual cars or do it in a general statistical way, whether you wanted to use ticks or a continuous model for time, etc. From there, you could come up with a representation of your data that lent itself to parallel or concurrent evaluation.
GHC provides several methods to leverage multiple hardware execution units, ranging from traditional semaphores and mutexes, to channels with lightweight threads (which could be used to implement an actor model like Erlang), to software transactional memory, to pure functional parallel expression evaluation, with strategies, and experimental nested data parallelism.
So yes, Haskell has many approaches to parallel execution that could certainly be used in traffic simulations, but you need to have a clear idea of what you're trying to do before you can choose the best digital representation for your concurrent simulation. Each approach has its own advantages and limits, including learning curve. You may even learn that concurrency is overkill for the scale of your simulations.
It sounds to me like you are trying to do a simulation, rather than real-world concurrency. This kind of thing is usually tackled using discrete event simulation. I did something similar in Haskell a few years ago, and rolled my own discrete event simulation library based on the continuation monad transformer. I'm afraid its owned by my employer, so I can't post it, but it wasn't too difficult. A continuation is effectively a suspended thread, so define something like this (from memory):
type Sim r a = ContT r (StateT ThreadQueue IO a)
newtype ThreadQueue = TQ [() -> Sim r ()]
The ThreadQueue inside the state holds the queue of currently scheduled threads. You can also have other types of thread queue to hold threads that are not scheduled, for instance in a semaphore (based on "IORef (Int, ThreadQueue)"). Once you have semaphores you can build the equivalent of MVars and MQueues.
To schedule a thread use "callCC". The argument to "callCC" is a function "f1" that itself takes a function "c" as an argument. This inner argument "c" is the continuation: calling it resumes the thread. When you do this, from that thread's point of view "callCC" just returned the value you gave as an argument to "c". In practice you don't need to pass values back to the suspended threads, so the parameter type is null.
So your argument to "callCC" is a lambda function that takes "c" and puts it on the end of whatever queue is appropriate for the action you are doing. Then it takes the head of the ThreadQueue from inside the state and calls that. You don't need to worry about this function returning: it never does.
If you need a concurrent programming language with a functional sequential subset, consider Erlang.
More about Erlang
I imagine you're asking if you could have one thread for each object in the system?
The GHC runtime scales nicely to millions of threads, and multiplexes those threads onto the available hardware, via the abstractions Chris Smith mentioned. So it certainly is possible to have thousands of threads in your system, if you're using Haskell/GHC.
Performance-wise, it tends to be a good deal faster than Erlang, but places less emphasis on distribution of processes across multiple nodes. GHC in particular, is more targetted towards fast concurrency on shared memory multicore systems.
Erlang, Scala, Clojure are languages that might suit you.
But I think what you need more is to find a suitable Multi-Agents simulation library or toolkit, with bindings to your favourite language.
I can tell you about MASON, Swarm and Repast. But these are Java and C libaries...
I've done one answer on this, but now I'd like to add another from a broader perspective.
It sounds like the thing that make this a hard problem is that each driver is basing their actions on mental predictions of what other drivers are going to do. For instance when I am driving I can tell when a car is likely to pull in front of me, even before he indicates, based on the way he is lining himself up with the gap between me and the car in front. He in turn can tell that I have seen him from the fact that I'm backing off to make room for him, so its OK to pull in. A good driver picks up lots of these subtle clues, and its very hard to model.
So the first step is to find out what aspects of real driving are not included in the failed models, and work out how to put them in.
(Clue: all models are wrong, but some models are useful).
I suspect that the answer is going to involve giving each simulated driver one or more mental models of what each other driver is going to do. This involves running the planning algorithm for Driver 2 using several different assumptions that Driver 1 might make about the intentions of Driver 2. Meanwhile Driver 2 is doing the same about Driver 1.
This is the kind of thing that can be very difficult to add to an existing simulator, especially if it was written in a conventional language, because the planning algorithm may well have side effects, even if its only in the way it traverses a data structure. But a functional language may well be able to do better.
Also, the interdependence between drivers probably means there is a fixpoint somewhere in there, which lazy languages tend to do better with.
I am quite excited by the possibility of using languages which have parallelism / concurrency built in, such as stackless python and erlang, and have a firm belief that we'll all have to move in that direction before too long - or will want to because it will be a good/easy way to get to scalability and performance.
However, I am so used to thinking about solutions in a linear/serial/OOP/functional way that I am struggling to cast any of my domain problems in a way that merits using concurrency. I suspect I just need to unlearn a lot, but I thought I would ask the following:
Have you implemented anything reasonably large in stackless or erlang or other?
Why was it a good choice? Was it a good choice? Would you do it again?
What characteristics of your problem meant that concurrent/parallel was right?
Did you re-cast an exising problem to take advantage of concurrency/parallelism? and
if so, how?
Anyone any experience they are willing to share?
in the past when desktop machines had a single CPU, parallelization only applied to "special" parallel hardware. But these days desktops have usually from 2 to 8 cores, so now the parallel hardware is the standard. That's a big difference and therefore it is not just about which problems suggest parallelism, but also how to apply parallelism to a wider set of problems than before.
In order to be take advantage of parallelism, you usually need to recast your problem in some ways. Parallelism changes the playground in many ways:
You get the data coherence and locking problems. So you need to try to organize your problem so that you have semi-independent data structures which can be handled by different threads, processes and computation nodes.
Parallelism can also introduce nondeterminism into your computation, if the relative order in which the parallel components do their jobs affects the results. You may need to protect against that, and define a parallel version of your algorithm which is robust against different scheduling orders.
When you transcend intra-motherboard parallelism and get into networked / cluster / grid computing, you also get the issues of network bandwidth, network going down, and the proper management of failing computational nodes. You may need to modify your problem so that it becomes easier to handle the situations where part of the computation gets lost when a network node goes down.
Before we had operating systems people building applications would sit down and discuss things like:
how will we store data on disks
what file system structure will we use
what hardware will our application work with
etc, etc
Operating systems emerged from collections of 'developer libraries'.
The beauty of an operating system is that your UNWRITTEN software has certain characteristics, it can:
talk to permanent storage
talk to the network
run in a command line
be used in batch
talk to a GUI
etc, etc
Once you have shifted to an operating system - you don't go back to the status quo ante...
Erlang/OTP (ie not Erlang) is an application system - it runs on two or more computers.
The beauty of an APPLICATION SYSTEM is that your UNWRITTEN software has certain characteristics, it can:
fail over between two machines
work in a cluster
etc, etc...
Guess what, once you have shifted to an Application System - you don't go back neither...
You don't have to use Erlang/OTP, Google have a good Application System in their app engine, so don't get hung up about the language syntax.
There may well be good business reasons to build on the Erlang/OTP stack not the Google App Engine - the biz dev guys in your firm will make that call for you.
The problems will stay almost the same inf future, but the underlying hardware for the realization is changing. To use this, the way of compunication between objects (components, processes, services, how ever you call it) will change. Messages will be sent asynchronously without waiting for a direct response. Instead after a job is done the process will call the sender back with the answer. It's like people working together.
I'm currently designing a lightweighted event-driven architecture based on Erlang/OTP. It's called Tideland EAS. I'm describing the ideas and principles here: http://code.google.com/p/tideland-eas/wiki/IdeasAndPrinciples. It's not ready, but maybe you'll understand what I mean.
mue
Erlang makes you think of the problem in parallel. You won't forget it one second. After a while you adapt. Not a big problem. Except the solution become parallel in every little corner. All other languages you have to tweak. To be concurrent. And that doesn't feel natural. Then you end up hating your solution. Not fun.
The biggest advantages Erlang have is that it got no global garbage collect. It will never take a break. That is kind of important, when you have 10000 page views a second.