Related
Hey so I have a question about FPGA's. If you look at the current lineup of xilinx products, specifically the 7 series, there is a massive price differential between each of the models. What I don't understand is if I could buy an Artix-7 with ~200k logic cells for $300 whereas a Virtex-7 with ~2000k logic cells costs in excess of $20,000. So could I just buy 10 Artix-7's and get the same performance? Furthermore is performance linearly related to the number of logic cells, and if not then how are they related? Is there any advantage to having more logic cells per core? I'm sure it depends on what you are doing but as my interest in the matter, although theoretical, lies in cryptograpic applications, my question relates specifically to implementations of the MD5, SHA-0/1/2/3, and similar encryption algorithms.
An FPGA doesn't have "performance" like a processor. It just has a bunch of logic elements (LEs) that you can use. If a high-end part has 2MLEs and a low-end part has 200kLEs, but you only need 20kLEs for your processing core, it makes little difference which one you use, all else being equal. Of course, if you have a problem that can easily be parallelized, then you can turn those extra LEs into extra performance by building more processing cores. But that's up to you to do.
Now, all else is not always equal, because there's a lot more to an FPGA than simply the number of logic cells. I can't speak for Xilinx parts (I work for another major FPGA vendor) but typically the high-end families will have things like very high-speed transceivers that the midrange and low-end families do not. In addition, sometimes they have different mixes of embedded RAM, DSP, etc.
So, can you use a bunch of small FPGAs instead of a large one? Remember that an FPGA will only have about 1000-2000 IOs, whereas there will be more like 100Ks of internal wires between the corresponding parts of the higher-end part. So not only will you have to build a pretty complicated board, you might find yourself IO-limited in getting signals off of one chip and onto another.
I'm a CS master student, and next semester I will have to start working on my thesis. I've had trouble coming up with a thesis idea, but I decided it will be related to Computer Graphics as I'm passionate about game development and wish to work as a professional game programmer one day.
Unfortunately I'm kinda new to the field of 3D Computer Graphics, I took an undergraduate course on the subject and hope to take an advanced course next semester, and I'm already reading a variety of books and articles to learn more. Still, my supervisor thinks its better if I come up with a general thesis idea now and then spend time learning about it in preparation for doing my thesis proposal. My supervisor has supplied me with some good ideas but I'd rather do something more interesting on my own, which hopefully has to do with games and gives me more opportunities to learn more about the field. I don't care if it's already been done, for me the thesis is more of an opportunity to learn about things in depth and to do substantial work on my own.
I don't know much about GPU programming and I'm still learning about shaders and languages like CUDA. One idea I had is to program an entire game (or as much as possible) on the GPU, including all the game logic, AI, and tests. This is inspired by reading papers on GPGPU and questions like this one I don't know how feasible that is with my knowledge, and my supervisor doesn't know a lot about recent GPUs. I'm sure with time I will be able to answer this question on my own, but it'd be handy if I could know the answer in advance so I could also consider other ideas.
So, if you've got this far, my question: Using only shaders or something like CUDA, can you make a full, simple 3D game that exploits the raw power and parallelism of GPUs? Or am I missing some limitation or difference between GPUs and CPUs that will always make a large portion of my code bound to CPU? I've read about physics engines running on the GPU, so why not everything else?
DISCLAIMER: I've done a PhD, but have never supervised a student of my own, so take all of what I'm about to say with a grain of salt!
I think trying to force as much of a game as possible onto a GPU is a great way to start off your project, but eventually the point of your work should be: "There's this thing that's an important part of many games, but in it's present state doesn't fit well on a GPU: here is how I modified it so it would fit well".
For instance, fortran mentioned that AI algorithms are a problem because they tend to rely on recursion. True, but, this is not necessarily a deal-breaker: the art of converting recursive algorithms into an iterative form is looked upon favorably by the academic community, and would form a nice center-piece for your thesis.
However, as a masters student, you haven't got much time so you would really need to identify the kernel of interest very quickly. I would not bother trying to get the whole game to actually fit onto the GPU as part of the outcome of your masters: I would treat it as an exercise just to see which part won't fit, and then focus on that part alone.
But be careful with your choice of supervisor. If your supervisor doesn't have any relevant experience, you should pick someone else who does.
I'm still waiting for a Gameboy Emulator that runs entirely on the GPU, which is just fed the game ROM itself and current user input and results in a texture displaying the game - maybe a second texture for sound output :)
The main problem is that you can't access persistent storage, user input or audio output from a GPU. These parts have to be on the CPU, by definition (even though cards with HDMI have audio output, but I think you can't control it from the GPU). Apart from that, you can already push large parts of the game code into the GPU, but I think it's not enough for a 3D game, since someone has to feed the 3D data into the GPU and tell it which shaders should apply to which part. You can't really randomly access data on the GPU or run arbitrary code, someone has to do the setup.
Some time ago, you would just setup a texture with the source data, a render target for the result data, and a pixel shader that would do the transformation. Then you rendered a quad with the shader to the render target, which would perform the calculations, and then read the texture back (or use it for further rendering). Today, things have been made simpler by the fourth and fifth generation of shaders (Shader Model 4.0 and whatever is in DirectX 11), so you can have larger shaders and access memory more easily. But still they have to be setup from the outside, and I don't know how things are today regarding keeping data between frames. In worst case, the CPU has to read back from the GPU and push again to retain game data, which is always a slow thing to do. But if you can really get to a point where a single generic setup/rendering cycle would be sufficient for your game to run, you could say that the game runs on the GPU. The code would be quite different from normal game code, though. Most of the performance of GPUs comes from the fact that they execute the same program in hundreds or even thousands of parallel shading units, and you can't just write a shader that can draw an image to a certain position. A pixel shader always runs, by definition, on one pixel, and the other shaders can do things on arbitrary coordinates, but they don't deal with pixels. It won't be easy, I guess.
I'd suggest just trying out the points I said. The most important is retaining state between frames, in my opinion, because if you can't retain all data, all is impossible.
First, Im not a computer engineer so my assumptions cannot even be a grain of salt, maybe nano scale.
Artificial intelligence? No problem.There are countless neural network examples running in parallel in google. Example: http://www.heatonresearch.com/encog
Pathfinding? You just try some parallel pathfinding algorithms that are already on internet. Just one of them: https://graphics.tudelft.nl/Publications-new/2012/BB12a/BB12a.pdf
Drawing? Use interoperability of dx or gl with cuda or cl so drawing doesnt cross pci-e lane. Can even do raytracing at corners so no z-fighting anymore, even going pure raytraced screen is doable with mainstream gpu using a low depth limit.
Physics? The easiest part, just iterate a simple Euler or Verlet integration and frequently stability checks if order of error is big.
Map/terrain generation? You just need a Mersenne-twister and a triangulator.
Save game? Sure, you can compress the data parallelly before writing to a buffer. Then a scheduler writes that data piece by piece to HDD through DMA so no lag.
Recursion? Write your own stack algorithm using main vram, not local memory so other kernels can run in wavefronts and GPU occupation is better.
Too much integer needed? You can cast to a float then do 50-100 calcs using all cores then cast the result back to integer.
Too much branching? Compute both cases if they are simple, so every core is in line and finish in sync. If not, then you can just put a branch predictor of yourself so the next time, it predicts better than the hardware(could it be?) with your own genuine algorithm.
Too much memory needed? You can add another GPU to system and open DMA channel or a CF/SLI for faster communication.
Hardest part in my opinion is the object oriented design since it is very weird and hardware dependent to build pseudo objects in gpu. Objects should be represented in host(cpu) memory but they must be separated over many arrays in gpu to be efficient. Example objects in host memory: orc1xy_orc2xy_orc3xy. Example objects in gpu memory: orc1_x__orc2_x__ ... orc1_y__orc2_y__ ...
The answer has already been chosen 6 years ago but for those interested to the actual question, Shadertoy, a live-coding WebGL platform, recently added the "multipass" feature allowing preservation of state.
Here's a live demo of the Bricks game running on Gpu.
I don't care if it's already been
done, for me the thesis is more of an
opportunity to learn about things in
depth and to do substantial work on my
own.
Then your idea of what a thesis is is completely wrong. A thesis must be an original research. --> edit: I was thinking about a PhD thesis, not a master thesis ^_^
About your question, the GPU's instruction sets and capabilities are very specific to vector floating point operations. The game logic usually does little floating point, and much logic (branches and decision trees).
If you take a look to the CUDA wikipedia page you will see:
It uses a recursion-free,
function-pointer-free subset of the C
language
So forget about implementing any AI algorithms there, that are essentially recursive (like A* for pathfinding). Maybe you could simulate the recursion with stacks, but if it's not allowed explicitly it should be for a reason. Not having function pointers also limits somewhat the ability to use dispatch tables for handling the different actions depending on state of the game (you could use again chained if-else constructions, but something smells bad there).
Those limitations in the language reflect that the underlying HW is mostly thought to do streaming processing tasks. Of course there are workarounds (stacks, chained if-else), and you could theoretically implement almost any algorithm there, but they will probably make the performance suck a lot.
The other point is about handling the IO, as already mentioned up there, this is a task for the main CPU (because it is the one that executes the OS).
It is viable to do a masters thesis on a subject and with tools that you are, when you begin, unfamiliar. However, its a big chance to take!
Of course a masters thesis should be fun. But ultimately, its imperative that you pass with distinction and that might mean tackling a difficult subject that you have already mastered.
Equally important is your supervisor. Its imperative that you tackle some problem they show an interest in - that they are themselves familiar with - so that they can become interested in helping you get a great grade.
You've had lots of hobby time for scratching itches, you'll have lots more hobby time in the future too no doubt. But master thesis time is not the time for hobbies unfortunately.
Whilst GPUs today have got some immense computational power, they are, regardless of things like CUDA and OpenCL limited to a restricted set of uses, whereas the CPU is more suited towards computing general things, with extensions like SSE to speed up specific common tasks. If I'm not mistaken, some GPUs have the inability to do a division of two floating point integers in hardware. Certainly things have improved greatly compared to 5 years ago.
It'd be impossible to develop a game to run entirely in a GPU - it would need the CPU at some stage to execute something, however making a GPU perform more than just the graphics (and physics even) of a game would certainly be interesting, with the catch that game developers for PC have the biggest issue of having to contend with a variety of machine specification, and thus have to restrict themselves to incorporating backwards compatibility, complicating things. The architecture of a system will be a crucial issue - for example the Playstation 3 has the ability to do multi gigabytes a second of throughput between the CPU and RAM, GPU and Video RAM, however the CPU accessing GPU memory peaks out just past 12MiB/s.
The approach you may be looking for is called "GPGPU" for "General Purpose GPU". Good starting points may be:
http://en.wikipedia.org/wiki/GPGPU
http://gpgpu.org/
Rumors about spectacular successes in this approach have been around for a few years now, but I suspect that this will become everyday practice in a few years (unless CPU architectures change a lot, and make it obsolete).
The key here is parallelism: if you have a problem where you need a large number of parallel processing units. Thus, maybe neural networks or genetic algorithms may be a good range of problems to attack with the power of a GPU. Maybe also looking for vulnerabilities in cryptographic hashes (cracking the DES on a GPU would make a nice thesis, I imagine :)). But problems requiring high-speed serial processing don't seem so much suited for the GPU. So emulating a GameBoy may be out of scope. (But emulating a cluster of low-power machines might be considered.)
I would think a project dealing with a game architecture that targets multiple core CPUs and GPUs would be interesting. I think this is still an area where a lot of work is being done. In order to take advantage of current and future computer hardware, new game architectures are going to be needed. I went to GDC 2008 and there were ome talks related to this. Gamebryo had an interesting approach where they create threads for processing computations. You can designate the number of cores you want to use so that if you don't starve out other libraries that might be multi-core. I imagine the computations could be targeted to GPUs as well.
Other approaches included targeting different systems for different cores so that computations could be done in parallel. For instance, the first split a talk suggested was to put the renderer on its own core and the rest of the game on another. There are other more complex techniques but it all basically boils down to how do you get the data around to the different cores.
I got into a bit of a debate yesterday with my boss about the proper role of optimization when building software. Essentially, his position was that optimization needs to be a primary concern during the entire process of development.
My opinion is that you need to make the right algorithmic decisions during development, but you should never be counting cycles during development. In fact, I feel so strongly about this I had to walk away from the conversation. I've seen too many bad programming decisions in the name of "optimization", and too much bad code defended with the excuse "this way is faster".
What does the StackOverflow.com community think?
"We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil. Yet we should not pass up our opportunities in that critical 3%."
- Donald Knuth
I think the premature optimization quote is used by too many to avoid thinking about the hard stuff concerning how well the application will run. I guarantee the users want you to think about how to design it so it will run as fast as possible.
This is not to say you should be timing everything, but the design phase is the easiest place to optimize and not cost lots of time later.
There are often several ways to do anything, you should pick in the design phase the one which is most likely to perform the best (if it turns out to be one of the times when it isn't the best, then optimize later). This should trump the need to have easy to read code.
If you aren't considering performance in the design phase, you aren't going to have a well designed system. That doesn't mean it should be the only concern (although in a database I'd rate it as 3rd in importance, right after data integrity and security), but trying to fix a system where poorly performing techniques were used throughout because the developers thought they were easier to understand is a nightmare. Being a user of such a system where you have to wait for minutes everytime you want to move from one screen to another is a nightmare (developers reallly should spend all day everyday for at least a week, using their systems!) for everyone who is stuck with the badly designed system. It costs less to design properly than to fix later and considering performance is critical to designing properly.
I work somewhere where the orginal developers drank the koolaid about premature optimization and did everything the way they thought was simplest (but which in almost every case was the wrong choice from a performance perspective). Now we are at 10 times the size we were three years ago and every screen on every website takes 30 seconds or so to load (or worse times out) and we are losing customers because of it. But changing it will be too hard because at the base they designed the database without considering how it would perform and redesigning a database with many many gigabytes of data into a new structure is way too time consuming and costly. If it had been designed to perform from the start it would be both easier to maintain and faster for the clients. We aren't talking about the need to performance tune the top 10 slowest queries here, we are talking about the fact that the overall structure requires a drastic change (one that would affect virtually every query against the system) to perform well.
Yes don't do micro optimization until you nmeed to, but please do the macro stuff. Consider if is this the best way before you commit to the path. Don't write cursors to hit tables with millions of records when a set-based statement will do. Don't try to have as few tables as possible becasue that seems to be a more elegant solution when the tables are storing disparate items (such as people, places, and vehicles) causing every single query to hit the same table and causing every delete to check all sorts of foreign key tables that will not ever have a record for that type of entity (it takes minutes to delete one record from the main table in our database, it's a real joy when something goes wrong in an import (bad data from a client usually) and we have to delete 200,000 let me tell you).
Optimization is almost tautologically a tradeoff: you gain runtime efficiency at the cost of other things (readability, maintainability, flexibility, compile times, etc.). As such, it's really never a good idea to do unless you know what the tradeoff is and why it's worthwhile.
Even worse, thinking about "how do I do X fast" can be very distracting. To much energy in that direction can easily lead to you misisng out on method Y which is much better (and often faster --- this is how "optimization" can make your code slower). Particularly if you do too much of this on a big project from the beginning, it represents a lot of momentum. If you can't afford to overcome that momentum, you can easily become locked into a bad design because you can't afford the time to restructure it. This way lie dragons
What your boss may be thinking of is more of an issue of writing bad code via inappropriate representations and algorithms. It's not really the same thing as optimizing, but an approach where you pay no attention whatsoever to appropriate data structures etc. can result in a codebase that is slow everywhere, and (much like the above "lock in") requires heroic effort to fix.
In general though, premature optimization really honestly is a terrible idea. Particularly when you end up with a complex, finely tuned, well documented (because that's the only way you can understand it) piece of code you end up not using anyway. And that's not even getting into the issue of subtle bugs that are often introduced when "optimizing"
[edit: pshaw, of course a Knuth quote encapsulates this well. That's what I get for typing too much]
Engineer throughout, optimize at the end.
Since with going with pithy, I'll say that optimization is as important as the impact of not doing it.
I think the "premature optimization is root of all evil" has to be understood literally - it does not say when is premature, and does not say you should optimize only at the end. Just not too early. Also, the "use the right algorithm, O(n^2) vs O(N)" is a bit dangerous if taken literally - because for many problems, the N is actually small, etc...
I think it depends a lot of the type of software you are doing: some software are such as every part is very independent, and can be optimized separately. But that's not always the case. For many (most ?) applications, speed just does not matter at all, the brute force but obviously correct way is the best one. But for projects where speed matters, it often has to be taken into account early - maybe that's another possible interpretation of Knuth's saying: many applications don't need to be optimized at all, just know which ones need and plan ahead.
Optimization is a primary concern through development when you have a good reason to expect that performance will be unfixably bad if optimization is a secondary concern.
This depends a lot what kind of code you're writing, but there are often better reasons to believe that your code will be unfixably difficult to use; or maintain; or full of bugs; or late; if all those things become secondary to tweaking performance.
Bad managers say, "all of those things are our primary concerns". Good managers work to find out which are dangers for this project.
Of course, good design does have to consider all these things, and the earlier you have a back-of-the-envelope estimate of any of them, the better. If all your manager is saying, is that if you never think about how fast your code will run then too much of your code will be dog-slow, then he's right. I just wouldn't say that makes optimization your "primary" concern.
If the USP of your software is that it's faster than your competitors', then optimization is a primary concern. With experience, you can often predict what sorts of operations will be the bottlenecks, design those with optimization in mind right from the start, and more-or-less ignore optimization elsewhere. A lot of projects won't even need this: they'll be fast enough without much effort, provided you use sensible algorithms and don't do anything stupid. "Don't do anything stupid" is always a primary concern, with no need to mention performance in particular.
I think that code needs to be, first and foremost, readable and understandable. So, optimisations that are done, should not be at the expense of readability. However, optimisation is often a trade-off.
Whether or not you should optimise your code depends on your application domain. If you are working on an embedded processor with only 8Mb of memory, then optimisation is probably something that every team member needs to keep in mind, when writing code - optimising for space vs speed.
However, pre-mature optimisation is not useful unless your system has been clearly spec'ed and understood. This is because most programmers do not make good optimisation decisions unless they can factor in the influence of the overall system, including processor architectural factors such as cache memory, hardware threads, pipelines, etc.
From 2 years of building highly optimized Java code (and that needed to be optimized that way) I would say that there is a time-spent rule that governs optimization:
optimizing on the spot: 5%-10% of your development time, because you have to do it countless times (every single time you have to amend your design)
optimizing just when you have had it working: 2% of your development time (you do it only once)
going back to it and optimizing when it's too slow: 30% of your development time, because you have to plunge yourself back into the system
SO I would come to the conclusion that there is a right time and a right way to optimize: do it entity by entity (class by class, if you have classes that have a single, well defined job to do, and can be tested and optimized), test well, make sure the logic is working, optimize just afterward, and forget about that entity's implementation details forever.
When developing, just keep it simple. IMHO, most performance problems are caused by over-engineering - making mountains out of molehills, often because of wanting "the right algorithm".
Periodically, stress test with a big data set, profiling or (my favorite technique) manual random sampling. You find a problem, you fix it. You find another, you fix it.
That way you avoid creating slugs (slowness bugs), and when they do arise, you kill them.
Added: If I can just elaborate on point 1. OO is seemingly the law of the land, and it certainly has good reasons behind it. Unfortunately it causes many young programmers to feel that programming is all about having lots of data structure, with layers upon layers of abstraction. Not that those are inherently bad, but combine that with the natural tendency to assume that the time something takes is roughly proportional to the number of characters you have to type to invoke it, and that this tendency multiplies over the layers (and besides, the machines are really fast), it's easy to create a perfect storm of cycle-waste.
Quote from a friend: "It's easier to make a working system efficient than to make an efficient system work".
I think it is important to use smart practices and patterns from the start, but get the system actually running for small test cases then do performance analysis. Frequently the areas of code that have poor performance aren't anticipated at the beginning, so get some real data and then optimize the bottlenecking 3% (or 20%, or whatever it is).
I think your boss is a lot more right than you are.
Allt too often the user experience is lost only to be "rediscovered" at the last possible moments when performance activites are prohibitively costly and inefficient. Or when it is discovered that the batch program that will process today's transactions requires forty hours to run.
Things such as database organization, when and when not to do which SELECTs are examples of design decisions that can make or break an application. Still you run the risk of a single programmer deciding to to otherwise, misinterpret or simply not understand what to do. Following-up on performance during an entire project decreases the risk that such things will happen. It also allows design decisions to be changed when there is a need for that without puting the entire poroject at risk.
"You need to make the right algorithmic decisions during development" is certainly true yet how many mainstream programmers are able to do that? Browsing the net for information does not guarantee finding a high quality solution. "Right" could be interpreted as meaning that it is ok to choose a poor algorithm because it is easy to understand and implement (= less development time, lower cost) than a more complicated one (= more development time, higher cost).
The pendulum of quantity vs quality is almost always on the quantity side because more code per hour or faster development time means money in the short term. The quality side means money in the long term.
EDIT
This article discusses performance and optimization quite thoroughly.
Performance preacher Rico Mariana sums it up in the short statement "never give up your performance accidentally."
Premature optimization is the root of all evil...There is a fine balance between, but I would say 95% of the time you need to optimize at the end; however, there are decisions you can make early on to help prevent issues. For example assume we are talking about an e-commerce web site. You have a requirement to display the catalog. Now you can grab all 100,000 items and display 50 of them, or you can grab just 50 from the database. These type of decisions should be made up front.
Cycle counting should only be done when a problem has been identified.
Your boss is partly right, optimisation does need to be considered throughout the development lifecycle but it is rarely the primary concern. Also, the term 'optimisation' is vague - it's an adjective, 'optimise for ...' which could be 'memory', 'speed', 'usability', 'maintainability' and so on.
However, the OP is right that counting cycles is pointless for many projects. For most PC applications the CPU is never the bottleneck. Also, the IA32 is not consistent - what worked well on one architecture, performs poorly on another. Cycle counting should only ever be done when it will actually make a difference - usually in CPU limited code or code with very specific timing needs.
Optimisation, of any kind, must always be driven by hard evidence. Never assume anything about the system or how the code is behaving. In an ideal world, application performance / constraints will be specified in the initial product design and tools to monitor the application's performance during development will be added early on in the development phase to guide the programmers as the product is made.
We should develop on slow boxen because it forces us to optimize early.
Randall Hyde points out in The Fallacy of Premature Optimization, there are plenty of misconceptions around the Hoare quote:
We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil.
In particular, even though machines these days scream compared with those in Hoare's day, it doesn't mean "optimization should be avoided." So does my respected colleague have a point when he suggests that we should develop on boxes of modest tempo? The idea is that performance bottlenecks are more irritating on a slow box and so they are likely to receive attention.
This should be community wiki since it's pretty subjective and there's no "right" answer.
That said, you should develop on the fastest machine available to you. Yes, anything slower will introduce irritation and encourage you to fix the slowdowns, but only at a very high price:
Your productivity as a programmer is directly related to the number of things you can hold in your head, and anything which slows down your process or impedes you at all lengthens the amount of time you have to hold those ideas in short term-memory, making you more likely to forget them, and have to go re-learn them.
Waiting for a program to compile allows the stack of bugs, potential issues, and fixes to drop out of your head as you get distracted. Waiting for a dialog to load, or a query to finish interrupts you similarly.
Even if you ignore that effect, you've still got the truth of the later statement - early optimization will leave you chasing yourself round in circles, breaking code that already works, and guessing (with often poor accuracy) about where things might get bogged down. Design your code properly in the first place, and you can forget about optimization until it's had a chance to settle for a bit, at which point any necessary optimization will be obvious.
Slow computers are not going to help you find your performance problems.
If your test data is only a few hundred rows in a table your db will cache it all and you'll never find badly written queries or bad table/index design. If your server application is not multi-threaded server you will not find that out until you stress test it with 500 users. Or if the app bottlenecks on bandwidth.
Optimization is "A Good Thing" but as I say to new developers who have all sorts of ideas about how to do it better 'I don't care how quickly you give me the wrong answer'. Get it right first, then make it faster when you find a bottleneck. An experienced programmer is going to design and build it reasonably well to start with.
If performance is really critical (real time? millisecond-transactions?) then you need to design and implement a set of benchmarks and tools to scientifically prove to yourselves that your changes are making it faster. There are way too many variables out there that affect performance.
Plus there's the classic programmer excuse they will bring out - 'but it's running slow because we have deliberately picked slow computers, it will run much faster when we deploy it.'
If your colleague thinks its important give him a slow computer and put him in charge of 'performance' :-)
I guess it would depend on what you're making and what the intended audience is.
If you're writing software for fixed hardware (say, console games) then use equipment (at least test equipment) that is similar or the same as what you will deploy on.
If you're developing desktop apps or something in that realm then develop on whatever machine you want and then tune it afterward to run on the desired min-spec hardware. Likewise, if you're developing in-house software, there is likely to be a min-spec for the machines that the company wants to buy. In that case, develop on a fast machine (to decrease development time and therefore costs) and test against that min-spec.
Bottom line, develop on the fastest machine you can get your hands on, and test on the minimum or exact hardware that you'll be supporting.
If you are programming on hardware that is close to the final test and production environments, you tend to find that there are less nasty surprises when it comes time to release the code.
I've seen enough programmers get side-swiped by serious, but unexpected problems caused by their machines being way faster than their most of their users. But also, I've seen the same problem occur with data. The code is tested on a small dataset and then "crumbles" on a large one.
Any differences in development and deployment environments can be the source of unexpected problems.
Still, since programming is expensive and time-consuming, if the end-user is running slow out-of-date equipment, the better solution is to deal with it at testing time (and schedule in a few early tests just to check usability and timing).
Why cripple your programmers just because you're worried about missing a potential problem? That's not a sane development strategy.
Paul.
for the love of Codd, use profiling tools, not slow development machines!
Optimization should be avoided, didn't that give us Vista? :p
But in all seriousness, its always a matter of tradeoffs. Important questions to ask yourself
What platform will your end users be using?
Can I drop cycles? What will happen if I do?
I agree with most that initial development should be done on the fastest or most efficient (not neccesarily the same) machine available to you. But for running tests, run it on your target platform, and test often and early.
Depends on your time to delivery. If you are in a 12 month delivery cycle then you should develop on a box with decent speed since your customers' 12 months from now will have better "average" boxes than the current "average".
As your development cycle approaches "today", your development machines should approach the current "average" speed of your clients' boxes.
I typically develop on the fastest machine I can get my hands on.
Most of the time I'm running a debug build, which is slow enough already.
I think it is a sound concept (but maybe because it works for me).
If my developer workstation is too fast I find I don't think ideas through thorougly enough simply because there is little time-penalty in re-generating the software image or downloading it to the target. I'd say at least half my downloads were unnecessary because I just remembered something I'd missed right before I was going to debug the code.
The target machine could well contain a throttled processor. If - on an embedded MCU - you have half the FLASH, RAM and clock cycles per second chances are developers will be a lot more careful when designing their code. I once suggested byte variables for the lengths of individual records in a data area (not in RAM but in a serial eeprom) and received the reply "we don't need to be stingy." A few months later they hit the RAM ceiling (128KiB). My reflection was that for this app there would never be any records larger than 256 bytes simply because there was no RAM to copy them to.
For server applications I think it would be a great idea to have a (much) lower-performing hardware to test on. Two or four cores instead of sixteen (or more). 1.6 GHz istdo 2.8. The list goes on. A server is usually - due to the very fact that everyone talks to it - a bottleneck in the system architecture. And that is long before you start developing the (server) application for it.
A two parter:
1) Say you're designing a new type of application and you're in the process of coming up with new algorithms to express the concepts and content -- does it make sense to attempt to actively not consider optimisation techniques at that stage, even if in the back of your mind you fear it might end up as O(N!) over millions of elements?
2) If so, say to avoid limiting cool functionality which you might be able to optimise once the proof-of-concept is running -- how do you stop yourself from this programmers habit of a lifetime? I've been trying mental exercises, paper notes, but I grew up essentially counting clock cycles in assembler and I continually find myself vetoing potential solutions for being too wasteful before fully considering the functional value.
Edit: This is about designing something which hasn't been done before (the unknown), when you're not even sure if it can be done in theory, never mind with unlimited computing power at hand. So answers along the line of "of course you have to optimise before you have a prototype because it's an established computing principle," aren't particularly useful.
I say all the following not because I think you don't already know it, but to provide moral support while you suppress your inner critic :-)
The key is to retain sanity.
If you find yourself writing a Theta(N!) algorithm which is expected to scale, then you're crazy. You'll have to throw it away, so you might as well start now finding a better algorithm that you might actually use.
If you find yourself worrying about whether a bit of Pentium code, that executes precisely once per user keypress, will take 10 cycles or 10K cycles, then you're crazy. The CPU is 95% idle. Give it ten thousand measly cycles. Raise an enhancement ticket if you must, but step slowly away from the assembler.
Once thing to decide is whether the project is "write a research prototype and then evolve it into a real product", or "write a research prototype". With obviously an expectation that if the research succeeds, there will be another related project down the line.
In the latter case (which from comments sounds like what you have), you can afford to write something that only works for N<=7 and even then causes brownouts from here to Cincinnati. That's still something you weren't sure you could do. Once you have a feel for the problem, then you'll have a better idea what the performance issues are.
What you're doing, is striking a balance between wasting time now (on considerations that your research proves irrelevant) with wasting time later (because you didn't consider something now that turns out to be important). The more risky your research is, the more you should be happy just to do something, and worry about what you've done later.
My big answer is Test Driven Development. By writing all your tests up front then you force yourself to only write enough code to implement the behavior you are looking for. If timing and clock cycles becomes a requirement then you can write tests to cover that scenario and then refactor your code to meet those requirements.
Like security and usability, performance is something that has to be considered from the beginning of the project. As such, you should definitely be designing with good performance in mind.
The old Knuth line is "We should forget about small efficiencies, say about 97% of the time: premature optimization is the root of all evil." O(N!) to O(poly(N)) is not a "small efficiency"!
The best way to handle type 1 is to start with the simplest thing that could possibly work (O(N!) cannot possibly work unless you're not scaling past a couple dozen elements!) and encapsulate it from the rest of the application so you could rewrite it to a better approach assuming that there is going to be a performance issue.
Optimization isn't exactly a danger; its good to think about speed to some extent when writing code, because it stops you from implementing slow and messy solutions when something simpler and faster would do. It also gives you a check in your mind on whether something is going to be practical or not.
The worst thing that can happen is you design a large program explicitly ignoring optimization, only to go back and find that your entire design is completely useless because it cannot be optimized without completely rewriting it. This never happens if you consider everything when writing it--and part of that "everything" is potential performance issues.
"Premature optimization is the root of all evil" is the root of all evil. I've seen projects crippled by overuse of this concept. At my company we have a software program that broadcasts transport streams from disk on the network. It was originally created for testing purposes (so we would just need a few streams at once), but it was always in the program's spec requirements that it work for larger numbers of streams so it could later be used for video on demand.
Because it was written completely ignoring speed, it was a mess; it had tons of memcpys despite the fact that they should never be necessary, its TS processing code was absurdly slow (it actually parsed every single TS packet multiple times), and so forth. It handled a mere 40 streams at a time instead of the thousands it was supposed to, and when it actually came time to use it for VOD, we had to go back and spend a huge amount of time cleaning it up and rewriting large parts of it.
"First, make it run. Then make it run fast."
or
"To finish first, first you have to finish."
Slow existing app is usually better than ultra-fast non-existing app.
First of all peopleclaim that finishign is only thing that matters (or almost).
But if you finish a product that has O(N!) complexity on its main algorithm, as a rule of thumb you did not finished it! You have an incomplete and unacceptable product for 99% of the cases.
A reasonable performance is part of a working product. A perfect performance might not be. If you finish a text editor that needs 6 GB of memory to write a short note, then you have not finished a product at all, you have only a waste of time at your hands.. You must remember always that is not only delivering code that makes a product complete, is making it achieve capability of supplying the costumer/users needs. If you fail at that it matters nothing that you have finished the code writing in the schedule.
So all optimizations that avoid a resulting useless product are due to be considered and applied as soon as they do not compromise the rest of design and implementation proccess.
"actively not consider optimisation" sounds really weird to me. Usually 80/20 rule works quite good. If you spend 80% of your time to optimize program for less than 20% of use cases, it might be better to not waste time unless those 20% of use-cases really matter.
As for perfectionism, there is nothing wrong with it unless it starts to slow you down and makes you miss time-frames. Art of computer programming is an act of balancing between beauty and functionality of your applications. To help yourself consider learning time-management. When you learn how to split and measure your work, it would be easy to decide whether to optimize it right now, or create working version.
I think it is quite reasonable to forget about O(N!) worst case for an algorithm. First you need to determine that a given process is possible at all. Keep in mind that Moore's law is still in effect, so even bad algorithms will take less time in 10 or 20 years!
First optimize for Design -- e.g. get it to work first :-) Then optimize for performance. This is the kind of tradeoff python programmers do inherently. By programming in a language that is typically slower at run-time, but is higher level (e.g. compared to C/C++) and thus faster to develop, python programmers are able to accomplish quite a bit. Then they focus on optimization.
One caveat, if the time it takes to finish is so long that you can't determine if your algorithm is right, then it is a very good time to worry about optimization earlier up stream. I've encountered this scenario only a few times -- but good to be aware of it.
Following on from onebyone's answer there's a big difference between optimising the code and optimising the algorithm.
Yes, at this stage optimising the code is going to be of questionable benefit. You don't know where the real bottlenecks are, you don't know if there is going to be a speed problem in the first place.
But being mindful of scaling issues even at this stage of the development of your algorithm/data structures etc. is not only reasonable but I suspect essential. After all there's not going to be a lot of point continuing if your back-of-the-envelope analysis says that you won't be able to run your shiny new application once to completion before the heat death of the universe happens. ;-)
I like this question, so I'm giving an answer, even though others have already answered it.
When I was in grad school, in the MIT AI Lab, we faced this situation all the time, where we were trying to write programs to gain understanding into language, vision, learning, reasoning, etc.
My impression was that those who made progress were more interested in writing programs that would do something interesting than do something fast. In fact, time spent worrying about performance was basically subtracted from time spent conceiving interesting behavior.
Now I work on more prosaic stuff, but the same principle applies. If I get something working I can always make it work faster.
I would caution however that the way software engineering is now taught strongly encourages making mountains out of molehills. Rather than just getting it done, folks are taught to create a class hierarchy, with as many layers of abstraction as they can make, with services, interface specifications, plugins, and everything under the sun. They are not taught to use these things as sparingly as possible.
The result is monstrously overcomplicated software that is much harder to optimize because it is much more complicated to change.
I think the only way to avoid this is to get a lot of experience doing performance tuning and in that way come to recognize the design approaches that lead to this overcomplication. (Such as: an over-emphasis on classes and data structure.)
Here is an example of tuning an application that has been written in the way that is generally taught.
I will give a little story about something that happened to me, but not really an answer.
I am developing a project for a client where one part of it is processing very large scans (images) on the server. When i wrote it i was looking for functionality, but i thought of several ways to optimize the code so it was faster and used less memory.
Now an issue has arisen. During Demos to potential clients for this software and beta testing, on the demo unit (self contained laptop) it fails due to too much memory being used. It also fails on the dev server with really large files.
So was it an optimization, or was it a known future bug. Do i fix it or oprtimize it now? well, that is to be determined as their are other priorities as well.
It just makes me wish I did spend the time to reoptimize the code earlier on.
Think about the operational scenarios. ( use cases)
Say that we're making a pizza-shop finder gizmo.
The user turns on the machine. It has to show him the nearest Pizza shop in meaningful time. It Turns out our users want to know fast: in under 15 seconds.
So now, any idea you have, you think: is this going to ever, realistically run in some time less than 15 seconds, less all other time spend doing important stuff..
Or you're a trading system: accurate sums. Less than a millisecond per trade if you can, please. (They'd probably accept 10ms), so , agian: you look at every idea from the relevant scenarios point of view.
Say it's a phone app: has to start in under (how many seconds)
Demonstrations to customers fomr laptops are ALWAYS a scenario. We've got to sell the product.
Maintenance, where some person upgrades the thing are ALWAYS a scenario.
So now, as an example: all the hard, AI heavy, lisp-customized approaches are not suitable.
Or for different strokes, the XML server configuration file is not user friendly enough.
See how that helps.
If I'm concerned about the codes ability to handle data growth, before I get too far along I try to set up sample data sets in large chunk increments to test it with like:
1000 records
10000 records
100000 records
1000000 records
and see where it breaks or becomes un-usable. Then you can decide based on real data if you need to optimize or re-design the core algorithms.