I’ve been attending the Future of Web Applications conference in London. I spoke to several attendees in the evening and the general perception was that the event had been weaker than usual so far. Complaints concerned uninspiring sessions, lack of deep technical content, and information on HTML 5 that was really nothing new.
That said, several said how much they enjoyed a session from Hilary Mason at bit.ly on data analysis. Bit.ly does url shortening, with 70% of so of its traffic coming from Twitter clients, and Mason is a statistical expert who has worked on analysing and visualising the resulting data. She told us, for example, that news links are more popular than sports links, and sports links more popular than food links. She was also able to discover the best time to post a link for any particular Twitter account, if you want maximum clicks. There is no quick way to discover this, so this type of analysis is valuable for companies using Twitter as a PR tool. Another snippet of information was the half-life of a typical bit.ly link – in other words, the time interval by which it has recorded 50% of its likely total clicks – which in the example she showed us was between 20 and 25 minutes.
The consequence was that I went into the next session, on social gaming, with data analysis on my mind. The session was presented by Kristian Segerstrale at Playfish, part of Electronic Arts focused on casual games for Facebook and the like. Gaming by the way is a huge part of Facebook, accounting for 30% to 40% of overall engagement, according to Segerstrale. As an insight into the future of gaming, it was a good session, but perhaps did not connect well with typical FOWA attendees.
Nevertheless, Segerstrale made a compelling point about how his company’s games evolve, which is also applicable to other kinds of web applications. He said that there is intense analysis of what works and what does not work, based on the flow of data that is available with web applications. You can see who is playing, when they are playing, which features are used, and get a level of insight into the strengths and weaknesses of your application which is typically unavailable for desktop applications. I imagine this works particularly well within Facebook, because of the rich user profile information there. If you take advantage of that data, you can get a lead over the competition; if you fail to make use of it, you will likely fall behind. There is now a data analytics skills gap, Segerstrale told us.
It was thought-provoking to see how data analytics was a common thread between such different sessions.
I’m waiting for Microsoft CEO Steve Ballmer to speak at the London School of Economics, which seems a good moment to reflect on his well-known war cry “Developers Developers Developers”.
Behind the phrase is a theory about how to make your platform succeed. The logic is something like this. Successful platforms have lots of applications, and applications are created by developers. If you make your platform appealing to developers, they will build applications which users will want to run, therefore your platform will win in the market.
Today though we have an interesting case study – Apple’s iPhone. The iPhone has lots of apps and is winning in the market, but not because Apple made it appealing to developers. In fact, Apple put down some roadblocks for developers. The official SDK has one programming language, Objective C, which is not particularly easy to use, and unlikely to be known other than by existing Apple platform developers. Apps can only be distributed through Apple’s store, and you have to pay a fee as well as submit to an uncertain approval process to get your apps out there. Some aspects of iPhone (and iPad) development have improved since its first launch. A clause in the developer agreement forbidding use of languages other than Objective C was introduced and then removed, and the criteria for approval have been clearly stated. Nevertheless, the platform was already successful. It is hard to argue that the iPhone has prospered thanks to Apple’s developer-friendly policies.
Rather, the iPhone succeeded because its design made it appealing to users and customers. Developers went there because Apple created a ready market for their applications. If Apple CEO Steve Jobs were prone to shouting words in triplicate, they might be “Design Design Design” or “Usability usability usability”. And as for developers, what they want is “Customers customers customers.”
Well, there are vicious and virtuous circles here. Clearly it pays, in general, to make it easy for developers to target your platform. Equally, it is not enough.
Microsoft’s own behaviour shows a shift in focus towards winning customers through usability, thanks no doubt to Apple’s influence and competition. Windows 7 and Windows Phone 7 demonstrate that. Windows Phone 7 is relatively developer-friendly, particularly for .NET developers, since applications are built on Silverlight, XNA and the .NET Framework. If it succeeds though, it will be more because of its appeal to users than to developers.
What do developers want? Customers customers customers.
I’m at the Future of Web Applications conference in London, a crazy mixture of tips for web start-ups and general discussion about application development in a web context. The first session was from Google’s Michael Mahemoff who enthused about HTML5 and open web standards, while refusing to be pinned down on what HTML5 is, which standards are in and which may in the end be out.
Microsoft is here showing off IE9; but one of my reflections is that while the HTML5 support in IE9 is impressive in itself, there are going to be important parts of what, say, Google considers to be part of HTML5 that will not be in IE9, and given the pace of Microsoft’s browser development, probably will not turn up for some time. In other words, the pressure to switch to Chrome, Firefox or some other browser will likely continue.
I digress. Mahemoff identified four key features of web apps – by which he means something different than just an application on the web. These are:
Local storage – encompassing local storage API and also local SQL, though the latter is not yet well advanced
Application cache – Cache Manifest in HTML 5 that lets your app run offline
Local installation – interesting as this is something which is not yet widely used, but clearly part of Google’s vision for Chrome, and also in IE9 to some extent.
Payments
The last of these is interesting, and I sensed Mahemoff showing some discomfort as he steered his way between open web standards on the one hand, and Google-specific features on the other. He presented the forthcoming Chrome Web Store as the solution for taking payments for your web app, whether one-time or subscription.
I asked how this would work with regard to the payment provider – could you freely use PayPal, direct debits or other systems? He said that you could do if you wanted, but he anticipated that most users would use the system built into Chrome Web Store which I presume is Google Checkout. After all, he said, users will already be logged in, and this will offer the smoothest payment experience for them.
The side effect is that if Chrome Web Store takes off, Google gets to make a ton of money from being the web’s banker.
Outside in the exhibition area Vodafone is promoting its 360 app store, with payments going through the mobile operator, ie in this case Vodafone. Vodafone’s apps are for mobile not for web, but it is relevant because it is trying to draw users away from Google’s Android Marketplace and onto its own store. PayPal is here too, showing its developer API.
The app store and payment provider wars will be interesting to watch.
There has always been an uneasy balance between Java as a cross-platform, cross-vendor standard; and Java as a proprietary technology. Under Sun’s stewardship the balance was tilted towards the cross-platform standard. Eventually, Java was open-sourced as the OpenJDK. However, Sun, and therefore now Oracle following its acquisition of Sun, still owns Java. The official Java specification is determined by the optimistically-named Java Community Process (JCP). The JCP is a democratic organisation up to a point, the point in question being clause 5.9 in the JCP procedures:
EC ballots to approve UJSRs for new Platform Edition Specifications or JSRs that propose changes to the Java language, are approved if (a) at least a two-thirds majority of the votes cast are "yes" votes, (b) a minimum of 5 "yes" votes are cast, and (c) Sun casts one of the "yes" votes. Ballots are otherwise rejected.
In other words, nothing happens without Sun’s approval.
Now the Register reports that Oracle and the JCP have fallen out. According to this report, the JCP does not like Oracle’s suit against Google; and does not have confidence in Java FX or Java ME both of which were promoted at the recent OpenWorld/JavaOne conference (though Java FX is to change significantly). The JCP still wants true independence – as, amusingly, proposed by Oracle in 2007:
… that the JCP become an open independent vendor-neutral Standards Organization where all members participate on a level playing field with the following characteristics:
members fund development and management expenses
a legal entity with by-laws, governing body, membership, etc.
a new, simplified IPR Policy that permits the broadest number of implementations
stringent compatibility requirements
dedicated to promoting the Java programming model
Oracle seems now to have changed its mind, wanting to tighten rather than loosen control over Java. Oracle still needs to work through the JCP in order to progress the Java specification so it will need either to mend relationships or reform the JCP somehow in order to deliver what was promised at JavaOne.
What does this mean for Java and its future? Perhaps surprisingly little. Alex Handy at the sdtimes reports this comment from Rod Johnson, now at VMware, whose SpringSource business was built on building Java frameworks outside the JCP:
There’s been very little activity on the [JCP] executive committee. I think we just have to wait and see what Oracle comes up with for JavaOne," he said. "The rest of the world is moving along fairly quickly. It’s not like we need Oracle or the EC of the JCP to get things done.
Java is the world’s most popular programming language. Further, Oracle is a smart company and although it is doing a good job of alienating members of the Java community – not least inventor James Gosling, now a loose cannon on deck – its technical work on Java will likely be excellent. That said, we are heading into an increasingly fractured world in terms of development platforms, especially in mobile, and that looks unlikely to change.
Blackberry has announced its pitch for the emerging tablet market, the 7” screen PlayBook. It has a new OS base on QNX Neutrino, a webkit-based web browser, Adobe Flash and AIR – offline Flash applications – front and rear cameras for video conferencing as well as taking snaps, and includes a USB port and HDMI out. There is wi-fi and Bluetooth but no 3G in the first release; you can connect to the Internet via your Blackberry. Storage is not yet specified as far as I can tell. There is no physical keyboard, which is surprising in some ways as the keyboard is the reason I hear most often for users choosing a BlackBerry smartphone over Apple’s iPhone.
Alongside the PlayBook RIM has announced a new developer platform. WebWorks is an HTML5 platform extended with access to local APIs, and targets both the Tablet OS and BlackBerry smartphones:
BlackBerry WebWorks applications can tap into the always-on, notification-based, push-enabled, contextual and social attributes of the BlackBerry smartphone. These applications can also access hardware features and integrate with other apps, and are powerful Super Apps that are fully integrated into the BlackBerry Application Platform.
In order to access local resources you need to package your app as a Blackberry application. Java and native C applications are also supported.
A winner? Well, there is a widespread industry presumption that we all want tablets; for example NVIDIA CEO Jen-Hsun Huang is planning on this basis, judging by what he said to the press last week. It is certainly a market in which every vendor wants a presence. There are a number of open questions though. The new tablet market is really defined by Apple’s iPad, and success for other operating systems and form factors is yet to be demonstrated. Personally I am not sure about the 7-inch screen, which is perhaps too large for a pocket and too small for the desktop-like web browsing you can do on an iPad. Here are the dimensions:
BlackBerry PlayBook: 7-inch 1024×600 screen,130mm x 193mm x 10mm
Apple iPad: 9.7-inch 1024×768 screen, 189.7mm x 242.8mm x 13.4mm
I doubt there will be much enthusiasm for carting around a phone, a small tablet, and a laptop, so in order to be viable as a portable device for work it has to be a laptop replacement. I do see this happening already with the iPad, though for me personally a netbook is both cheaper and more practical.
Apps are another key factor. It is smart of RIM to support Flash and AIR, which along with HTML 5 web applications are likely the best bet for supporting something like the PlayBook without a lot of device-specific work.
Exhibiting here at the NVIDIA GPU Technology Conference is a Cambridge-based company called tidepowerd, whose product GPU.NET brings GPU programming to .NET developers. The product includes both a compiler and a runtime engine, and one of the advantages of this hybrid approach is that your code will run anywhere. It will use both NVIDIA and AMD GPUs, if they support GPU programming, or fall back to the CPU if neither is available. From the samples I saw, it also looks easier than coding CUDA C or OpenCL; just add an attribute to have a function run on the GPU rather than the CPU. Of course underlying issues remain – the GPU is still isolated and cannot access global variables available to the rest of your application – but the Visual Studio tools try to warn of any such issues at compile time.
GPU.NET is in development and will go into public beta “by October 31st 2010” – head to the web site for more details.
I am not sure what sort of performance or other compromises GPU.NET introduces, but it strikes me that this kind of tool is needed to make GPU programming accessible to a wider range of developers.
NVIDIA CEO Jen-Hsung Huang spoke to the press at the GPU Technology Conference and I took the opportunity to ask some questions.
I asked for his views on the cloud as a supercomputer and whether that would impact the need for local supercomputers of the kind GPU computing enables.
Although we expect more and more to happen in the cloud, in the meantime we’re going to keep buying devices with more and more solid state memory. The way to think about it is, storage is simply a surrogate for bandwidth. If we had infinite bandwidth none of us would need storage. As bandwidth improves the requirement for storage should reduce. But there’s another trend which is that the amount of data we collect is growing incredibly fast … It’s going to be quite a long time before our need for storage will reduce.
But what about local computing power, Gigaflops as opposed to storage?
Wherever there is storage, there’s GigaFlops. Local storage, local computing.
Next, I brought up a subject which has been puzzling me here at GTC. You can do GPU programming with NVIDIA’s CUDA C, which only works on NVIDIA GPUs, or with OpenCL which works with other vendor’s GPUs as well. Why is there more focus here on CUDA, when on the face of it developers would be better off with the cross-GPU approach? (Of course I know part of the answer, that NVIDIA does not mind locking developers to its own products).
The reason we focus all our evangelism and energy on CUDA is because CUDA requires us to, OpenCL does not. OpenCL has the benefit of IBM, AMD, Intel, and ourselves. Now CUDA is a little difference in that its programming approach is different. Instead of an API it’s a language extension. You program in C, it’s a different model.
The reason why CUDA is more adopted than OpenCL is because it is simply more advanced. We’ve invested in CUDA much longer. The quality of the compiler is much better. The robustness of the programming environment is better. The tools around it are better, and there are more people programming it. The ecosystem is richer.
People ask me how do we feel about the fact that it is proprietary. There’s two ways to think about it. There’s CUDA and there’s Tesla. Tesla’s not proprietary at all, Tesla supports OpenCL and CUDA. If you bought a server with Tesla in it, you’re not getting anything less, you’re getting CUDA more. That’s the reason Tesla has been adopted by all the OEMs. If you want a GPU cluster, would you want one that only does OpenCL? Or does OpenCL and CUDA? 80% of GPU computing today is CUDA, 20% is OpenCL. If you want to reach 100% of it, you’re better off using Tesla. Over time, if more people use OpenCL that’s fine with us. The most important thing is GPU computing, the next most important thing to us is NVIDIA’s GPUs, and the next is CUDA. It’s way down the list.
Next, a hot topic. Jen-Hsun Huang explained why he announced a roadmap for future graphics chip architectures – Kepler in 2011, Maxwell in 2013 – so that software developers engaged in GPU programming can plan their projects. I asked him why Fermi, the current chip architecture, had been so delayed, and whether there was good reason to have confidence in the newly announced dates.
He answered by explaining the Fermi delay in both technical and management terms.
The technical answer is that there’s a piece of functionality that is between the shared symmetric multiprocessors (SMs), 236 processors, that need to communicate with each other, and with memory of all different types. So there’s SMs up here, and underneath the memories. In between there is a very complicated inter-connecting system that is very fast. It’s nearly all wires, dense metal with very little logic … we call that the fabric.
When you have wires that are next to each other that closely they couple, they interfere … it’s a solid mesh of metal. We found a major breakdown between the models, the tools, and reality. We got the first Fermi back. That piece of fabric – imagine we are all processors. All of us seem to be working. But we can’t talk to each other. We found out it’s because the connection between us is completely broken. We re-engineered the whole thing and made it work.
Your question was deeper than that. Your question wasn’t just what broke with Fermi – it was the fabric – but the question is how would you not let it happen again? It won’t be fabric next time, it will be something else.
The reason why the fabric failed isn’t because it was hard, but because it sat between the responsibility of two groups. The fabric is complicated because there’s an architectural component, a logic design component, and there’s a physics component. My engineers who know physics and my engineers who know architecture are in two different organisations. We let it sit right in the middle. So the management lesson learned – there should always be a pilot in charge.
Huang spent some time discussing changes in the industry. He identifies mobile computing “superphones” and tablets as the focus of a major shift happening now. Someone asked “What does that mean for your Geforce business?”
I don’t think like that. The way I think is, “what is my personal computer business”. The personal computer business is Geforce plus Tegra. If you start a business, don’t think about the product you make. Think about the customer you’re making it for. I want to give them the best possible personal computing experience.
Tegra is NVIDIA’s complete system on a chip, including ARM processor and of course NVIDIA graphics, aimed at mobile devices. NVIDIA’s challenge is that its success with Geforce does not guarantee success with Tegra, for which it is early days.
The further implication is that the immediate future may not be easy, as traditional PC and laptop sales decline.
The mainstream business for the personal computer industry will be rocky for some time. The reason is not because of the economy but because of mobile computing. The PC … will be under disruption from tablets. The difference between a tablet and a PC is going to become very small. Over the next few years we’re going to see that more and more people use their mobile device as their primary computer.
[Holds up Blackberry] There’s no question right now that this is my primary computer.
The rise of mobile devices is a topic Huang has returned to on several occasions here. “ARM is the most important CPU architecture, instruction set architecture, of the future” he told the keynote audience.
Clearly NVIDIA’s business plans are not without risk; but you cannot fault Huang for enthusiasm or awareness of coming changes. It is clear to me that NVIDIA has the attention of the scientific and academic community for GPU computing, and workstation OEMs are scrambling to built Tesla GPU computing cards into their systems, but transitions in the market for its mass-market graphics cards will be tricky for the company.
Update: Huang’s comments about the reasons for Fermi’s delay raised considerable interest as apparently he had not spoken about this on record before. Journalist Nico Ernst captured the moment on video:
I’m at NVIDIA’s GPU tech conference in San Jose. The central theme of the conference is that the capabilities of modern GPUs enable substantial performance gains for general computing, not just for graphics, though most of the examples we have seen involve some element of graphical processing. The reason you should care about this is that the gains are huge.
Take Matlab for example, a popular language and IDE for algorithm development, data analysis and mathematical computation. We were told in the keynote here yesterday that Matlab is offering a parallel computing toolkit based on NVIDIA’s CUDA, with speed-ups from 10 to 40 times. Dramatic performance improvements opens up new possibilities in computing.
Why has GPU performance advanced so rapidly, whereas CPU performance has levelled off? The reason is that they use different computing models. CPUs are general-purpose. The focus is on fast serial computation, executing a single thread as rapidly as possible. Since many applications are largely single-thread, this is what we need, but there are technical barriers to increasing clock speed. Of course multi-core and multi-processor systems are now standard, so we have dual-core or quad-core machines, with big performance gains for multi-threaded applications.
By contrast, GPUs are designed to be massively parallel. A Tesla C1060 has not 2 or 4 or 8 cores, but 240; the C2050 has 448. These are not the same as CPU cores, but nevertheless do execute in parallel. The clock speed is only 1.3Ghz, whereas an Intel Core i7 Extreme is 3.3Ghz, but the Intel CPU has a mere 6 cores. An Intel Xeon 7560 runs at 2.266 Ghz and has 8 cores.The lower clock speed in the GPU is one reason it is more power-efficient.
NVIDIA’s CUDA initiative is about making this capability available to any application. NVIDIA made changes to its hardware to make it more amenable to standard C code, and delivered CUDA C with extensions to support it. In essence it is pretty simple. The extensions let you specify functions to execute on the GPU, allocate memory for pointers on the GPU, and copy memory between the GPU (called the device) and the main memory on the PC (called the host). You can also synchronize threads and use shared memory between threads.
The reward is great performance, but there are several disadvantages. One is the challenge of concurrent programming and the subtle bugs it can introduce.
Another is the hassle of copying memory between host and device. The device is in effect a computer within a computer. Shifting data between the two is relatively show.
A third is that CUDA is proprietary to NVIDIA. If you want your code to work with ATI’s equivalent, called Streams, then you should use the OpenCL library, though I’ve noticed that most people here seem to use CUDA; I presume they are able to specify the hardware and would rather avoid the compromises of a cross-GPU library. In the worst case, if you need to support both CUDA and non-CUDA systems, you might need to support different code paths depending on what is detected at runtime.
It is all a bit messy, though there are tools and libraries to simplify the task. For example, this morning we heard about GMAC, which makes host and device appear to use a single address space, though I imagine there are performance implications.
NVIDIA says it is democratizing supercomputing, bringing high performance computing within reach for almost anyone. There is something in that; but at the same time as a developer I would rather not think about whether my code will execute on the CPU or the GPU. Viewed at the highest level, I find it disappointing that to get great performance I need to bolster the capabilities of the CPU with a specialist add-on. The triumph of the GPU is in a sense the failure of the CPU. Convergence in some form or other strikes me as inevitable.
Runtime Revolution has renamed its software development IDE and runtime to LiveCode, which it says is a “modern descendent of natural-language technologies such as Apple’s HyperCard.” The emphasis is on easy and rapid development using visual development supplemented with script.
It is now a cross-platform development platform that targets Windows, Mac and Linux. Android is promised soon, there is a pre-release for Windows Mobile, and a new pre-release targets Apple’s iOS for iPad and iPhone.
LiveCode primarily creates standalone applications, but there is also a plug-in for hosting applets in the browser, though this option will not be available for iOS.
Now that Apple has lifted its restrictions on cross-platform development for iOS, it is Microsoft’s Windows Phone 7 that looks more of a closed device. The problem here is that Microsoft does not permit native code on Windows Phone 7, a restriction which also prohibits alternative runtimes such as LiveCode. You have to code applications in Silverlight or XNA. However, Adobe is getting a special pass for Flash, though it will not be ready in time for the first release of Windows Phone 7.
If Windows Phone 7 is popular, I imagine other companies will be asking for special passes. The ubiquity of Flash is one factor holding back Silverlight adoption, so in some ways it is surprising that Microsoft gives it favoured treatment, though it makes a nice selling point versus Apple’s iPhone.
I’m not able to attend the whole of Oracle OpenWorld / JavaOne, but I have sneaked in to MySQL Sunday, which is a half-day pre-conference event. One of the questions that interests me: is MySQL in safe hands at Oracle, or will it be allowed to wither in order to safeguard Oracle’s closed-source database business?
It is an obvious question, but not necessarily a sensible one. There is some evidence for a change in direction. Prior to the takeover, the MySQL team was working on a database engine called Falcon, intended to lift the product into the realm of enterprise database management. Oracle put Falcon on the shelf; Oracle veteran Edward Screven (who also gave the keynote here) said that the real rationale for Falcon was that InnoDB would be somehow jiggered by Oracle, and that now both MySQL and InnoDB were at Oracle, it made no sense.
Context: InnoDB is the grown-up database engine for MySQL, with support for transactions, and already belonged to Oracle from an earlier acquisition.
There may be something in it; but it is also true that Oracle has fine-tuned the positioning of MySQL. Screven today emphasised that MySQL is Oracle’s small and nimble database manager; it is “quite performant and quite functional”, he said; the word “quite” betraying a measure of corporate internal conflict. Screven described how Oracle has improved the performance of MySQL on Windows and is cheerful about the possibility of it taking share from Microsoft’s SQL Server.
It is important to look at the actions as well as the words. Today Oracle announced the release candidate of MySQL 5.5, which uses InnoDB by default, and has performance and scalability improvements that are dramatic in certain scenarios, as well as new and improved tools. InnoDB is forging ahead, with the team working especially on taking better advantage of multi-core systems; we also heard about full text search coming to the engine.
The scalability of MySQL is demonstrated by some of its best-known deployments, including Facebook and Wikipedia. Facebook’s Mark Callaghan spoke today about making MySQL work well, and gave some statistics concerning peak usage: 450 million rows read per second, 3.5 million rows changed per second, query response time 4ms.
If pressed, Screven talks about complexity and reliability with critical data as factors that point to an Oracle rather than a MySQL solution, rather than lack of scalability.
In practice it matters little. No enterprise currently using an Oracle database is going to move to MySQL; aside from doubts over its capability, it is far too difficult and risky to switch your database manager to an alternative, since each one has its own language and its own optimisations. Further, Oracle’s application platform is built on its own database and that will not change. Customers are thoroughly locked in.
What this means is that Oracle can afford to support MySQL’s continuing development without risk of cannibalising its own business. In fact, MySQL presents an opportunity to get into new markets. Oracle is not the ideal steward for this important open source project, but it is working out OK so far.
