Lately I’ve been experimenting with Cyberlink MediaShow Espresso, a simple video conversion utility that happens to support both CUDA and ATI Stream extensions. The idea was to compare performance across various Nvidia and ATI GPUs, but I quickly discovered a problem: MediaShow isn’t scriptable. To use it, you need to be there to click the right buttons at the right times. That’s not exactly an efficient approach for a test like this, and it invites human error.

Happily, a brief web search turned up the perfect solution: a free automation system named AutoIt. Using a BASIC-like scripting language, you can direct AutoIt to generate the necessary keypresses and mouse movements to automatically control any Windows application or feature.

It’s not limited to dumbly repeating a predefined series of actions, either. Scripts can monitor what’s happening on the desktop, and then sleep, branch, loop or manipulate windows and files in response. A generous library of example scripts and functions, along with contextual keyword help and a syntax-highlighting editor, makes it easy to implement some quite sophisticated logic. If you can explain to a friend how to use an application, you can use AutoIt to script it.

The package has a few bonus tricks too – you can compile AutoIt scripts to standalone executables, and even create GUIs for them. It’s such a powerful tool that I was surprised not to have heard of it before, especially since it’s been knocking around, through various versions, for more than a decade. But I suppose this kind of automation is a niche interest: writing scripts isn’t exactly fashionable, and while AutoIt’s active it effectively takes over your desktop, so it’s not exactly a time-saver.

All the same, I’m sure many of us have established a few repetitive desktop routines – for example, you might regularly open a group of applications and arrange their windows in a particular way, or perhaps you have a troublesome tool that requires you to click through multiple requesters every time it launches. Knock up a simple script and AutoIt can do the legwork for you.

Update: Thanks to those of you who have suggested AutoHotKey as an alternative. I confess, that’s a package I really ought to have known about, not least because it regularly features in the “Essential Programming” section of the PC Pro cover disc! Now it’s been brought to my attention I’ll be sure to check it out. And of course, any other comments or suggestions are very welcome…

In my last post I suggested that DirectX 11’s extensive GPGPU support could mark the end of the road for CUDA. And I do expect that mass market GPU applications will quickly move to DirectX rather than restricting themselves to a single architecture.

But the other day I was discussing DX11 with Bit-Tech editor Tim Smalley, and I found him very reluctant to write CUDA off just yet. He pointed out that CUDA retains one big advantage over DX11, in that developers can knock up CUDA routines directly in C – or Fortran or even Matlab – without having to deal with the DirectX API.

Two different markets

Thinking about this, I’ve realised that there are in fact two wholly separate markets for GPU computing. As a mainstream technology, it’s a great way for application developers to wring extra performance from whatever hardware the user happens to already own. In this market, it makes sense to write code that will benefit as many users as possible, which today means DirectX.

But the truth is that there aren’t many desktop tasks that really benefit all that much from GPU acceleration – everyone talks about video transcoding, physics simulations and AI for games, but once you move past those very specific applications it’s slim pickings. For now, the real potential of massively parallel computing appears to lie in its ability to accelerate scientific research.

An academic question

And when scientists and engineers are choosing a research platform, they don’t really care about issues like market share. Their code only needs to run on a handful of machines, and it’s no problem to design those machines to suit the task, rather than vice versa. Here, CUDA is a no-brainer, because it lets researchers program in familiar languages, producing code that can be maintained and expanded without having to learn a new API.

Nvidia realises this, of course, and rather than continuing to talk about games, it’s been carefully positioning CUDA as a friend of academia. As the photo above shows, the hallways at GTC this morning were filled with boards – not whiteboards, like at IDF, but display boards showing summaries of research projects. Some focused on graphical techniques; others targeted problems in biology, physics or engineering. But all of them had something in common…

It’s a bold display. It cleverly makes CUDA look like serious business while ATI is still worrying about copying DVDs onto iPods.

And, what’s more, it’s persuaded me that Mr Smalley does have a point. Clearly, CUDA isn’t going to vanish from these environments overnight. Why would it, when DirectX offers no advantage?

The road ahead

But while academia may be a respectable market, you have to question how much actual revenue Nvidia sees from each of these projects. And though the use of (more or less) industry standard C is working for CUDA right now, it will work against it when a real rival comes along – a rival, perhaps, that can offer even greater programmability, backed up by the kind of muscle that doesn’t need to worry about revenue.

So here’s my crazy prediction. Some form of high-level GPGPU interface will survive alongside DX11 for the foreseeable future. And for the time being that will be CUDA as a matter of default. But I don’t think CUDA will ever be a real money-maker for Nvidia. And within five years I predict that Nvidia’s share of the GPU computing market will be swallowed up… by Larrabee.

(The picture, in case you’re wondering, is a strange engine-type affair that’s been set up at the entrance to the delegates’ hotel, apparently to welcome us as we arrive. I guess that’s how they communicate with one another down here in the Valley.)

Under pressure

Yet despite the company’s outward confidence, you have to wonder whether the green team really enjoys its 72 hours under the spotlight. Last year the recurrent themes of the event were faulty notebook GPUs and Larrabee – hardly the issues Nvidia will have wanted to focus on.

This year the notebook GPU issue seems to have died away, and it’s become clear that Larrabee is no threat for the immediate future. Yet Nvidia is still on the back foot in several fights. In the desktop market, it’s been playing catch-up with ATI for over a year: it’s rumoured that a new generation of GeForce cards may be unveiled in the coming days, but they’ll have to be very impressive to compete with the ATI Radeon HD 5870.

Challenged by new technology

In the mobile and lightweight markets, meanwhile, the company’s Tegra and Ion platforms have attracted praise, but will soon be challenged by Intel’s two new Atom architectures – Moorestown for handhelds and the CE4100 for media devices. Nvidia did well to get to these markets first, but can it really cling onto the territory now Intel’s tanks are rolling into town?

And lastly there’s CUDA. This time last year, Nvidia was touting the flexibility of CUDA as a unique benefit of Nvidia hardware, and since then the technology has shown its potential in some real-world applications – notably the built-in CUDA acceleration for Premiere Pro CS4. But now that DirectX 11 is here, its extensive platform-agnostic GPGPU functions make CUDA’s exclusivity look like a liability.

With all this in mind, I think it will be very interesting to see what the company has to say for itself over the coming days. Stay tuned and I’ll report back with all the news and nuggets that emerge…

The 9400M is the key that Nvidia hopes will allow it to eat away at Intel’s share. Combining the north bridge, south bridge and GPU into one chip less than half the size of Intel’s GMA X4500HD, it could be the great leap forward we’ve been waiting so long for. The integrated graphics solution that can actually run the latest games – we’d almost given up hope.

With 16 parallel processing cores, the chip is at least 70% GPU, with the rest of the essentials crammed around the edges where there’s room (as the die image clearly shows).

It offers full support for DirectX 10, PhysX, CUDA and even contains dedicated hardware for high definition video processing – Nvidia claims full-spec Blu-ray capability on a single battery charge, including all the PiP and BD Live features.

Gaming power

To demonstrate its power, Nvidia’s notebook general manager Rene Haas gave us a little side-by-side demonstration – pitting the brand new GeForce 9400M-equipped Macbook against a Centrino 2 Sony Vaio FW. Running Call of Duty 4 on both, at 1,024 x 768 and medium settings, the results were eye-opening.

The Sony exhibited all the stuttering motion and painful hangs we’ve grown accustomed to in low-end laptops, barely gettting close to 10fps throughout the level. By contrast, the MacBook ran happily at what we’d estimate to be around 25fps throughout, barely even hiccuping when the action hotted up and the effects began flying. Sure, it’s not Crysis at Very High settings, but it is a laptop with integrated graphics running a cutting-edge game at a playable framerate – and that’s something no manufacturer has yet offered us.

Photoshop

As exciting as this is, it’s in non-gaming applications that more useful advances may be appreciated. Haas fired up Photoshop CS4 on both demo laptops and opened up a whopping 3GB image ready for manipulation. As anyone who runs Photoshop on a laptop will expect, rotating and zooming the image on the Intel-equipped Sony was patchy at best: zooming was achieved in paused steps, while a 90-degree rotation left us with a progress bar for more than a minute before we saw any change.

The 9400M-equipped MacBook, on the other hand, simply flew through the image. Rotation was handled on the fly, while zooming exhibited the same smoothness we loved in Microsoft’s Deep Zoom technology when it was first demoed.

(Click thumbnail to play video)

In short, laptops with integrated GPUs can now benefit from the same CUDA-enabled technological advances as those with power-hungry discrete cards taking up the slack.

Nvidia claims the 9400M will offer five times the performance in the same power envelope as its Intel integrated rival – a bold claim indeed, but one backed up by the demos we’ve seen. We’ll be running it through our own intensive performance and battery tests in the coming weeks as laptop manufacturers unveil their first offerings with the integrated GPU.

Future potential

I spoke to Haas after the demo and asked if he thought the success of this chipset would be detrimental to Nvidia’s discrete laptop chips, but he was unequivocal. “Are we cannibalising ourselves by releasing this? I don’t think we are. Larger laptops will always have the need for discrete graphics, and there are still plenty of people who’ll prefer the power of a discrete chip in smaller laptops.”

He also threw up a few very interesting prospects on the horizon for the Geforce 9400M. CUDA-based upscaling of DVDs to 720p or more is on the agenda, potentially doing away with the need for expensive HD drives. And your low-end integrated GPU could also soon be enhancing YouTube videos – pixel interpolation of grainy streamed video is a highly parallel-processor-intensive task which lends itself well to CUDA, and Haas was hopeful it could even be with us before the end of 2008.

Whether we get there this year or next, it’s clear that Nvidia’s newest baby has the potential to shake up the integrated graphics market in a huge way. Intel has had it pretty easy for years now, despite every new advance coming with claims of genuine 3D power, and every one failing to live up to that promise. Nvidia may be struggling in the graphics card market right now, but by finally proving that an integrated chip can handle gaming it’s just diverted a whole new revenue stream towards its own coffers. Intel needs to pull its socks up.

This is the ATI Radeon HD 4870, rumoured before its release to be one of the fastest cards around, and it has just landed in our labs, only one week after it was announced.

It’s benchmarking downstairs in our labs as we speak – hence the odd angle in the image above – but we can already tell you that it looks to be incredible value for money.

It’s not as fast as Nvidia’s GTX 280, as we hoped it might be earlier this month, but it looks to be matching the speed of the GTX 260 – while almost chopping its price in half.

We’ll have full benchmark results and a review on the way soon, so keep an eye out on the site.

They spent a great deal of time assuring the assembled members of the IT press that it was a waste of time – every game since before the turn of the Millennium (indeed, since the demise of voxels) because every game is made using polygons and that developers wouldn’t want to alter their techniques and systems around a new, somewhat experimental technology.

So, why have Nvidia gone and bought a ray-tracing company?

It’s certainly a bit of a strange move for a company that’s previously denounced the system as pretty worthless for Nvidia’s main market: games. Then again, they did buy Ageia and their Physx technology, which has barely made a ripple in the virtual oceans of games like Crysis and Oblivion, so they do have a history of odd investments.

But, for all this prophesising, their position has recently take a bit of an about-face. The company’s CTO, David Kirk, claimed that ray-tracing was suddenly part of their plans – and that they could integrate it with traditional rendering techniques to make games and graphical applications look even better. Even so, he still doesn’t sound entirely convinced, emphasising that ray-tracing is only ‘part of the future‘, admitting that, at the moment, ‘ray tracing is currently significantly slower than rasterization’.

So, what’s caused such a dramatic turnaround?

We’ve theorised before that Nvidia, for all their bluster, seem to be making various loud, angry noises out of fear. Sure, very little is known about Larrabee, but it’s another competitor in Nvidia’s main marketplace which, aside from ATI, they’ve had little recent competition in.

This just seems like more evidence to support this line of thought – if Intel are going to be incorporating their own ray-tracing and physics technology in their new GPU – developed in-house – then Nvidia need to compete. What better way, then, than by buying up companies already specialised in these things?

Or is there another method to Nvidia’s (apparent) madness?

Evidently, it’s decided to come out on the offensive: Nvidia boss Jen-Hsun Huang recently lambasted Intel’s integrated graphics, which have long been a staple of PCs that don’t need to play games and edit demanding videos, as ‘a joke’. He also boasted of his plans to ‘open a can of whoop-ass’ onto Intel, which must be quaking in its boots – after all, its CPUs haven’t done that well, and they certainly not market leaders with no real competitors. Ahem.

In a move that could be likened to David squaring up to Goliath and promising to break his legs with his arms tied behind his back and a blindfold on, Nvidia has also now announced its new chipset. The MCP79 is designed to stomp all over Intel’s well-defended backyard after recent claims that Intel “can’t write a graphics driver to save their life”.

It’s certainly a brave position to take – at a recent press event, Nvidia was keen to stress that, as well as excellent processors, Intel’s mighty marketing department was responsible for recent success. Given its dominance in the CPU market, I certainly wouldn’t put it past Intel to seriously frighten Nvidia with its new graphics card – in fact, judging by the aggressive and almost petulant reaction emerging from leading Nvidia figures these days, it could be reasoned that the company’s already pretty worried.

Even so, it makes for a tantalising future for graphics, as there hasn’t been a serious third player in the market for a fair few years – and a bit of playground scrapping could work wonders for the price of new GPU technology.