Slashdot Mirror

You'll be waiting a long time, I think. People have been trying this foolhardy idea of optical computing for a long time, and the field is pretty much dead.

The field isn't so much dead. What is "dead in the water" (at least for now, and as long as there is no further breaktrough) is the idea of building processors using photons the sale way as the current processors are using electrons.

However, optical processing has other applications that could be put into good use. Computing Fourier transforms like Lenslet is doing could be such an example.

This research has been going on way before Harvard in Israel. Go to http://www.lenslet.com/index.asp

They use light for DSPs and have for 6 years. Check it out its a cool project

You're probably thinking of Lenslet's optical digital signal processor, performing a variety of vector-matrix and fast fourier transform operations. 256 optical digital inputs, 256 o.d. outputs, and electronic selection of the operation.

White paper here

Now if there is a cheap clean way to do serial-to-parallel conversion on a gigabit/s optical digital datastream, Intel has created a neat device for feeding the Lenslet beast.

There's a nice picture of the processor here

And

Also, a demo video here

There's a nice picture of the processor here

And

Also, a demo video here

There is a nice picture of it here.

here

Analog multipliers like this have been built before, but this one has a major improvement: The matrix of constants can be changed. In most signal processing systems (particulary surface acoustic wave (SAW) devices), the operation performed is fixed. Changing the matrix of constants is probably much slower than the data rate of the optical components.

The only operations performed optically are analog multiplication by constants (which is straightforward) and analog addition (which is even easier). They haven't cracked the hard problem of developing a useful optical element which can do logical operations like NOT or NAND. So this doesn't lead to a general purpose processor.

It's going to be useful for radars and such, but it's overhyped.

The article says "The Ablaze(TM) is the Spatial Light Modulator (SLM) in the optical core of the EnLight256(TM)". Going by the graphic in the Inquirer article, they shine a row of blinking lights through a LCD-like device (and some lenses and mirrors I assume) and collect the results in a column of light sensors on the other end.

Each pattern of on/off elements on the LCD-like device gives them a different transformation running at however fast you could emit and sense the light. I doubt they mechanically move the optical arrangment so that would seem to limit the number of transformations. Some of the LCD patterns might give useful transformations. A vector multiply, a Fast Fourier Transform (maybe) or a sort (I doubt it)?

If the numbers are an analog light intensity level the precision would depend on how precise the light emitters and sensors you have are. Packaging the mirrors and lenses small enough is a neat trick. Having a problem that fits the available transformations and can supply data in and out fast enougth is another. I wonder anything useful can be done by quickly switching LCD matrix pattern, or directly feeding outputs back as inputs?

Okay, what the hell IS one of these things and how does it WORK?

There is a diagram on lenslet.com that shows how their optical processor works. There are three parts, a row of lasers, a row of photodetectors, and a big grid of Multiple Quantum Well (MQW) Spatial Light Modulator. I assume this grid is where the matrix operations actually take place. I don't even care about the math, I never could understand it. But from a physical point of view, how is this thing constructed? What _IS_ it?

I tried searching google but all I got were incomprehensible scientific papers. I'm not a layman but I'm also not an electro-optical engineer. Can someone explain this thing in language I can understand?

... at the Lenslet page, the unit actually has several components. The VMM (vector matrix multiplier) does 8000 MAC (matrix array calculations) but there is a VPU (vector processing unit) that comes in at 128 Giga-ops and which would be the bottleneck in the whole setup. No question this is a huge improvement BUT to put it in perspective, it is a DSP only, not a computer system (although some neural network weenies might see a way of turning this into something more than just a DSP). In any case, the bottlenecks will come from the equipment it has to operate with both onboard and off.

Still, note that it's developed with Matlab. Now surely that is the Holy Grail of research, a bitchin' language with an awesome tailored processor. Imagine the logo Matlab [Lenslet Inside].

Also, a demo video here

There's a nice picture of the processor here

Check out the Englight256... Coming soon to a military installation near you...