jfk: they demoed a prototype storage system which was a crystal cube which could be 'written' by firing 2 laser beams at a point within the cube to break a molecular bond. Data could be read back by analysing the scatter of a beam aimed through the same point.
What they haven't got is a way of re-writing data, reading the data even as fast as 1X CD, creating the material substrate with good purity or finding a substrate that is durable.
We will get there at some point, but these things take time.
Holographic is assuredly the way forward for compact disc storage. However, the areas that really need focus when we're talking about both primary and secondary storage are latency and speed.
Both are now woefully inadequate when you consider that even a bottom-end CPU can process a few billion operations a second. A cache-miss means the loss of over a thousand CPU cycles whilst it waits for the data to be dragged from RAM. If that data isn't in RAM but has been paged to HD, you're looking at millions of cycles before the CPU gets its data.
HD access latencies really haven't seen much improvement over the past few years. The only real improvements have been platter capacity, noise reduction and transfer speeds. NCQ has been a small improvement, but rarely applicable outside of a web server scenario.
RAM latency is pretty much subject to the speed of electricity through the circuitry involved. The only way forward is RAM directly incorporated into the CPU package or even the CPU die. This could perhaps be in the form of much extended L2 cache or something as low tech as a GDDR3 module incorporated into the CPU package, something AMD could probably manage quite well considering they already have their memory controller in the CPU die.
As always with any sooth-saying about technology, we can only wait and see.
What they haven't got is a way of re-writing data, reading the data even as fast as 1X CD, creating the material substrate with good purity or finding a substrate that is durable.
We will get there at some point, but these things take time.
Holographic is assuredly the way forward for compact disc storage. However, the areas that really need focus when we're talking about both primary and secondary storage are latency and speed.
Both are now woefully inadequate when you consider that even a bottom-end CPU can process a few billion operations a second. A cache-miss means the loss of over a thousand CPU cycles whilst it waits for the data to be dragged from RAM. If that data isn't in RAM but has been paged to HD, you're looking at millions of cycles before the CPU gets its data.
HD access latencies really haven't seen much improvement over the past few years. The only real improvements have been platter capacity, noise reduction and transfer speeds. NCQ has been a small improvement, but rarely applicable outside of a web server scenario.
RAM latency is pretty much subject to the speed of electricity through the circuitry involved. The only way forward is RAM directly incorporated into the CPU package or even the CPU die. This could perhaps be in the form of much extended L2 cache or something as low tech as a GDDR3 module incorporated into the CPU package, something AMD could probably manage quite well considering they already have their memory controller in the CPU die.
As always with any sooth-saying about technology, we can only wait and see.