|
Kish on: |
||||
|
|
||||
|
|
||||
For a consumer device, the Flash is the third most expensive component, and in a thin-client [desktop], it is the second most expensive component, after the processor. For the latter, [Flash implementation] accounts for about 30 to 40 percent of the cost. This includes the electrical system required to support Flash, like battery, etc, etc.
Do you derive any technical benefits by forgoing Flash memory?
Chief advantage is that the applications don't exist when the device is turned on. More importantly, the device can take a number of different identities anytime it turns on. For example, we have some very large customers which are in the process of shifting from Microsoft- to Linux-based apps, and they want one device to do this. So they don't want to have a device that already has an operating system on it because then when they want to go to Linux, they have to throw that one away, and throw the licensing fee away. So our devices are such that you can run XP on it one day and the next day, run Linux. You can also run Blazer, which is an OS that we have. For a thin-client, the embedded OS is actually very highly-optimized. You don't need all of the OS because a thin-client doesn't do a number of the things that a typical PC might do, like managing a disk, for example.
What you want to think about is a device which effectively is nothing more than a flat memory space, with perhaps some communications infrastructure and display. This flat memory space can be made to do anything. What we are doing is simply addressing that space on boot-up and shifting over to it OS or app fragments that we choose to run locally.
So this new streaming technology is to allow customers to optimize what's running on the device for their particular installation. So, as an example, in Japan, it just passed a data protection act which came into effect on April 1, making it illegal to store citizen's data on a PC. So all of sudden, there is a great deal of interest on how can we do computing without storing data on it--which is of course exactly how a thin-client is designed to do, the data is never resident on a thin-client, the data is resident in the network and the servers, and all the thin-client is that it is a windows to the network.
What is a flat memory space?
It is volatile memory that is addressable, meaning when you write program or application for it, you can grab memory pointers by yourself, and you don't have to do it with an operating system. Like I used to write database applications, and one of the things you have to do in a database is to define an area of memory that you are effectively caching data to all the time, and you address that memory directly.
Just think of such a space as an empty floor. The OS will come in just like a chair where you can it place on the floor. And then the application that comes in will be put on the chair.
The advantage to that is that volatile memory is the cheapest form of memory, it doesn't require anywhere near the kinds of electronic overhead that you'll need with Flash memory, or RAM. Volatile memory is only about 10 to 15 percent the cost of Flash memory.



















There are currently no comments for this post.