RAM FOR FREE…
Lei Yang, Robert P. Dick, Haris Lekatsas, and Srimat Chandrakar
Memory, often the Random access memory (RAM), forms one of the major components of the computer. It, to an extent, defines the performance of the computer. In the early seventies, computers used a RAM which had a capacity which was a ten thousandth of what we use today. The technology is zooming in a lane without an impasse. Advancements in VLSI have led to the miniaturization of electronic circuits. We use computers in every nook and corner of our daily activities, for instance mobile phones, PDAs, microwave ovens and also elevators. The main point of focus made in the given article “RAM for Free” by Lei Yang, Robert P. Dick, Haris Lekatsas, and Srimat Chandrakar is on the re-usability of the RAM by using algorithms for compressing data and vice-versa.
With the advent of fabrication technology, we have devices whose sizes are of the order of micro-meters or even nanometre. The drastic fall in the costs of memory devices haven’t been of great help in the overall cost reduction of a machine .This is mainly due to tremendous increase in the demand for faster memory by the newer software. In this article, a possible solution is provided. Data compression concept has been the author’s core idea. Different algorithms, problems, solution and finally its application are discussed.
The major parameters considered in the design of the embedded system were taken to be power consumption and response time. RAM is fast, has relatively a fractional response time when compared to the secondary memory. The author in this article mainly deals with a method that enhances the memory utilization by using the Lempel-Ziv-Oberhumer algorithm (LZO). The algorithm selection was the initial step, then a method had to be devised for decompressing the compressed data in the memory.
The author found success in the above mentioned compression technique when it was implemented over a PDA’s Linux operating system. Here the data was compressed and stored on RAM itself. By doing so, a fraction of the RAM was only used up. This saved a lot of memory and remaining portion could be effectively utilized for other software allocation. The authors came up with another algorithm which was intended to improve the speed and increase the degree of compression. This algorithm was similar to the one invented by Jon Louis Bentley of
The modified algorithm used a technique similar to Morse code, i.e; it used shorter codes for frequently occurring code and longer ones for infrequent ones. To identify the patterns for different codes a dictionary was built dynamically. This was updated with changes in the more frequently occurring patterns. The compressed data was the main source for the dictionary. Though the algorithm didn’t beat out other algorithms ,as it couldn’t compress certain files like imaging files, it could still compress data up to 60%, and the operational speed wasn’t affected much. Hence an embedded system could save up half its memory or in other words we have simply added an additional RAM without practically adding one for free.
In January 2007, the author mentions that they had applied for patent and named their technique as CRAMES, short for Compressed RAM for Embedded System. CRAMES faced a few initial hiccups in identifying the compressed data’s source, but the authors together solved it by keeping a record of the compressed data and application to which it belonged. CRAMES was tested on a mobile phone board. It worked well even after data compression.
Initially the authors had the idea of implementing their idea (CRAMES) only on embedded systems, but work was also done to implement CRAMES on general purpose computers. As today we see that computers are designed with more than a processor at cost of lesser cache sizes. So CRAMES could be a possible solution for this problem. The use of CRAM concept would not only enhance memory usage but also reduce the expense of adding an additional RAM.
In my opinion, this concept would drastically impact the ongoing technological advancements in a positive nature. I appreciate the concept of data compression which has led to economical use of memory. The first major advantage is the cost reduction. Secondly, we are able to store more data especially in smaller devices like PDA, mobile phones, MP3 players...Etc. Lastly, credit should be given to the contributors who had devised algorithms like LZO and the pattern matching algorithms designed at BELL labs, as these algorithms were the stepping stones for CRAMES.
Posts
No comments:
Post a Comment