Going on 50
So I haven't been blogging for a long time, I was well what busy. Seems like this job of creating an OS is really getting onto me, when I started I had all the ideas
a) Implementation of Synthesis on x86.
b) Usage of an object oriented programming to implement the kernel.
By far Synthesis is the most revolutionary concept in the overall Operating Systems. Forget Microkernel even Nano Kernel. It is sort of having a virtual machine { Sriram are you listening } in the kernel itself only. Advantages are very obvious. The code can be synthesised at runtime taking into consideration the resources and the running the optimizing code. Consider
for ( int i = 0; i < N; ++i)
do_something();
There are 3 ways to execute it.
a) Everytime the loop is executed read the value of N, bad very bad.
b) Put the value of N in some register. Good but x86 has only 7 General purpose registers.
c) Take N as intermediate in the computation. This is possible because if N is not volatile we can assume that it won't change its value.
so the whole thing can be put in as
cmpl i, 55 ; N is assumed to be 55 here.
Idea number 2, Implementation of Linux in C++. A very bright idea :-). However there are issues here.
a) Noone has studied Object oriented systems in much detail till now.
b) Noone is ready to look at Linux++.
How do you change public opinions. However when I started with the thought process of the whole OS thing I realized that I really did not know much about Operating systems except that last year I did my internship had a bit to do about Operating Systems research. Eventhough my university never had a course in Operating Systems
Linux++ yeah sure, but I am a patient guy{50 lines of code is bad very bad I know}, thinking big is quite good but just like Abraham Lincoln said
"If I am given 6 hours to chop a tree, I will spend the first 4 sharpening my axe"
And O we had our farewell yesterday. Nice party they tell me. But first VM.
1 Comments:
And I forgot to mention about
6.828
http://www.pdos.lcs.mit.edu/6.828/2004/tools.html
and synthesis
http://citeseer.ist.psu.edu/massalin92synthesi.html
4:34 AM, February 26, 2005
Post a Comment
<< Home