"Don't buy that new PC just yet!"

[lazzhar]

.....



What do you mean RISC wokstation race? RISC was never designed for workstations.

...

I'm not into computing but I mean SGI workstations were all based completly on RISC technoligy. They were the leaders but now after losing the race with thr dumb PC they have axed making desktop workstations. As well, apple sent it out through the window when they left the PowerPC chips to switch to intel's ones.
I may be wrong but other than simulations I don't know a time consuming applications like 3d rendering.

[lazzhar]

...

Finally, you are merely talking about performnec in Graphics, FYI I have written actual code that renders scenes....

Any hope to share it or opensource it?

[k_s]

I see, Silicon Graphics in this case uses MIPS RISC not for workstations, it is actually for their special purpose systems, which is the correct thing to do, since they only need few instructions, thus a RISC is more suitable, there is no pint using a slow CISC clock when you don`t actually use most of it`s 80% instructions expecially when those instructions that you don`t use are the ones which make the clock speed slow.

Any simulation of real world is extremely time consuming and uses often mathematical formulae.

In the case of rendering a series of transformations are appplied on a scene to simulate how we view the scene and light models are used to calculate the brightness of each point in the scene.

Take for example, infobiotics community, they too have huge data sets with a lot of processing, and simulations tend to use a lot of computation, unlike applications which are more about transfering data and networking (yes simulations often use networking but it is usually for cooperating multiple computing resources inparalel to speed up the simulations, in other words all of the computers are doing the same thing and adding the result togather at the end).

[k_s]

Any hope to share it or opensource it?

It was part of my second year lab, it`s somewhere buried in the file system, but yes I studied in depth the graphics pipeline, including advanced thechniques and data structures on how to speed it up(such as those ones used in games to make wrting interractive responsive games possible).

You don`t have to see it , as it`s obviously simpler than what`s available and may contain bugs, we just wrote it to learn and actually go through the process.

In all other cases reinventing the wheel is something engineers should never attempt to do.

[lazzhar]

...

In all other cases reinventing the wheel is something engineers should never attempt to do.

Who knows? You can come up with something more efficient than others. I remember reading about a guy who was working for a company called Nextlimit developping a Fluid Simulation application called RealFlow, the original code was all his concept but he got fired later then he decided to write something similar from the scratch and he came up with a new faster program -called glue3d- wich pushed the company to sue him but in vain because they compared both codes line by line and he has a completely different concept.
Sometimes indivuals could beat great companies like this examples and others as you certainly know.

[k_s]

You are certainly right.
When a software component is not good enough, then yes you are actually obliged to make a new one that either:

1- Is good enough to enable adequate performence of the (whole) system.
2- Beats the competition.

What I wrote was not something new, I was learning.

Reusability is very tricky, in UNIX the recommendation is to build from scratch (like you suggested) instead of reusing existing components and programs why?

First: Reusable components tend to carry a lot of assumptions about the environment (the project).

Second: Reusable components tend to have hidden side effects (which were important for the correct operation of a previous system).

For these two reasons, a software project may fall apart as a result of bugs, consequently schedule delays result and costs increase which often cause projects to be cancelled.

Reusabiliy is one of the cornerstones of engineering, if not used properly, it may cause one to blow their own foot.