Why use SGI for video editing/effects

[Rockazella]

As a hobby I do lots of video editing/effects type stuff. Recently I have started to look for a new beefier system. I've been scanning through video forums searching for hardware recommendations. I came across one thread about SGI's stuff and how it's tops in speed when it comes to graphics work.

Although SGI's workstations are way way way out of my budget, I am just curious what's with their hardware layout that makes them so superior to PCs and Macs when it comes to graphics?

 

[dduardo]

Because SGI uses dual 64bit processors and a proprietary graphics card with the IRIX OS (64 bit Unix) and graphics software that actually takes advantage of the system. They also have 8 Gigs of Ram.

Not only is the hardware expensive but 64 bit video processing apps are equally expensive.

Intel and apple doesn't make consumer products with hardware similiiar to SGI because consumers can't afford it.

Even though apple and amd are marketing 64bit processors to consumers, they aren't true 64 bit systems as the software can't take advantage of the hardware

 

[Rockazella]

Even though apple and amd are marketing 64bit processors to consumers, they aren't true 64 bit systems as the software can't take advantage of the hardware

Aren't true 64bit? how can a processor be 64bit, but not true 64bit?

 

[dduardo]

by "true 64 bit systems", I am talking about hardware and software. Yes the processor can run in 64bit mode, but the software isn't, therefore the system isn't true 64bit. It is a 64bit processor running 32bit code in emulation mode.

 

[russ_watters]

Originally posted by dduardo
by "true 64 bit systems", I am talking about hardware and software. Yes the processor can run in 64bit mode, but the software isn't, therefore the system isn't true 64bit. It is a 64bit processor running 32bit code in emulation mode.

Slight clarification: older apple's advertised even 128 bit. The AMD Athlon 64 is a 32/64 bit proc in that it runs both equally well. MS is working on the operating system so for now it will only run in 32 bit mode. Intel has the Itanium which runs 64 bit and emulates 32 bit (badly). It runs 64 bit Unix software right now. Apple's new G5 is 64 bit (I think).

 

[Rockazella]

I don't know too much about these processy gizmos, or about any computer hardware really. For clarification, a 64bit processor is one that can process 64bits at a time, or per clock cycle? Also, you say that the OS needs to be specifically designed for the bit number of the proc...I'm a bit confused about that...or maybe it's just that I wasn't aware of it before. Ah whatever, bet there's plenty of things I don't know about computers.

One more question:
Can any computer process a file of any bit depth? I know 8 bits is a byte, and a byte (and multiples of a byte) hold some significance in computers. Yet some video(true HiDef) is 10bit. Since 10bit isn't a multiple of 8 do you need special hardware to process it, or is it just the software that needs to be designed for the files bit depth?

 

[russ_watters]

The best way to explain it is in terms of numerical precision and memory addressing (though there is more to it than that). A 32 bit proc processes data in 32 bit chunks. 32 bits is 2^32 or 4,294,967,296 different combinations or a numerical precision of 1/4,294,967,296. It also means that if you want to catalog something, its label has that many combinations. 4,294,967,296 bytes is 4 gigabytes - the current limit in a PC's memory capacity.

So if you kick it up to 64 bits, now you have 18,446,744,073,709,551,616 combinations. The precision of your calculations is far higher and now you can address significantly more memory.

 

[Rockazella]

The best way to explain it is in terms of numerical precision and memory addressing (though there is more to it than that). A 32 bit proc processes data in 32 bit chunks. 32 bits is 2^32 or 4,294,967,296 different combinations or a numerical precision of 1/4,294,967,296. It also means that if you want to catalog something, its label has that many combinations. 4,294,967,296 bytes is 4 gigabytes - the current limit in a PC's memory capacity.

So if you kick it up to 64 bits, now you have 18,446,744,073,709,551,616 combinations. The precision of your calculations is far higher and now you can address significantly more memory.

numerical precision? memory adressing? memory label combinations?
Sorry, I've never really been given a good explanation of the processes of processing before, and I think I'd need one before that would make much sense.

Was that intended to answere my second question as well? I'll ask it again anyway:

Can any computer process a file of any bit depth? I know 8 bits is a byte, and a byte (and multiples of a byte) hold some significance in computers. Yet some video(true HiDef) is 10bit. Since 10bit isn't a multiple of 8 do you need special hardware to process it, or is it just the software that needs to be designed for the files bit depth?

Also,
Russ, dduardo, or anyone...can you recommend a good site or maybe a book that could thoroughly teach me the basics of this stuff?

 

[dduardo]

Rockazella, try this website:

http://www.arstechnica.com/cpu/03q1/x86-64/x86-64-1.html

 

[russ_watters]

"memory addressing" is exactly what it sounds like - its the address (location) of a specific piece of data in the computer's memory. Just like your address is the specific location of your house in your neighborhood. As I showed with the math, there are only enough specific addresses to use 4GB of memory using 32 bit addressing.

Numerical precision is just decimal places: 1.00 is ten times more precise than 1.0.

Not too tough, right?

Can any computer process a file of any bit depth?

Yes. Essentially (simple answer) its just a software issue.