Slightly OT: what would this processor mean for editing (if it ever comes out)?

musman wrote on 2/10/2005, 6:42 PM
Kinda surprised I haven't heard anyone talking about this yet, but IBM, Sony, and Toshiba announced this week a processor supposedly 10x faster than what we have today. I don't really understand it, but apparently it has 10 cores, each of which should run at 4.5 Ghz. Sony has plans to add it o Playstation 3. But, the processor is not supposed to be out on computers until sometime in 2006 and AMD and Intel are expected to have something similar by that time. Also, AMD and Intel supposed are coming out with multicore processors later this year.
Here's a link to an article about it:

http://news.yahoo.com/news?tmpl=story&u=/pcworld/20050207/tc_pcworld/119602

I was wondering:
1) What would this kind of speed mean for us if it actually does arrive as promised?
2) What is the difference b/t multicore and hyperthreading or hypertransport?
At any rate, it does sound promising.

Comments

BillyBoy wrote on 2/10/2005, 6:53 PM
A processor 10X faster then today's fastest would mean Vegas would render in less than real time. Whoopee! Don't hold your breath such a chip with be here in just a couple years.
TheHappyFriar wrote on 2/10/2005, 7:02 PM
oh yeah, the cell processsor. One of the advantegs of gaming for a hobby is that you find this stuff out the second it gets public. That would of bee months ago.

See, the one thing that they don't mention with this is that both Microsoft & Sony didn't think much of this technology (at least enough to use it) until now. Nintendo went with IBM years ago for it's chips, like apple.

This would also require a new OS to run it. Now how often do people like to reinstall their os? :)
busterkeaton wrote on 2/10/2005, 7:10 PM
I saw that too. Does it run on Windows or does it require a new OS.
clyde2004 wrote on 2/10/2005, 7:50 PM
The heck with the 'cell' I'm holding out for a quantum computer. I know it is going to be a long wait but if it'll run Vegas .....
JohnnyRoy wrote on 2/10/2005, 8:05 PM
> Does it run on Windows or does it require a new OS

It doesn't require a new OS. They only say that it runs multiple operating systems simultaneously. I think its safe to assume that one of them is Linux (since Linux already runs on IBM PowerPC’s). Perhaps this is the Windows killer? When the fastest desktop on the block (by a magnitude of 10) is running Linux, the CPU intensive applications will surely follow. ;-) Lindows anyone?

~jr
musman wrote on 2/10/2005, 8:09 PM
I don't believe anything until it's out and proven, but the word I read was that it would be able to work on any OS. There's a discussion of it here:

http://www.geek.com/news/geeknews/2005Feb/bch20050209029060.htm

Confused if it's okay to copy and paste from another forum, but one person said there:

The Cell is not PowerPC baased. It is POWER based. There is a difference. When it says "10 instruction sequences" it means two on the central CPU and two on it's Synergistic core, each of which can perform up to four 64-bit floating-point operations per clock, for a total of 10 "operations" per clock. At 4GHz, that would yield about 8 BOPS (CPU) + 32 GFLOPS. Great for graphics and audio. The beauty of the cell processor is not it's internal core architecture, but it's ability to connect to other Cell "resources" and share their combined vector processing power over a network.
Hulk wrote on 2/10/2005, 8:48 PM
I have to agree with the above poster. These types of announcements really don't mean anything until the product is actually in the hands of users. The rest is just hype.

It seems as though the MHz race has flattened out over the last year or two.

We've just gone through a big push for parallel instruction execution in processors.

Now, with the advent of multicores we're entering the era of parallel threads.

A long, long time ago I used to do some assembly programming on my old Atari 800. The speed increase over Basic was simply not to be believed. Of course I'm comparing a very high level language to a very low level one, but the point is that it may be time for some of the heavily processor laden routines of many applications to be hand coded in assembly.

Does anybody remember SAW? Software Audio Workshop. I was mixing 8 tracks in real time on a Pentium 90 on that software in 1993. The entire application fit in about 800k and ran from the executable, no installation required. That was one very well designed piece of software. Hand coded by a very brilliant fellow by the name of Bob Lentini. He bypassed the entire Windows abstraction layer and coded in assembly.
Stonefield wrote on 2/10/2005, 9:28 PM
I agree that we may be on the forefront of the "next big thing" in computers. I remember few years ago when CPU's had flattened out at 400 - 450 Mghtz. That was when the Pentium changed to the Pentium II. ( Or was that the II to the III ? ) And as mentioned above, we seem to have leveled off at the 3.2-3.4 cpu speed. I got my eye on a 3.2 and they usually drop in price after a few weeks but it's been the same price for the last few months.

Not sure what's in store for real, but I'm sure we'll all benefit from the faster speeds. Rendering, Multi Video Track real time previews, multi opened applications.

Hmmm.....MUCH faster computers on the horizon just as HDV is becoming affordable.......good times ahead my friends.
Steve Mann wrote on 2/10/2005, 11:08 PM
The Cell Processor is none other than the application of the 20-year old theory of asynchronous processing. It will first be used in dedicated applications - most likely games - because there's no potential for incompatible applications or hardware as in the general PC. The processor will not in any way be compatible with current operating systems, but PC makers could write an emulator for current processors.
HPV wrote on 2/11/2005, 11:23 AM
Really interesting design. Could/should be a hugh deal for us "multimedia" and "HD" producers. Do a safety meeting before you read this deep overview, it'll keep your head from hurting (or exploding).
http://www.blachford.info/computer/Cells/Cell0.html

Craig H.