Intel Shows 48-core x86 Processor as Single-chip Cloud Computer

Started by Kadri, December 02, 2009, 11:01:04 PM

Previous topic - Next topic

Zairyn Arsyn

now if nVidia or ATi would come out with a prototype video card of the same calibar of intel's 48 core prototype processor, that would be sweet. :)

WARNING! WIZARDS! DO NOT PREDICT THE BEHAVIOR OF OTTERS UNLESS YOU OBEY BIG HAPPY TOES.

i7 2600k 3.4GHZ|G.skill 16GB 1600MHZ|Asus P8P67 EVO|Evga 770GTX 4GB|SB X-FI|Antec 750W
http://zlain81.deviantart.com/

TheBlackHole

Quote from: zaai999 on December 04, 2009, 02:20:02 PM
now if nVidia or ATi would come out with a prototype video card of the same calibar of intel's 48 core prototype processor, that would be sweet. :)


You think THAT would be awesome? My nVidia GeForce card is already pretty great! Still, you can't have TOO much computing power or graphics awesomeness then, can you?
They just issued a tornado warning and said to stay away from windows. Does that mean I can't use my computer?

Cyber-Angel

Quote from: Kadri on December 04, 2009, 01:41:57 PM
Njeneb if i remember correctly did Newtek bring Lightwave to PC in 1994 or 1995 ? That was the first time i used it i think ...?

Kadri.

It was 1994 that Newtek ported Lightwave to the PC though you might recall that it was introduced by Newtek as part of the Videotoaster on the Amiga platform in 1990.  ;D

Regards to you.

Cyber-Angel

Kadri

Quote from: Cyber-Angel on December 04, 2009, 08:40:16 PM
Quote from: Kadri on December 04, 2009, 01:41:57 PM
Njeneb if i remember correctly did Newtek bring Lightwave to PC in 1994 or 1995 ? That was the first time i used it i think ...?

Kadri.

It was 1994 that Newtek ported Lightwave to the PC though you might recall that it was introduced by Newtek as part of the Videotoaster on the Amiga platform in 1990.  ;D

Regards to you.

Cyber-Angel

Thanks , Cyber-Angel  . Glad that i remember some things (so or so ) at least right  ;D

Kadri.

Henry Blewer

Cyber is correct. I bought a Video Toaster just to use Lightwave in 1990. Then Newtek came out with the PC version in 1994. They also made the Video Flyer in 1993, but it did not really hit the consumer until 1995. (?) It was about this time that PC's broke the 16 color barrier in most computers sold.
http://flickr.com/photos/njeneb/
Forget Tuesday; It's just Monday spelled with a T

Cyber-Angel

Quote from: njeneb on December 05, 2009, 08:41:51 AM
Cyber is correct. I bought a Video Toaster just to use Lightwave in 1990. Then Newtek came out with the PC version in 1994. They also made the Video Flyer in 1993, but it did not really hit the consumer until 1995. (?) It was about this time that PC's broke the 16 color barrier in most computers sold.

That's interesting, the Commodore Amiga 500 which was around at that time could use a maximum of 4069 colors,  but generally people tended to restrict them self's to a 256 color range!  ;D

Regards to you.

Cyber-Angel   

Kadri

Hmm...Amiga... I only years latter realized how good they were .

Kadri.

Henry Blewer

The 4000 could do 226,000 in a native HAM8 format. I had the Video Toaster in the 3000, and a graphics card in the 4000. Both the Toaster and the graphics card could do 16 million colors, but the Toaster was analog video. It looked really cool on the 200lb. 32"  Toshiba I had. I used a tiny 13" CRT monitor on the 4000. The Amiga 1200 had many of the same capabilities as the 4000, minus the 6 expansion slots, plus one video slot.
http://flickr.com/photos/njeneb/
Forget Tuesday; It's just Monday spelled with a T

penang

Quote from: zaai999 on December 04, 2009, 02:20:02 PMnow if nVidia or ATi would come out with a prototype video card of the same calibar of intel's 48 core prototype processor, that would be sweet. :)
Huh ???

In terms of raw data crunching power, current crop of GPUs runs circles around CPUs.

Take Nvidia.

Nvidia's upcoming 300 series is 512-bit wide. That's 8 times Intel's CPU's 64-bit, which means data can be crunched 8 times as fast.

Plus how many stream processors are there? Definitely many times the "48" cores of Intel.

Take ATI.

Ati's newest 5970 is 256-bit wide, 4 times as wide Intel's 64-bit architecture, and yes, can crunch data 4 times as fast.

Plus, the HD5970 has 1,600 stream processors. Which means, at any given point in time, there are 1,600 "cores", all working in tandem, each crunching 256-bit of data !

Can Intel's 48-core CPU even begin to compete ?

latego

About processing power: yesterday evening I was updating a dusty Java toolkit and I run an FFT bechmark applet.

In 1985, I ran my first serious 1024-samples FFT using an Apple II (8 bit micro, 1 MHz, 40 bit floating point numbers done in software, BASIC) and it took 635 second for a run.

Now, the Java 1.6 version, running on a 64 bit micro, at 3 GHz, hardware floating point, was 4.4 MILLION times fasters (140 microseconds).

And all this on ONE core.

And, at the same time, you still find legions of idiots hand optimizing obscure loops.

Bye ::)

Henry Blewer

I learned to write tight code in assembler. Memory was expensive and always scarce, so you had to write the program as small as possible. Now even the compilers generate sloppy code. But that's just me being a curmudgeon. There are some great programmers out there. I think we are all fortunate to have Matt and Terragen 2. 
http://flickr.com/photos/njeneb/
Forget Tuesday; It's just Monday spelled with a T

penang

Quote from: latego on December 06, 2009, 05:00:38 AMyou still find legions of idiots hand optimizing obscure loops.
Maybe we look idiotic to you, but do you know the difference between a highly optimized code and a so-so one?

I am no code guru, but I have experienced a 20 fold increased in speed.

If i were a real code guru, perhaps the result could be 20 X 20 folds :)

Henry Blewer

I programmed for size more than speed. I was a good programmer, but I knew people who were much better.
http://flickr.com/photos/njeneb/
Forget Tuesday; It's just Monday spelled with a T

penang

Quote from: njeneb on December 08, 2009, 10:03:25 AMI programmed for size more than speed. I was a good programmer, but I knew people who were much better.
depends on situation

on embedded projects size takes precedent

but on other projects i take performance seriously

whether it be under C or ASM i try my best to achieve 2 objectives:

1. as few bugs as possible (impossible to have total bug free programs)

2. program runs as fast as possible

FrankB

Quote from: latego on December 06, 2009, 05:00:38 AM
About processing power: yesterday evening I was updating a dusty Java toolkit and I run an FFT bechmark applet.

In 1985, I ran my first serious 1024-samples FFT using an Apple II (8 bit micro, 1 MHz, 40 bit floating point numbers done in software, BASIC) and it took 635 second for a run.

Now, the Java 1.6 version, running on a 64 bit micro, at 3 GHz, hardware floating point, was 4.4 MILLION times fasters (140 microseconds).

And all this on ONE core.

And, at the same time, you still find legions of idiots hand optimizing obscure loops.

Bye ::)

The thing is, when you want to calculate a Fourier Transformation just once, to get THE one result, then the speed increase that comes with modern hardware is just awesome and more than enough. However, let's say you render an image and have to calculate multiple FFT's for each subpixel, then yeah, hand optimzing code can have a great impact on the overall performance of the application.

There are becnmark conversations that are beyond me though. For example, the neverending speed comparisons of Safari vs Firefox vs IE, vs...
Honestly, I've tried most of them, but never felt that any of them were slow. I think the last time I realized a browser could be faster in order to be convenient, was back in year 2000 or something... Since then, the hardware power made speed differences totally irrelevant in browsers - at least for me ;-) I'm still waiting longer for servers to serve the page with all the css files and such, than would take for the browser renderer to display the page. So in a case like this... hand optimzation is indeed not relevant ;)

Cheers,
Frank