AMD + ATI

[miriam_e]

"We may lose business on Intel boards, but we will break the Intel monopoly." With these words, AMD's CFO Bob Rivet announced the takeover of graphics chip maker, ATI, offering a future of joined-up shared processing, split between CPU and GPU.

The deal, announced today, goes back some time. Last year, at Computex in Taipei, it was apparent that ATI and AMD were falling in love with the idea of using the powerful graphics processor to run computer programs, not just for animating video.

Read more at http://www.theregister.co.uk/2006/07/24/amd_etc/

Comments

[damien-wise.livejournal.com]

It'll be interesting to see how this merger pans-out.

I think it makes sense for them to integrate their CPUs/GPUs more closely. They could really get a speed-improvement by coupling the graphics card and motherboard design...bandwidth between processor, main memory and graphics subsystem is now the bottleneck, and one which will only get worse unless someone overhauls the PC.

That's perhaps a medium-term view.
In the short-term, it could benefit them to share technology, manufacturing plants/tools, etc. to lower costs and further weaken Intel's undeserved monopoly (and compete with Nvidia).

What saddens me is that the Alpha was years ahead of its time, yet those who got the designers when Digital was split-up/taken-over didn't capitalise on the treasures they had.
I heard a rumour not so long ago that Google had been head-hunting those same Alpha people to develop customised processors/hardware.

[miriam-e.livejournal.com]

Wow! Google are taking this diversification thing seriously.

I didn't realise the Alpha team was broken up. It would be wonderful if they let them run with cool ideas again.

I feel pretty strongly that parallel design is where the future lies. Using different processors for different jobs is key to that. I hope that AMD + ATI works more in that direction. (I wonder what they'll call themselves, or if they'll simply keep the old names but effectively merge business and knowledge.)

I saw a talk a while back where a guy "showed" that a single fast processor would always be more efficient than multiple processors, mostly because of communication overheads, but he ignored what happens when it becomes too hard or too expensive or too energy-costly to build a faster processor. You can always add more processors, but you don't always have a faster processor available. The human brain is a showcase for massive parallel processing using trillions of incredibly simple, terribly slow processors. A neuron has a clock rate of about 10 cycles per second(!!!), yet we can pattern match absurdly complex images in a split second.