what happened to that amd fusion thing?

[MODEL3]

Originally posted by: Idontcare
I don't know if any of it has to do with problems of any sort...could just be that until recently it simply made no sense to create a fusion-like product as it would have had little-to-no net benefits to the end-user and little-to-no gross margin benefits to the manufacturer.

Yeah, i also see it that way...


I meant:

Or is it pure design choice/roadmap execution related? (i wrote problem)

 

[maddoctor]

Don't say AMD anymore, they will be dead soon.

 

[Idontcare]

Originally posted by: maddoctor
Don't say AMD anymore, they will be dead soon.

Interesting choice for your first post to the forums...

Welcome?

 

[DrMrLordX]

Originally posted by: IlllI
memory is foggy but i think i recall reading about it a couple years ago.. it was suppose to combine a gpu and cpu onto one chip, i guess for laptops etc.

There was a bit more to Fusion than just that.

AMD had (and may still have) plans to expand the functionality of Hypertransport so that GPUs could be plugged into sockets common to both CPUs and GPUs supported by any given HT3.0-capable platform. An alternative/complement to this aspect of Fusion was to offer HTX slots on Fusion/HT3.0 motherboards that would allow full cards to be plugged into an HT link to communicate at the same level of speed and priority as a CPU in a multi-CPU HT-based system.

The whole idea was to use HT to simplify the process of sharing computational resources and memory between anything utilizing HT and to eliminate additional busses (PCI-E) to reduce latency and so forth and so on.

So, if you know anything about HT (I must confess that my knowledge only scratches the surface), you'd know that trying to cram a GPU and a CPU into one package on an AMD platform isn't that far off from having a dual-socket AMD board with a CPU in one socket and a GPU in another (or a single-socket AMD board with an HTX slot hosting a CPU in the socket and a HT-aware vid card in the HTX slot) thanks to Hypertransport. There's hops involved which can change performance, but it is not an insurmountable barrier to functionality.

AMD's ability to implement CPU/GPU hybrids in a single package, or on a single die, isn't the only factor involved with Fusion. Now that they own ATI, they should have (theoretically) been able to offer HTX video cards that would utilize HT to communicate with processors on HT3.0 motherboards. For some reason, this aspect of Fusion has not yet materialized. It may be that AMD fears that offering HTX versions of their cards would involve a great deal of R&D/implementation expense to produce a product that, at the outset, would fill a niche market since HTX would most likely not be implemented on Intel platforms (meaning if they wanted to sell HTX version of their cards, they would not be able to sell them to anyone but those using the latest and greatest AMD systems). They would still have to sell PCI-E cards to those using older AMD platforms and Intel-based platforms to continue producing revenues from graphics card sales. That kind of product redundancy would be injurious to their overall financial situation, I would think.

So, the whole idea of CPU/GPU hybrids was more of the pie-in-the-sky, "we'll do it someday" aspect of Fusion when it was first announced (or at least, that was my understanding). We were supposed to see HT3.0 systems utilizing HTX or GPU-in-a-socket before that. Or, at least, that's the way I remember it.

Now Fusion is basically being pushed off until AMD can implement CPU/GPU hybrids. Apparently the financial outlay on doing that is better than trying to bring a non-standard video card slot to market.

 

[Idontcare]

Originally posted by: MODEL3

Originally posted by: Idontcare
I don't know if any of it has to do with problems of any sort...could just be that until recently it simply made no sense to create a fusion-like product as it would have had little-to-no net benefits to the end-user and little-to-no gross margin benefits to the manufacturer.

Yeah, i also see it that way...


I meant:

Or is it pure design choice/roadmap execution related? (i wrote problem)

Yeah, could also be just simple case of not enough R&D dollars to go around for funding everything AMD executives had on the table at the time.

Money spent on paying salaries for designers working on fusion in 2007 would have meant less money to spend on salaries for designers working on Shanghai and Istanbul.

 

[MODEL3]

http://en.expreview.com/2009/0...ssor-specs-leaked.html

http://www.fudzilla.com/content/view/15354/1/">http://www.fudzilla.com/content/view/15354/1/</a>

http://www.fudzilla.com/content/view/15364/1/

 

[IlllI]

for some reason i was reminded of a slide that showed that there will be several different configurations to appeal to different audiences. i managed to find that slide Text

i had completely forgotten about this until reading all these replies

 

[Idontcare]

Originally posted by: IlllI
i managed to find that slide Text

Ah yes, back when AMD was "smarter choice"...now they are "The Future is Fusion". So much progress in so little time.