[Deneb]

Quote:

Q2 2009 Unlike many reports all over the internet, the chip that many call RV870, the next generation ATI 40nm chip scheduled for Q2 2009 launch, won't be a multi chip module. You won't see two of these chips packed together. It will be a single chip and following tradition you can expect to see a dual chip X2 card for the high end market. We don’t have many details about the card, but we know its a single chip design, and not MCM. Graphics guys will stay away from this technology for the time being.

Source

What do you guys think?Its FUD anyway,but if its true?

 

[Diabolical999]

I think it's ALL speculation, until confirmed by a reputable source.

 

[p0isonapple]

I don't trust FUD from my left sock. :\\

 

[Deneb]

Yeah i know its just speculation since its FUD.But what do you guys think IF its true?

 

[Zonda]

FUD and INQ. I don't trust their news.

 

[ThePath]

Quote:

Originally Posted by Zonda FUD and INQ. I don't trust their news.

Me too

Both sources have history of posting many fake, untrue news

 

[003]

Quote:

Originally Posted by Zonda FUD and INQ. I don't trust their news.

Ignorance at its best. If you haven't noticed, pretty much everything they have reported on lately has been true. In fact, it has been quite a long time since an untrue story came from either of them.

To be suspicious of them is one thing, but to completely dismiss what they report on is another, and casts an aura of naivete.

 

[GymieSmallz]

All i'll say is FUD Off!

 

[dark_eternal_knight]

there kept being mention of both june release and late Q1 to early Q2 releases, now unless the quarter system is based on the british tax year, then Q2 is nowhere near june/july...

i mentioned a possibility of the first 40nm radeons simply being refreshes of the 4k series resembling a tick-tock model, alot like the 3k series being a shrunk 2k series, after that i was mainly quoted and mocked for the, somewhat logical enough idea...

as long as this does happen, and as long as the price drop from 55nm to 40nm is as extreme as the drop between 2900 and 3800 series i'll definitely be making such a purchase...

 

[Emitz989]

I think this is perfectly feasible. Why would ATI want to change a winning formula? Reducing their Fab process and increasing the SP count will improve performance dramatically - just look at the difference 38XX to 48XX. In my opinion a MCM makes no sense at the moment, they don't need to spend that sort of money on R&D yet.

 

[-iceblade^]

Quote:

Originally Posted by 003 Ignorance at its best. If you haven't noticed, pretty much everything they have reported on lately has been true. In fact, it has been quite a long time since an untrue story came from either of them. To be suspicious of them is one thing, but to completely dismiss what they report on is another, and casts an aura of naivete.

i agree.

best to at least consider what they say, and then when a more reputable source comes out with the news, be like, ah, knew that arleady

 

[dark_eternal_knight]

best way of putting it would be to say:

40nm refresh and a little tweak for ATi RV770 will be like codename penryn was for intel core 2...

 

[Cheetos316]

Quote:

Originally Posted by Emitz989 I think this is perfectly feasible. Why would ATI want to change a winning formula? Reducing their Fab process and increasing the SP count will improve performance dramatically - just look at the difference 38XX to 48XX. In my opinion a MCM makes no sense at the moment, they don't need to spend that sort of money on R&D yet.

Well... I think a MCM design would drastically reduce the cooling complexities of the card. The current X2s require multiple fans and/or multiple heatsinks that drive up the cost of the card. The PCB would probably be a bit shorter saving a tiny bit on materials cost. Plus, a MCM design would reduce latencies between the dies if/when they do start using the Sideport. MCM is probably going to be the next big thing for GPUs, so AMD might as well get a start on it now...

 

[jrharvey]

Either way, its gonna rock the house. If its anything like the rv770 it will still be awsome.

 

[MrMan]

um i dont uderstand. didnt they completely contradict themselves.

they said we wont see multichip modules but then they say we will see x2's? is that not the same thing?

 

[Licht]

AMD has been pushing multi-core GPUs since HD4K. The plans always called for it, then it wound up not getting in there. Guess we will see if HD5K gets it or not in 2 months.

 

[DuckieHo]

Quote:

Originally Posted by Cheetos316 Well... I think a MCM design would drastically reduce the cooling complexities of the card. The current X2s require multiple fans and/or multiple heatsinks that drive up the cost of the card. The PCB would probably be a bit shorter saving a tiny bit on materials cost. Plus, a MCM design would reduce latencies between the dies if/when they do start using the Sideport. MCM is probably going to be the next big thing for GPUs, so AMD might as well get a start on it now...

Cooling complexity actually increases with MCM. The dies are next to each other so they generate more heat in a smaller area. Each die affects the other so engineers have to do much more validation than two seperate chips.

Quote:

Originally Posted by MrMan um i dont uderstand. didnt they completely contradict themselves. they said we wont see multichip modules but then they say we will see x2's? is that not the same thing?

MCM is two dies on one package. Examples are the Pentium D (two single core dies) or Core 2 Quad (two dual core dies). The X2 configuration is two seperate different packages on a single PCB.