RV870 single-core, R870 dual-core, R800 quad-core?

[Macadamia]

Whether or not this is true, it would make sense to do, right? It would save on PCB costs significantly.

Keyword is probably PCB length instead of costs, though.


I think this would make it up nicely in a dualcore situation.

------------------------------------------------------------------------
GPU1 <-> GPU 1 Sideport Interface <-> GPU2 Sideport Interface <-> GPU2
------------------------------------------------------------------------
|
|
|
V
PCIe 2.0 x16

Aka GPU1's sideport acts just like the PLX chip (on the 4870X2 they act like extra traces)

If they have a crossbar approach it would be even more interesting to watch.

Top boss of Chiphell (pretty much always accurate) is implying that memory will be shared between the MCMs.

[orangekiwii]

lol this thread is funny

[russian boy]

4 cores = 4 Frames AFR input lag+microstuttering = good for benchers, bad for everyone else

[Final8ty]

4 cores = 4 Frames AFR input lag+microstuttering = good for benchers, bad for everyone else

Look at sig..i'm not complaining my arrangement with old tech 3870 is meant to have the worse lag+microstuttering going.

[Mr.BSEL]

According to this website, the rumors of 4 gpus on one card are false.

http://translate.google.de/translate...istory_state0=

[Calmatory]

According to this website, the rumors of 4 gpus on one card are false.

http://translate.google.de/translate...istory_state0=

Quite obvious.

[gamervivek]

4 cores = 4 Frames AFR input lag+microstuttering = good for benchers, bad for everyone else

rage3d did a 4870x2 CF review where they pointed out that having more gpu horsepower can actually decrease chances of MS,
http://www.rage3d.com/reviews/video/.../index.php?p=5

[NH|Delph1]

I'm not saying it's true, I'm saying that several not related sources are once again saying the same thing. So far I'm up to three.

I thought I was pretty clear on that...

I'm waiting for more information soon, and I will post a follow-up. I'm not going to let this circulate for long if it's not true. I'm skeptic myself, but like I said, I just wanted to forward the "information".

In the meantime:


EDIT:
As mentioned elsewhere, Lil' Dragon looks to be a rather straight forward upgrade from RV770, but knowing AMD/ATI they have brought some surprises in the past. Just that they managed to keep RV790 a secret for long, drawing all attention to RV740 instead was quite a trick. Not to mention the previously mentioned 480/800sp debacle with RV770.

EDIT2:

I checked around and after stumbling across a couple of sites I decided that this just isn't worth it. I try to forward what I hear, but in return I get ...

Due to misinterpretations and since we don't want to be pointed out as the source of misinformation that might distort people's image of what is about to come we've decided to pull this story. We will post an update on Lil Dragon, RV870, and its siblings in the future, but right now we're feeling a bit discouraged to do so.

Until next time,
NordicHardware Crew

STORY pulled

Sorry guys.

//Andreas

[orangekiwii]

interesting

i'm just interested in the time frame as i need a new gpu soon

[Helmore]

interesting

i'm just interested in the time frame as i need a new gpu soon

My guess would be, April for ATI's first 40 nm. chips and Q4 for RV870 (or whatever their first DX11 chip will be).

[spursindonesia]

So, for highend card, the roadmap is like this:

RV770 --> RV 790 (lil dragon ?) --> RV870 (DX 11 capable ?)

Quite probable ?

[Shintai]

So, for highend card, the roadmap is like this:

RV770 --> RV 790 (lil dragon ?) --> RV870 (DX 11 capable ?)

Quite probable ?

yep

[ColonelCain]

@ the chops!

As far as RV790, have we yet figured out anything else about it?

[Mechromancer]

Photoshop ftw! And the extra blur is a nice touch

Here is another bunch of photoshops.








Ye and those 100-200GB/sec external busses is lovely too and 512bit MCs.

Anybody notice those pics are signed "CONCEPT BY ILLUMINATI"? I do believe these are very FAKE. The idea looks very feasable and is convincing. Technically I think it can be done, but I doubt they would be very power efficient.

[renzokuken]

There isn't a salt mine big enough take with this, but we just wanted to notify you that the rumor has resurfaced. Beware.

Lol, this is a rofl pop.

[STaRGaZeR]

Anybody notice those pics are signed "CONCEPT BY ILLUMINATI"? I do believe these are very FAKE. The idea looks very feasable and is convincing. Technically I think it can be done, but I doubt they would be very power efficient.

[orangekiwii]

any ideas if nvidia is releasing its gt212 in march to compete with DA LIL DRAGON

[NH|Delph1]

So, for highend card, the roadmap is like this:

RV770 --> RV 790 (lil dragon ?) --> RV870 (DX 11 capable ?)

Quite probable ?

Don't mix up the code-names. RV870 = Lil Dragon.

//Andreas

[lm358]

Rumour seems very plausible to me. Intel used the same concept successfully with dual-die single package chips like Kentsfield.

It makes very good sense from a cost perspective and in terms of both die profits and yields, and also PCB complexity.

And ATI have already stated that multi-GPU scalability is where they are heading long term.

With clever drivers/compilation techniques, and/or possibly new memory controller techniques it seems like a natural progression from their current work.

[spursindonesia]

OK, expanding the roadmap by including competing products from nVidia and ATi dual GPU solution, will it look like it ?

GT200 --> RV770 & R700 --> GT200b (GT206 ?) --> RV790 & R7xx --> GT212 --> RV870 & R800 --> GT300

The timeframe seems quite coherent. With GT212, i think nVidia has a chance for a dual GPU card too, considering GT212's less than 300 mm^2 die size area.

[Hornet331]

Rumour seems very plausible to me. Intel used the same concept successfully with dual-die single package chips like Kentsfield.

Theres quite a difference between in the bandwidth requierement in syncing 2 cpus and 2 gpus. I doubt we see the mystical "shared memory" the next few years. Not without a drastic change in architecture form ati/nv.

[Helmore]

Theres quite a difference between in the bandwidth requierement in syncing 2 cpus and 2 gpus. I doubt we see the mystical "shared memory" the next few years. Not without a drastic change in architecture form ati/nv.

Current rumors put RV870 as an evolution of RV770, which is an evolution of R600. Where as the next generation after RV870, which we'll just call RV970, will be more of a revolution and a much more drastic change from the way ATI's current GPUs work. All just rumors ofcourse, but still interesting to know.

[ColonelCain]

Current rumors put RV870 as an evolution of RV770, which is an evolution of R600. Where as the next generation after RV870, which we'll just call RV970, will be more of a revolution and a much more drastic change from the way ATI's current GPUs work. All just rumors ofcourse, but still interesting to know.

Definatly interesting to know. But then again, everything is indeed just rumors, we just attempt to figure out which have ground, and which are just plain FUD. Like the 'chops by illuminati.

[Macadamia]

740 and 790 should arrive in March.


At least 740 will from what I've heard...

[Calmatory]

???->R300->R600->R900->???. :-)

What could RV870 be, if RV790 is 960 shaders and higher clocks @ 40nm? 1280 shaders with high clocks? Doesn't sound too big improvement over RV790.