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

From the well-informed site from North : Nordic Hardware

Quote:

---

The story of a dual-die GPU from AMD/ATI has been circulating since way before the launch of R700, Radeon HD 4870X2. While it did allow many to dream, it turned out to be just a rumor, or maybe it was just to soon? Either way a source has hinted that the new card R870 will be two GPU dies glued together sharing a memory bus. This is pretty much the exact same that was said before the R700 launch, and while we can't confirm this, similar information has been posted over at Chiphell.

Both are basically saying that AMD has glued two RV870 GPUs together to build world's first dual-die GPU, which is kind of amusing considering AMD's bashing of Intel's dual-die solutions. Anyhow, this means that AMD would then be able to build what many would come to call quad-fire on a stick. Four GPU dies on the same card, R800. It also means that for the first time we will have a middle between the RVx70 and Rx00.

Specification wise this would land us on something like 1000-1200 shaders and 50-60 TMUs with RV870, 2000-2400 shaders and 100-120 TMUs with R870, and 4000-4800 shaders and 200-240 TMUs with R800. In theory (IN THEORY!), AMD could launch them as Radeon HD 5870, Radeon HD 5870 X2 and Radeon HD 5870 X4.

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

---

Source

Good newz, :yepp:

W&S

If it's true, I'd like to see how CrossFireX scales, and if it'll support 8 GPUs (such as dual-HD5870X2, quad-HD4870X2). Other than that, I'm curious to the packaging, and I assume there'll still be a secondary chip to handle GPU-to-GPU communication. My only other question is how will memory be split? Will both have access to the full amount? Or will access be limited, such as current 2x1GB cards?

Uhmm...no, no and no for the nTh time.

Its pretty bad to make random forum posts into news. Do I even have to mention the 900 shaders in RV770 "news" that was from XS speculation/random fun. (That person actually believes the 900SP RV790 we madeup here at XS).

It would also mean you would need 4GB memory on a "quadcore" to have 1GB.

Quote:

---

In fact everyone thinks that R800 is the adoption of two dual-core RV870 the glue, I suddenly thought of AMD can make RV870 have the 1000-1200 stream processor 40NM ultra-small core, defined as the RV870, and the two high-level RV870 glue, causing R870, in fact, if two of the 1000 SP glue together RV870 will not vary much from the area, and can be shared memory, when a single card to sell ... ..., while the R800 is the R870 the two versions of X2 is, in fact, has four RV870 the RV870X4 ... ... every two RV870 composed of a memory can be shared R870, R870 with two similar R700 via the Internet to form the first internal CF ... ...

---

Hahaha, what a bunch of FUD. Its utter homemade crap. And the poster basicly try to shoehorn it in as people ask questions. 100% homemade!

Quote:

---

Suddenly have the feeling ... ... RV870 is two 40NM glue version of the RV790 ... ...

RV770 800SP 55NM ——> RV790 1000SP 55NM ——> RV870 (40NM RV790X2) 1000X2=2000SP RV770 800SP 55NM -> RV790 1000SP 55NM -> RV870 (40NM RV790X2) 1000X2 = 2000SP

---

Quote:

---

March-April release RV790 If 900SP, the performance of most high RV770 15% -20%, right?

---

Not sure why nordichardware pulls of such a dumb stunt. Even tho Andreas write alot of salt and theory. This just belongs nowhere.

so a single gpu card is like the low end... if they don't fix microstuttering & co i really hope this won't happen anytime soon.

I like this part of the article:

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

Well dual-die single sub would be easier to fab than trying to make a native dual-die (how would that work for GPUs, given theres hundreds of "cores" as it is?!, but thermal dissipation density could be horrible.

be awesome if it was true but i don't believe it at all

and if it does turn out to be true i'll be like AW SH*T WHYD I GET THIS GTX 200 CARD

bah ATI can't get their drivers right.

it took how many years for nvidia to get sli multi monitors to work no one is perfect stop while your ahead

Quote:

---

Originally Posted by safan80

bah ATI can't get their drivers right.

---

CF drivers? Never.

deja vu: RV770 single-core, R770 dual-core, R700 quad-core

R7xx was meant the same..

will it happen?.. eventually.. R8xx aint no guarantee

its about time cf/cfx moves @ die

Quote:

---

Originally Posted by STaRGaZeR

CF drivers? Never.

---

and opengl drivers

:shrug:

Quote:

---

Originally Posted by Shintai

It would also mean you would need 4GB memory on a "quadcore" to have 1GB.

---

Shared bus by dual-cores. Need 2GB memory on a "quadcore" to have 1GB.

@ stevil

nice.. but still micro stuttering

if cf/cfx moved @ die.. maybe just maybe ms be gone

Considering that IF RV790 exists as a 40nm part it will be a hell of a scaling up with 1200+ SPs already...


Why would RV870 take so much hassle in scaling instead? Make a 790-ish sized part (250-300mm^2), X2 it, no big worries.

very interesting pic.......

Quote:

---

Originally Posted by STEvil

:shrug:

---

a very skilled photoshopper....right? :eek:

@stevil, cooling that would be a nightmare.

I don't believe amd will make a dual core gpu anytime soon, why would they? I think it's more likely that they will just scale everything of the current gpu's up a bit.

Multiple dies on graphics cards are starting to sound like the rumours surrounding an integrated large ammount of EDRAM in ATI's GPUs a few years ago. Ever since the 9700 Pro came out there were rumours about ATI's next GPU with 10MB EDRAM. That proved too complicated for the "mess" that is the PC platform but eventually it came out in a console and was quite successfull.

My guess is that two GPUs sharing the same memory pool could be too dificult to get into a PC, but could very well be in the next XBox or Playstation (I won't say Nintendo because they're obviously not aiming for the "best tech" battle).

Quote:

---

Originally Posted by NapalmV5

@ stevil

nice.. but still micro stuttering

---

:shakes::shakes::shakes:

for the Nth time the 4870x2 has no micro stuttering.

4 chips (3 frames in queue) would give you LESS possible stuttering than 2, actually.


See Rage3D's investigations on it.

^ you mean 4 cores..

Quote:

---

Originally Posted by safan80

:shakes::shakes::shakes:

for the Nth time the 4870x2 has no micro stuttering.

---

:rofl:

then wth am i doing @ single gpu cards..

Quote:

---

Originally Posted by STEvil

:shrug:

---

Photoshop ftw! And the extra blur is a nice touch :p:

Here is another bunch of photoshops.

http://i114.photobucket.com/albums/n...870Concept.png

http://i114.photobucket.com/albums/n...5000Series.png

Quote:

---

Originally Posted by v_rr

Shared bus by dual-cores. Need 2GB memory on a "quadcore" to have 1GB.

---

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

Quote:

---

Originally Posted by safan80

bah ATI can't get their drivers right.

---

Flamebait. :)

^ lol not chips.. cores

Maybe if they ditch the PLX bridge chips and put a Hydra on there instead.. Outside of that it'll be utterly useless.

Quote:

---

Originally Posted by safan80

and opengl drivers

---

Does that happen with CF disabled?

Quote:

---

Originally Posted by STEvil

:shrug:

---

The blur ruins your Photoshop skills :(

Shouldn't we be worried that our salt supply won't last forever if we go on like this? I don't think this rumor has any credibility, although I have no idea what to expect from RV870 to be honest.

Quote:

---

Originally Posted by Helmore

Shouldn't we be worried that our salt supply won't last forever if we go on like this? I don't think this rumor has any credibility, although I have no idea what to expect from RV870 to be honest.

---

But they give more clicks to FUD sites!!!1!

Sorry NH but if this is a completely infundated rumor, just don't post it.

Quote:

---

Originally Posted by STaRGaZeR

But they give more clicks to FUD sites!!!1!

Sorry NH but if this is a completely infundated rumor, just don't post it.

---

...just as 480/800 SPs were.

Quote:

---

Originally Posted by RaZz!

so a single gpu card is like the low end... if they don't fix microstuttering & co i really hope this won't happen anytime soon.

---

Exactly my thoughts.

Quote:

---

Originally Posted by Helmore

although I have no idea what to expect from RV870 to be honest.

---

I would expect something like 256 or higher MC depending on GDDR5 speeds. 800-1200shaders. (Maybe 960-1200 after RV790). Else just improved tweaked parts inside for new SM, overall DX11 and OpenCL support. Perhaps higher clocks too.

Then the usual type of cards as we have today. Possibly tho with no PLX chip. But that require AMD solve the issue did long ago with 2 GPUs directly on a PCIe bus. Sideport could also be there if they didnt abandon that idea again.

Quote:

---

Originally Posted by STaRGaZeR

Does that happen with CF disabled?



The blur ruins your Photoshop skills :(

---

I did up some without blur, just tossed that in for "bad camera pic" effect to try and get it onto FUD or something a while ago :D

As to glued cores it will happen eventually.

Quote:

---

Originally Posted by Shintai

I would expect something like 256 or higher MC depending on GDDR5 speeds. 800-1200shaders. (Maybe 960-1200 after RV790). Else just improved tweaked parts inside for new SM, overall DX11 and OpenCL support. Perhaps higher clocks too.

Then the usual type of cards as we have today. Possibly tho with no PLX chip. But that require AMD solve the issue did long ago with 2 GPUs directly on a PCIe bus. Sideport could also be there if they didnt abandon that idea again.

---

I'd say that it will be at least 1200 shaders, if they don't alter them too much from the way they currently are. I don't think they will go for a higher than 256 bit memory controller and clocks may not even change that much. ATI will probably put more emphasis on idle power consumption and try to outdo NVIDIA on that front and they should thus have lower idle draw than the current GT200b (talking about ATI's top card that is).
But what I meant to say is, will RV870 be more of a revolution or more of an evolution? RV770 is close to being a revolution, but I still consider it to be more of an evolution as it is mainly a major refinement. So will RV870 be a bigger change than RV770 is or not? I'm leaning towards the former, but we have no idea which it will be.

That would be called the R800 - Spontaneous Combustion Edition

Quote:

---

Originally Posted by waver_01

a very skilled photoshopper....right? :eek:

---

I can do that with MS PAINT hehehe

Quote:

---

Originally Posted by GAR

I can do that with MS PAINT hehehe

---

show me cores side by side then and not inline gpu to gpu to the exhaust.
p.s put them close the fan area too

quad-core R800 :D

http://img403.imageshack.us/img403/9...orer800kb1.jpg

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

Quote:

---

Originally Posted by NapalmV5

quad-core R800 :D

http://img403.imageshack.us/img403/9...orer800kb1.jpg

---

:up::rofl::ROTF:

Quote:

---

Originally Posted by NapalmV5

quad-core R800 :D

http://img403.imageshack.us/img403/9...orer800kb1.jpg

---

Ultimate gaming card. Video card + Built-in CPU. :)

Quote:

---

Originally Posted by clayton

Ultimate gaming card. Video card + Built-in CPU. :)

---

It's coming...

Nvidia already has 240 cores, ATI doesn't have a chance with 2 or 4, so the news is obviously fake, I am sure they will use more than that... they better prepare for the can of whoopass.

ha, fanboi
ati has 160 cores capable of 5 simultaneous calculations (per gpu), if you want to count that way

nice photoshops btw.

though i dont see a reason for amd to go with 4 gpu's on a pcb, each with a different package

Quote:

---

Originally Posted by Kuroimaho

Nvidia already has 240 cores, ATI doesn't have a chance with 2 or 4, so the news is obviously fake, I am sure they will use more than that... they better prepare for the can of whoopass.

---

Thats sarcasm, right?

Quote:

---

Originally Posted by Cybercat

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

---

Especially when you think that because GPUs already have a parallel design the scaling would be good if the cores could access the same RAM. But as some already mentioned as such rumours surfaced before RV770 this would require quite a radical redesign.

Quote:

---

Originally Posted by clonez

though i dont see a reason for amd to go with 4 gpu's on a pcb, each with a different package

---

I gather the rumor referred in OP goes like this:
1 die = "HD5870"
2 dies on 1 chip package = "HD5870 X2"
4 dies on 2 chip packages = "HD5870 X4"

I guess it could be the single die cards would use 256bit GDDR5 of cheaper (low speed bin) RAM chips, but each die in both the multi-die MCMs have 128bit GDDR5 of premium grade (high speed bin) high density chips? So no proverbial shared memory, just dual-die MCM with one 128bit bus for each die.
:shrug:

And a Lucid chip on-board the X4...

Quote:

---

Originally Posted by STEvil

Thats sarcasm, right?

---

I thought the last part would give it away but that some took it seriously shows how dire the situation really is these days :rofl:.

Quote:

---

Originally Posted by Cybercat

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.

lol this thread is funny

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

Quote:

---

Originally Posted by russian boy

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

---

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

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

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

Quote:

---

Originally Posted by Mr.BSEL

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

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

---

Quite obvious.

Quote:

---

Originally Posted by russian boy

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

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:
http://www.firstscience.com/home/ima.../salt_mine.jpg

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 ...

Quote:

---

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

interesting

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

Quote:

---

Originally Posted by orangekiwii

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).

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

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

Quite probable ?

Quote:

---

Originally Posted by spursindonesia

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

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

Quite probable ?

---

yep

:rofl: @ the chops! :rofl:

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

Quote:

---

Originally Posted by Shintai

Photoshop ftw! And the extra blur is a nice touch :p:

Here is another bunch of photoshops.

http://i114.photobucket.com/albums/n...870Concept.png

http://i114.photobucket.com/albums/n...5000Series.png




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.

Quote:

---

Originally Posted by Olivon

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.

Quote:

---

Originally Posted by Mechromancer

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.

---

:lastweek:

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

Quote:

---

Originally Posted by spursindonesia

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

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.

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.

Quote:

---

Originally Posted by lm358

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.

Quote:

---

Originally Posted by Hornet331

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.

Quote:

---

Originally Posted by Helmore

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. :ROTF:

740 and 790 should arrive in March.


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

???->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. :shrug:

Quote:

---

Originally Posted by 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. :shrug:

---

Is it even confirmed that RV790 has 960 shaders? I wouldn't be surprised if RV770's replacement has more than that.

I'm expecting either a lot more, or none (aka speed bump).

At least 1200... for 790. ATI's shaders are small enough, and if they won't X2 this particular chip for now (maybe after shrink) they can have a lot of space to play.

Quote:

---

Originally Posted by Helmore

Is it even confirmed that RV790 has 960 shaders? I wouldn't be surprised if RV770's replacement has more than that.

---

No, but that would make some sense. If the pads aren't being shrunk, then anything below the current size would make the chip pad limited. 40 nm parts are 25 % smaller, 800 -> 960 is the closest match which would increase the shaders 25%(1000). However, the shader area of the core isn't too big, so they probably could do +35% increase in shaders(1080 and beyond) whilist keeping the chip under current size.

What will be done? If it is direct increase in shaders, I wouldn't be surprised if they went +1100 for RV790 and +1500 for RV870. If they are pad limited with 256-bit MC w/ GRRD5, they can not reduce the die area too much without being pad limited again, they can only make the chip bigger. The question is, how far are they willing to go with the die size? Obviously big dies do not fit their strategy. Their superior strategy is biting them right now?

For me it seems that both cores are going to be at least as big as RV770 is. RV790 - 5-10 % bigger, RV870 - 20-33 % bigger would be my bet. Now just calculate how many shaders fit to that area and that would be my bet. :shrug:

Quote:

---

Originally Posted by Calmatory

No, but that would make some sense. If the pads aren't being shrunk, then anything below the current size would make the chip pad limited. 40 nm parts are 25 % smaller, 800 -> 960 is the closest match which would increase the shaders 25%(1000). However, the shader area of the core isn't too big, so they probably could do +35% increase in shaders(1080 and beyond) whilist keeping the chip under current size.

What will be done? If it is direct increase in shaders, I wouldn't be surprised if they went +1100 for RV790 and +1500 for RV870. If they are pad limited with 256-bit MC w/ GRRD5, they can not reduce the die area too much without being pad limited again, they can only make the chip bigger. The question is, how far are they willing to go with the die size? Obviously big dies do not fit their strategy. Their superior strategy is biting them right now?

For me it seems that both cores are going to be at least as big as RV770 is. RV790 - 5-10 % bigger, RV870 - 20-33 % bigger would be my bet. Now just calculate how many shaders fit to that area and that would be my bet. :shrug:

---

How did you come up with 25% smaller? You are going from 55 nm. to 45 nm. and then to 40 nm. and I'd say you would end up about 45% smaller than at 55 nm. at least.
As for die size, they can make a smaller die by ditching Sideport and reducing power draw. The latter would mean that you would need less VDD and GND pins, which will also reduce your minimum necessary die size. So I don't think that a die size of about 200 mm^2 for RV790 would be too far off and entirely possible while still increasing unit count.

So they are not going to go directly to 40nm? Oh, you are right, ideal scaling would be 52% (40/55)^2 or 66 % (45/55)^2. So, expecting at least ~60 % or ~73 % smaller chips. Yet another sleepless night makes this thinking impossible.

So, if they are pad limited then RV790 should be at least 1280 shaders, even if they are going to 45nm first. If RV870 was 40 nm, then 1500-1600+ shaders should be very well possible.

Quote:

---

Originally Posted by Calmatory

So they are not going to go directly to 40nm? Oh, you are right, ideal scaling would be 52% (40/55)^2 or 66 % (45/55)^2. So, expecting at least ~60 % or ~73 % smaller chips. Yet another sleepless night makes this thinking impossible.

---

No, I only meant to imply that ATI is making a full node jump. There probably will not be 45 nm. chips made by TSMC for ATI/AMD.

Quote:

---

Originally Posted by Calmatory

So, if they are pad limited then RV790 should be at least 1280 shaders, even if they are going to 45nm first. If RV870 was 40 nm, then 1500-1600+ shaders should be very well possible.

---

Or they will increase the capability and/or efficiency of the shaders in RV870, in which case the shaders will take up more space, but they will perform better than the actual jump in the amount of shaders would suggest. They have to add some extra logic to RV870's shaders anyway to make it compatible with DX11's specs, although I wouldn't be surprised if RV770 is already reasonably close to having all that is needed to support DX11.
RV790 will have the exact same units as used in RV770, but there will simply be more of 'em. Just my guess though, I wouldn't mind to be proven wrong.

Quote:

---

Originally Posted by Helmore

Or they will increase the capability and/or efficiency of the shaders in RV870, in which case the shaders will take up more space, but they will perform better than the actual jump in the amount of shaders would suggest. They have to add some extra logic to RV870's shaders anyway to make it compatible with DX11's specs, although I wouldn't be surprised if RV770 is already reasonably close to having all that is needed to support DX11.
RV790 will have the exact same units as used in RV770, but there will simply be more of 'em. Just my guess though, I wouldn't mind to be proven wrong.

---

I am pretty sure as well that ATI already has most of the additional logic for DX11 from their bid on DX10.1.

I also am pretty confident that at least the efficiency of the shaders will be improved, as they were able to do this in RV670 -> RV770 , which resulted in "smaller" units (more performance per mm^2 comparitively). As far as capability, that is the one that I am unsure of, as they already did some minor tweaks to the shaders to make them more GPGPU "friendly"; though who knows what changes DX11 brings to the table in the GPGPU arena (from a design standpoint).

Back to RV790 though: last time that I checked the "RV790 and RV740 in March 2009" thread, a few people were certain that RV790 would be 55nm, and just a slightly tweaked version of RV770 to yield higher clocks.
But now, you are saying that RV790 will be 40nm like RV740?
:confused:

Quote:

---

Originally Posted by ColonelCain

I am pretty sure as well that ATI already has most of the additional logic for DX11 from their bid on DX10.1.

I also am pretty confident that at least the efficiency of the shaders will be improved, as they were able to do this in RV670 -> RV770 , which resulted in "smaller" units (more performance per mm^2 comparitively). As far as capability, that is the one that I am unsure of, as they already did some minor tweaks to the shaders to make them more GPGPU "friendly"; though who knows what changes DX11 brings to the table in the GPGPU arena (from a design standpoint).

Back to RV790 though: last time that I checked the "RV790 and RV740 in March 2009" thread, a few people were certain that RV790 would be 55nm, and just a slightly tweaked version of RV770 to yield higher clocks.
But now, you are saying that RV790 will be 40nm like RV740?
:confused:

---

I'd say it wouldn't be wise of them to do another high(er) end chip on 55 nm. You have to consider that ATI only has a limited amount of resources and another 55 nm. would not be a wise investment of those resources. A 40 nm. chip will get them some extra practice with TSMC's new process node and it would make it easier to make a more capable chip and thus more compatible chip. A new high end chip at 55 nm. would simply not provide enough of a performance jump to be able to compete with it long enough to regain the investment you made to develop it. That's because NVIDIA will probably come in Q2 with it's 40 nm. parts, or that is probably what ATI is expecting from them. So all in all, a 40 nm. chip would be far better to do for ATI from a strategic/tactical point of view than a new 55 nm. one and that's why I'm guessing RV790 will be made on 40 nm. That's just my prediction though, although w0mbat and some other folks over at Beyond3D seem to be pointing in the same direction so to speak.

That depends a lot on the existence of GT212, a supposed less than 300 mm^2 40 nm chip with 384 SPs 96 TMUs and 32 ROPs on 256 bot GDDR5. If that chip successfully makes appereance in late Q2, ATi MUST have something competitive against it, on 40 nm node too.

ATI-Forums.de seem to think that RV790 might not even be launched at CeBit and will thus arrive later, although they aren't completely sure about that either. If that's the case than it is even more unlikely for RV790 to be made on 55 nm.
Source: http://news.ati-forum.de/index.php/n...och-kein-rv790
Google translate: http://translate.google.com/translat...90&sl=de&tl=en

Quote:

---

Originally Posted by Helmore

ATI-Forums.de seem to think that RV790 might not even be launched at CeBit and will thus arrive later, although they aren't completely sure about that either. If that's the case than it is even more unlikely for RV790 to be made on 55 nm.
Source: http://news.ati-forum.de/index.php/n...och-kein-rv790
Google translate: http://translate.google.com/translat...90&sl=de&tl=en

---

Maybe because there is no RV790 since RV870 is still "relatively" on schedule.

Schedule being? Long walk till the summer, Q3 to be more precise.

Quote:

---

Originally Posted by LordEC911

Maybe because there is no RV790 since RV870 is still "relatively" on schedule.

---

There is a RV790, that i can tell u for shure. Recent rev is A11 w/ 750MHz coreclk and 900MHz GDDR5 memclk (just like RV770).

I wonder if they're going to make a refresh of the dual gpu part ?

Quote:

---

Originally Posted by Calmatory

Schedule being? Long walk till the summer, Q3 to be more precise.

---

being the same as it has the past few years. At least I imagine that's the implication. Process shrink after 6 months from prev. generation, then ~6 months later(1 year from previous gen. release), new generation.

Quote:

---

Originally Posted by ryboto

being the same as it has the past few years. At least I imagine that's the implication. Process shrink after 6 months from prev. generation, then ~6 months later(1 year from previous gen. release), new generation.

---

That would mean RV790 is late quite a bit. Hopefully RV870 isn't. :)

Quote:

---

Originally Posted by Calmatory

That would mean RV790 is late quite a bit. Hopefully RV870 isn't. :)

---

I used "~", plus, the 4870X2, last of the 4800 series came in Aug, 6 months from then is Feb. Also, while it's ATi's intention to do things this way, they do have to wait for TSMC to have a process ready.

My edumacated guess is that a lot of the RV870 speculation might be confused with RV790 if the 750/900 clocks are true. And to provide a performance boost, it's going to have expanded units with a goal of 1.5 TFlops.

Quote:

---

Originally Posted by zerazax

My edumacated guess is that a lot of the RV870 speculation might be confused with RV790 if the 750/900 clocks are true. And to provide a performance boost, it's going to have expanded units with a goal of 1.5 TFlops.

---

Or a little more. RV790 is a 40nm "fill-in" for RV870 coming out Q3/Q4?
A nice RV790 would be 1200SPs, 24SPs in 10clusters, @750mhz = 1.8TFlops.
~170mm2? 1.15b trannies? 40-50% performance increase over RV770?

Definitely leaves room for a 2000SP RV870.

Quote:

---

Originally Posted by LordEC911

Or a little more. RV790 is a 40nm "fill-in" for RV870 coming out Q3/Q4?
A nice RV790 would be 1200SPs, 24SPs in 10clusters, @750mhz = 1.8TFlops.
~170mm2? 1.15b trannies? 40-50% performance increase over RV770?

Definitely leaves room for a 2000SP RV870.

---

Very possible too if it is 40nm

Either way, I don't believe the RV870 only has 1000-1200 SP's if its 40nm, unless they really do MCM and put two of them together. Then again, people were all over the 480 SP rumor and look what happened.

Quote:

---

Originally Posted by LordEC911

Or a little more. RV790 is a 40nm "fill-in" for RV870 coming out Q3/Q4?
A nice RV790 would be 1200SPs, 24SPs in 10clusters, @750mhz = 1.8TFlops.
~170mm2? 1.15b trannies? 40-50% performance increase over RV770?

Definitely leaves room for a 2000SP RV870.

---

1.15 billion trannies? Sounds like a crazy party.

Quote:

---

Originally Posted by bowman

1.15 billion trannies? Sounds like a crazy party.

---

That's only about 200 million more than RV770 already has, doesn't sound too far off and it probably is a pretty conservative estimate. I'd say that RV790 has more than that, if it has a 256-bit memory bus.

@LordEC911: I don't think they'll change the amount of SPs per cluster/SIMD-core, that would be too much of a redesign and it would increase their ALU/TEX ratio even further (to 6:1 in your case). That's something they won't do with what is supposed to be a reasonably straight shrink. I guess we will only see an increase in the amount of clusters. I've no idea what they will do with their RBEs/ROPs, I don't think they will change the way they work and I therefor don't really expect an increase in ROPs/RBEs. Or they would add more ROP/RBE clusters with each a 32-bit memory controller instead of a 64-bit one, I think that would require the least amount of changes to the ROP/RBE section while still being able to increase the amount of ROPs/RBEs. Oh well, in the end it is all guess work for us and we may all be completely wrong.

Quote:

---

Originally Posted by Helmore

That's only about 200 million more than RV770 already has, doesn't sound too far off and it probably is a pretty conservative estimate. I'd say that RV790 has more than that, if it has a 256-bit memory bus.

@LordEC911: I don't think they'll change the amount of SPs per cluster/SIMD-core, that would be too much of a redesign and it would increase their ALU/TEX ratio even further (to 6:1 in your case). That's something they won't do with what is supposed to be a reasonably straight shrink. I guess we will only see an increase in the amount of clusters. I've no idea what they will do with their RBEs/ROPs, I don't think they will change the way they work and I therefor don't really expect an increase in ROPs/RBEs. Or they would add more ROP/RBE clusters with each a 32-bit memory controller instead of a 64-bit one, I think that would require the least amount of changes to the ROP/RBE section while still being able to increase the amount of ROPs/RBEs. Oh well, in the end it is all guess work for us and we may all be completely wrong.

---

I know I had this discussion over at B3D but do you recall how exactly the TMUs are setup? Is it a quad for each SIMD? I thought that was my original assumption when looking at some of the architecture shots of R600/RV670 but I recall being told something different.

If they don't increase the number of SPs per SIMD, from 16-24, then there really isn't much they can do with RV790. They could up the SIMDs to 12 like was talked about before that isn't much of an increase.

AMD/ATi really needs to go back to back to back with strong GPUs if they really want to take over even more marketshare.

Quote:

---

Originally Posted by LordEC911

I know I had this discussion over at B3D but do you recall how exactly the TMUs are setup? Is it a quad for each SIMD? I thought that was my original assumption when looking at some of the architecture shots of R600/RV670 but I recall being told something different.

---

The TMUs are setup differently on RV770 than on R600/RV670. On R600/RV670 a TMU block was assigned to 4 SP blocks, each block from a different SIMD 'core'. This approach has its benefits, but it is a bit tough to scale it and that's why they abandoned it for RV770. RV770 has 4 TA units and 4 TF units per SIMD 'core'. Just check here: http://www.rage3d.com/reviews/video/.../index.php?p=5
There's one odd part though, and that is that there are 'only' 32 interpolators. But that will only become a bottleneck if you INT8 bilinear filtering, you won't notice it once you apply some more AF for example and then the bottleneck will shift to the filtering units. It's still weird though, as you'd expect a texture related unit to reside with the TF/TA units and thus also scale in capability along with those.

Quote:

---

Originally Posted by LordEC911

If they don't increase the number of SPs per SIMD, from 16-24, then there really isn't much they can do with RV790. They could up the SIMDs to 12 like was talked about before that isn't much of an increase.

AMD/ATi really needs to go back to back to back with strong GPUs if they really want to take over even more marketshare.

---

I don't think they will increase the number of SPs per SIMD 'core', not with their first 40 nm. chips and probably not even with RV870. To me it seems like it should be fairly easy for them to increase the amount of SIMD 'cores' on a chip. They increased the amount of SIMD 'cores' that RV770 would get at the last minute when they needed to fill up some extra space, so to me it seems like you don't need to do a lot of redesigning/work to increase the SIMD 'core' count. I'd say 16 SIMD 'cores' for RV790 shouldn't be too hard to do for them.
I'd like to know what they would to with their RBEs if they do put 16 SIMD 'cores' on there though, although I'm not really sure if yet if 16 RBEs would become a bottleneck.

It will simply be guesswork for us until it's launched I guess, still interesting to think about though.:)

Quote:

---

Originally Posted by Helmore

I don't think they will increase the number of SPs per SIMD 'core', not with their first 40 nm. chips and probably not even with RV870. To me it seems like it should be fairly easy for them to increase the amount of SIMD 'cores' on a chip. They increased the amount of SIMD 'cores' that RV770 would get at the last minute when they needed to fill up some extra space, so to me it seems like you don't need to do a lot of redesigning/work to increase the SIMD 'core' count. I'd say 16 SIMD 'cores' for RV790 shouldn't be too hard to do for them.
I'd like to know what they would to with their RBEs if they do put 16 SIMD 'cores' on there though, although I'm not really sure if yet if 16 RBEs would become a bottleneck.

It will simply be guesswork for us until it's launched I guess, still interesting to think about though.:)

---

Keep in mind that this won't be their first 40nm chip...
Anyways, it wouldn't be much harder to add more SPs into a SIMD than to add SIMDs. Adding more SIMDs will cause a reworking of the scheduler and will cause more overhead/latency. As for the ROPs, they sorta seem to be related to the memory controllers, like Nvidia, but not directly. I don't think more ROPs is really needed right now but maybe for RV870, ROPs are large and take up a lot of bandwidth.