AMD Plans RV775XT To Compete GeForce GTX285

[ubuntu83]

. There are and have been many rumors of a RV740 coming, which would be ATI's first 40 nm. chip, but nobody knows (except ATI and some people whom are under NDA) what the chip will be like. It could simply be a RV730 done on 40 nm., but it just as well be something different.
Rumors on RV740 so far say it has a 128-bit memory bus and 8-SIMDs, which makes it rather similar to a RV730. But those 8-SIMDs could just as well consist of SIMDs with 16 shader execution units per SIMD. Which would mean the card has 640 SPs, unlike the RV730 which has 8 shader units per SIMD and thus 320 SPs. To me the version with 640 SPs seems plausible as this would slightly increase die size to make sure the chip does not get to small to not allow for a 128-bit bus. By that I mean that the current RV730 might get too small to still be able to put the 128-bit memory bus on there if they did a direct shrink. Just my speculation though.

If they add GDDR5 to it then it could become a HD4830/9800GT killer. They should make it 560 SP's instead of 640 to make it competitive with 9600GT instead of their own if they have plans to use GDDR5. later they can make HD5670 with 640 SP's and GDDR5 to replace HD4670. For now these are all my fantasies. We don't even know if they plans for releasing a RV740.

[SNiiPE_DoGG]

you all have seen the great strides r700 arch makes at higher clocks, if AMD releases a GPU the can do 950mhz at stock settings (nothing more than some voltage and cooling at this point for some cards) then I think it will give the 285 a run for its money. if it is lower clocks it will be a failed attempt

[LordEC911]

I'm confused is this how it is?

R740 ~ 40nm R730 (4670)?
R775 = high binned R770
R780 ~ 40nm R770
R790 ~ 40nm R775
R870 = 40nm new GPU with DX11
R800 = 2xR870

I was more thinking-
RV740 = 40nm ~4830/4850 performance.
RV775=RV790= new RV770 revision on 55nm
R780= 2xRV775/790
RV870= 40nm nextgen evolution of RV770
R800= 2xRV870

[Loque]

useless, we need 2gb gfx cards to play gta4 on max settings :p

[Helmore]

If they add GDDR5 to it then it could become a HD4830/9800GT killer. They should make it 560 SP's instead of 640 to make it competitive with 9600GT instead of their own if they have plans to use GDDR5. later they can make HD5670 with 640 SP's and GDDR5 to replace HD4670. For now these are all my fantasies. We don't even know if they plans for releasing a RV740.

Why should they? The 4830 is made by disabling parts on the die, which is a bit of a waste once you do have good yields. By the time something like this (hypothetical RV740) comes along the yields on their RV770 should be well enough to not need the scavenged part (4830). So the hypothetical RV740 would be perfect for AMD to replace their by then silicon wasting 4830. Just my theory though.

[YukonTrooper]

110% increase in performance over GTX 280? Holy crap, eh? I'm down.

[LordEC911]

110% increase in performance over GTX 280? Holy crap, eh? I'm down.

It offers 110% of the performance of a GTX280, 10% performance increase.

Should read the original source...
http://en.expreview.com/2008/12/11/5...gpu-cards.html

[ToTTenTranz]

Cherry-picked RV770s with better cooling and better power regulation could hit the 1Ghz mark, making it highly competitive to the GTX280.

I just hope this is not another lame-and-bundled-watercooling-only edition, those always suck..

[ColonelCain]

So what, kinda in the neighborhood of an 8800 Ultra?

As far as 40nm, I can see an RV740, but smack-dab midrange performance. However, I can't see a 40nm derivitive of RV770. I'm sure that shrinking the RV730 would be much easier compared to shrinking the RV770. It makes sense from an engineering standpoint.

[LordEC911]

So what, kinda in the neighborhood of an 8800 Ultra?

As far as 40nm, I can see an RV740, but smack-dab midrange performance. However, I can't see a 40nm derivitive of RV770. I'm sure that shrinking the RV730 would be much easier compared to shrinking the RV770. It makes sense from an engineering standpoint.

I would imagine the RV740 is close to a 40nm RV770.
Shrinking RV770 to 40nm would put it in the 140-150mm2 range, even without the sideport it doesn't seem possible to fit a 256bit bus on there.

My Speculation for RV740-
40nm
~600-700mil trannies
~100mm2
700-800mhz core
480-640SPs
128bit bus
3.6-4.4ghz GDDR5 in 256/512mb/1Gb flavors
58-70Gbps of bandwidth
32-40TMUs
8-16ROPs
~100w TDP

RV730 was ~3870 performance, IMO they need something to fit between the 4670 and the 4830 which should be the RV740.
If RV740 turns out to be a near straight shrink of RV730 then I expect this speculation to be a RV830 in Q3.

[madcho]

fake.

It's a lot more easyer to start selling RV770 at 850mhz with a better cooler


& in Anandtech review about RV770, AMD said they working to not get the halo in performance, but the halo in mainstream. Creating a RV775 would be a big mistake, and will cost a lot for nothing to gain

[gamervivek]

http://en.expreview.com/2008/12/12/a...a-a-punch.html

900Mhz stock,840 shaders.

[Stukov]

http://en.expreview.com/2008/12/12/a...a-a-punch.html

900Mhz stock,840 shaders.

RV870 not out before Oct of 09? I think that seals the deal, this is misinformation by AMD so Nvidia doesn't know what to expect.

[Calmatory]

RV870 not out before Oct of 09? I think that seals the deal, this is misinformation by AMD so Nvidia doesn't know what to expect.

This reply is necessarily not for you, but more for everyone.

I am quite sure Nvidia has a tad better people guessing whether or not the specs are fake and when it can be expected. Heck, these people have been working on this kind of stuff like over a decade and then there is this "Xtremesystems forums" where people suddenly know it better than engineers at Nvidia/ATI.

[Stukov]

This reply is necessarily not for you, but more for everyone.

I am quite sure Nvidia has a tad better people guessing whether or not the specs are fake and when it can be expected. Heck, these people have been working on this kind of stuff like over a decade and then there is this "Xtremesystems forums" where people suddenly know it better than engineers at Nvidia/ATI.

How good were they on the RV770/R700 launch, price, and performance? The indication is they were not. It wasn't until about a month before GTX 200 launch when AMD realized how well priced/performance their RV770 was. I am pretty sure they go off the same news reports we go off of, which are sources from other vendors in Taiwan, China, etc.

[Calmatory]

How good were they on the RV770/R700 launch, price, and performance? The indication is they were not. It wasn't until about a month before GTX 200 launch when AMD realized how well priced/performance their RV770 was. I am pretty sure they go off the same news reports we go off of, which are sources from other vendors in Taiwan, China, etc.

They still have a lot more knowledge and experience with marchitectures, especially VLIW ones than any of us has. They know what is possible and what is impossible by way more accurately than any of us.

[mascaras]

AMD Sets RV775 to Chase Down G200b?


Sources tell Expreview that AMD is working on a new graphics processor (apart from the mysterious RV790), as an immediate reaction to the 55nm high-end GPUs NVIDIA is set to release. NVIDIA on its part is working on GeForce GTX 260 (55nm), GeForce GTX 285 (successor to GeForce GTX 280) and the dual-GPU monstrosity, GeForce GTX 295. To rival this, according to the source, AMD is working on the RV775 and derivatives based on its yields' performance.

These GPUs are slated for January thru March 2009. It has some very interesting specifications. To begin with, 840 stream processors. While stream processors on a current generation RV770 cannot be added/subtracted in quantities of 40 stream processors, perhaps, the increment of 40 SPs is derived from an extra ALU cluster the RV770 comes with by default (which helps in die harvesting, where the spare ALU cluster makes up for any damaged cluster, thereby improving yields). This would require near-perfect GPUs in the making (if re-engineering of the RV770 on a large scale is to be avoided). Another interesting point is regarding the smaller die. At this point we are clueless as to how that is possible without re-engineering the RV770 at a large scale. Treat yourselves to the specifications meanwhile:



Source: Expreview


http://forums.techpowerup.com/showpo...76&postcount=1

regards

[demonkevy666]

http://en.expreview.com/2008/12/12/a...a-a-punch.html

900Mhz stock,840 shaders.

can't add that little amount of shaders and get less transistors.

it would have over 1 billion and 900 shaders for a top part. adding half a row a shaders isn't going to do much t it's levels already.

[LexDiamonds]

Also, I dont believe AMD will ever launch a performance part with 512Mb from this point forward. As we have seen with the 4870, 1Gb gives definate performance boost. It is also good marketing along side a potential 285 w/ 1gb.

[alexio]

These numbers may very well be real. However, I didn't ask around myself lately, so I can't comment on their validity. I'm just saying that from my understanding a chip with these specs could exist with current technology.

[cadaveca]

Smaller die size is attributed to removal of UVD function, now that STREAM functions are online?

[mascaras]

These numbers may very well be real. However, I didn't ask around myself lately, so I can't comment on their validity. I'm just saying that from my understanding a chip with these specs could exist with current technology.

yep

and we already had rumors about a possible "refresh " >>> http://www.xtremesystems.org/forums/...+RV770+Refresh


regards

[LordEC911]

Smaller die size is attributed to removal of UVD function, now that STREAM functions are online?

Plus the possible removal of the sideport.

Confused by the 1400mhz, 2800mhz GDDR, clock. GDDR4?
Why? 90Gbps vs 64Gbps...

Have to agree seems like misinformation.

[cadaveca]

Well, the whole XT nomenclature guarantees this info isn't from a 100% "reliable" source...

[JohnJohn]

A big piece of FUD in my honest opinion