• We’re currently investigating an issue related to the forum theme and styling that is impacting page layout and visual formatting. The problem has been identified, and we are actively working on a resolution. There is no impact to user data or functionality, this is strictly a front-end display issue. We’ll post an update once the fix has been deployed. Thanks for your patience while we get this sorted.

[bit-tech] Radeon HD 7000-series rumoured for May production

Page 7 - Seeking answers? Join the AnandTech community: where nearly half-a-million members share solutions and discuss the latest tech.
I've been saying this all along - it should be very easy for AMD to produce a 32nm GPU because that's what they've been trying to do all along. They've probably had it waiting in the wings for some time now.

If they do come out with Bulldozer, Llano, and this new GPU all on 32nm and all at the same time, things will look very good for the company. They will surely have a terrific product lineup that can compete with both Intel and nVidia.
 
If they will be taping out first samples in May/June, then Q4 seems much more likely for 7xxx, which is in line with previous new series launches. Q2/early Q3 seems BS.
 
iCyborg said:
If they will be taping out first samples in May/June, then Q4 seems much more likely for 7xxx, which is in line with previous new series launches. Q2/early Q3 seems BS.
Click to expand...

good assumption. amd likes to get their releases out right before the holiday sales season.
 
More than likely the HD7xxx will be an extension of the same VLIW-4 design of Cayman. Cayman had to be back ported to TSMC's 40nm process, so we'll probably see something akin to what AMD originally meant for HD69xx to be.
 
(sic)Klown12 said:
It is. There is nothing suggesting that AMD will switch to GloFo's 32nm process for discrete graphics cards, so I don't understand why people bring it up. GloFo is going to be busy enough with Llano and Bulldozer.
Click to expand...

I wouldn't be so sure.

AMD will not utilize 32nm GPU process because TSMC and GloFo have shifted to 28nm process. But, if you meant to say 28nm, there is a possibility GloFo may provide some 28nm GPUs for AMD.

http://www.xbitlabs.com/news/other/..._Expects_Two_28nm_Tape_Outs_This_Quarter.html

"We expect several 28nm tape-outs during the current quarter. We balance our loading according to risk performance and price. But we also said that there is, of course, a natural incentive for us at this point to work closely with Globalfoundries, and that is, of course, something we continue to do," said Thomas Seifert, interim chief executive officer of AMD and chief financial officer.
 
Last edited:
iCyborg said:
If they will be taping out first samples in May/June, then Q4 seems much more likely for 7xxx, which is in line with previous new series launches. Q2/early Q3 seems BS.
Click to expand...


It's all rumors, at this point, but the rumor is "mass production" in May. Not Tape out in May or June. While no one's expecting AMD to release Southern Islands at the same time as Bulldozer, it's hard to argue with how big that could be for them to pull that off. Along those lines, if they were to do it, I don't think they'd release a midrange "pipe cleaner" SKU, if they were to pull off a multiple release on both GPU and CPU platforms.

It's all just speculation, but imagine the splash Bulldozer and Southern Islands would make if released together. 8 core +4GHz CPU (including turbo boost) and the "7970" (+3000 spu) being reviewed as a simultaneous release. It would be a great day in tech. 🙂
 
Didn't Charile say that a SI chip taped out back in late march? The two chips taping out could be mid-range cards rather than high end or the witchita APU.
 
3DVagabond said:
It's all rumors, at this point, but the rumor is "mass production" in May. Not Tape out in May or June. While no one's expecting AMD to release Southern Islands at the same time as Bulldozer, it's hard to argue with how big that could be for them to pull that off. Along those lines, if they were to do it, I don't think they'd release a midrange "pipe cleaner" SKU, if they were to pull off a multiple release on both GPU and CPU platforms.
Click to expand...
The last rumour (post #148) talks about a tape-out with the source being T. Seifert, so coming from AMD directly. And in the article they also think it's more likely it's a GPU than a new APU. The original story from the opening post comes from "their sources". Yes, it's all rumours, but I'm putting more eggs into the tape-out basket.
 
No, because GDDR6 doesn't exist.
Possibly it will have a larger memory bus, or a memory controller which is capable of higher clocks (less likely).
 
Lonyo said:
No, because GDDR6 doesn't exist.
Possibly it will have a larger memory bus, or a memory controller which is capable of higher clocks (less likely).
Click to expand...

You think a larger bus is more likely than an improved memory controller?
 
Skurge said:
You think a larger bus is more likely than an improved memory controller?
Click to expand...

A larger bus is much easier than an improved memory controller, given how much of an improvement they would need to get to make it a similar bump to an increased memory bus.
They are at 5.5Gbps now, GDDR5 tops out at 7Gbps, and they've had a hard time getting to even 5.5Gbps according to various AT articles (so have NV, so it's not like there's a magic breakthrough to be found).
So getting +25% or so from an improved memory controller is probably going to be a lot harder than getting +25~50% from going to a 320~384-bit memory bus.
 
Lonyo said:
No, because GDDR6 doesn't exist.
Possibly it will have a larger memory bus, or a memory controller which is capable of higher clocks (less likely).
Click to expand...

so it will be improved memory controller especially with 28nm AMD have luxury to improve it after all there some cayman GDDR5 ram doesn't fully enabled( iI mean it doesn't clocked as high as recommended )
 
Lonyo said:
A larger bus is much easier than an improved memory controller, given how much of an improvement they would need to get to make it a similar bump to an increased memory bus.
They are at 5.5Gbps now, GDDR5 tops out at 7Gbps, and they've had a hard time getting to even 5.5Gbps according to various AT articles (so have NV, so it's not like there's a magic breakthrough to be found).
So getting +25% or so from an improved memory controller is probably going to be a lot harder than getting +25~50% from going to a 320~384-bit memory bus.
Click to expand...

thats because high speed GDDR5 memory controller require complex MC, thats why fermi and bart can't use the higher speed like cayman because bart and fermi already die space constrain.

btw if AMD going with odd bus config it will eventually hit fermi complex like GTX 580 whre 1,5 Gb is not enough and 3 Gb is over kill, so i don't think AMD will go that route
 
Lonyo said:
A larger bus is much easier than an improved memory controller, given how much of an improvement they would need to get to make it a similar bump to an increased memory bus.
They are at 5.5Gbps now, GDDR5 tops out at 7Gbps, and they've had a hard time getting to even 5.5Gbps according to various AT articles (so have NV, so it's not like there's a magic breakthrough to be found).
So getting +25% or so from an improved memory controller is probably going to be a lot harder than getting +25~50% from going to a 320~384-bit memory bus.
Click to expand...

A larger bus is going to give them other problems like odd memory configurations and have the die larger than it has to be. Cayman is already larger than AMD would like, probably.

The controller on Cayman is likely just an evolution of the RV770 controller so that could explain the difficulties in reaching higher speeds. As for nvidia, GF100 was their first High-End GPU to use GDDR5 and it started out at the same speeds AMD started with the 4870.

It wouldn't be too far fetched for them to have a brand new controller ready for S.I. It might have been one of the things left out of Cayman when they had to back-port it to 40nm.
 
http://forums.anandtech.com/showpost.php?p=31609460&postcount=107

Cracking open the 6970 we find the PCB with the Cayman GPU at the center in all its 389mm2 glory. Around it are 8 2Gb Hynix GDDR5 chips, rated for 6Gbps, 0.5Gbps higher than what the card actually runs at. As we’ve said before the hardest part about using GDDR5 at high speeds is the complexity of building a good memory bus, and this continues to be the case here. AMD has made progress on getting GDDR5 speeds up to 5.5Gbps primarily through better PCB designs, but it looks like hitting 6Gbps and beyond is going to be impractical, at least for a 256bit bus design. Ultimately GDDR5 was supposed to top out at 7Gbps, but with the troubles both AMD and NVIDIA have had, we don’t expect anyone will ever reach it.
Click to expand...

Bus, not controller, but the point is that getting higher memory speeds isn't the easiest thing according to AT.
 
A lot of 580s are being sold recently. Point to anything?

Either way, May has come and gone uneventfully.
 
It's all quiet in the woods.. boring no news, not even leaks.

Guess more time for forum campers to actually put their GPUs to good use and do some serious gaming.
 
You must log in or register to reply here.

TRENDING THREADS

Back
Top