Wednesday, May 8th 2013
AMD's Answer to GeForce GTX 700 Series: Volcanic Islands
GPU buyers can breathe a huge sigh of relief that AMD isn't fixated with next-generation game consoles, and that its late-2013 launch of its next GPU generation is with good reason. The company is building a new GPU micro-architecture from the ground up. Codenamed "Volcanic Islands," with members codenamed after famous islands along the Pacific Ring of Fire, the new GPU family sees AMD rearranging component-hierarchy within the GPU, in a big way.
Over the past three GPU generations that used VLIW5, VLIW4, and Graphics CoreNext SIMD architectures, the component hierarchy was essentially untouched. According to an early block-diagram of one of the GPUs in the series, codenamed "Hawaii," AMD will designate parallel and serial computing units. Serial cores based on either of the two architectures AMD is licensed to use (x86 and ARM), could handle part of the graphics processing load. The stream processors of today make up the GPU's parallel processing machinery.
We can't make out text in the rather blurry block-diagram, but are rather convinced that if it's authentic, then AMD is making some big changes. Another reason for AMD's delay could be silicon fab process. "Tahiti" as implemented on Radeon HD 7970 GHz Edition, already poses high thermal envelope. AMD doesn't want the 28 nm process to restrict its next-generation architecture development, and is holding out till the 20 nm process is in place at TSMC. The fab set Q4 as its tentative bulk manufacturing date for the process.
The source that leaked the block-diagram also posted specifications of the chip that's codenamed "Hawaii," which appears to be the flagship part.
Source:
ChipHell
Over the past three GPU generations that used VLIW5, VLIW4, and Graphics CoreNext SIMD architectures, the component hierarchy was essentially untouched. According to an early block-diagram of one of the GPUs in the series, codenamed "Hawaii," AMD will designate parallel and serial computing units. Serial cores based on either of the two architectures AMD is licensed to use (x86 and ARM), could handle part of the graphics processing load. The stream processors of today make up the GPU's parallel processing machinery.
We can't make out text in the rather blurry block-diagram, but are rather convinced that if it's authentic, then AMD is making some big changes. Another reason for AMD's delay could be silicon fab process. "Tahiti" as implemented on Radeon HD 7970 GHz Edition, already poses high thermal envelope. AMD doesn't want the 28 nm process to restrict its next-generation architecture development, and is holding out till the 20 nm process is in place at TSMC. The fab set Q4 as its tentative bulk manufacturing date for the process.
The source that leaked the block-diagram also posted specifications of the chip that's codenamed "Hawaii," which appears to be the flagship part.
- 20 nm silicon fab process
- 4096 stream processors
- 16 serial processor cores
- 4 geometry engines
- 256 TMUs
- 64 ROPs
- 512-bit GDDR5 memory interface
145 Comments on AMD's Answer to GeForce GTX 700 Series: Volcanic Islands
The specs do make sense, but like Wizzard said, it could be a fabricated news. I hope it's not, cuz BF4 is coming out later this year, and I want to upgrade to a worthy card that can max BF4 out on a 120Hz 2560x1440 res. display.
on another note: Are we getting any 8 bit, backlight-strobed displays this year with 120Hz refresh rate? I mean with those 8 bit panels that are rated at 6ms pixel transition times, it's perfectly possible with a strobing backlight to achieve perfectly smooth and clear animation that surpasses that of 120Hz TN displays with 2-3ms pixel response ratings.
Anyways, this thread is not even about Titan, so moving on.... it seems like AMD is shooting for a GPU with more general purpose focus, if these cores are ARM cores they will use a cut down version of this GPU for mobile devices, depending on power consumption, this might actually be the breakthrough AMD needs to get their designs into more mobile devices :rockout:
Let's not forget that the green team is also an ARM licensee, and they already have parts that couple A15 cores with Kepler cores in the pipeline, GPU designs take over aprox 24 mos. to reach the market, so they may be working on a similar design using Maxwell cores as well, but at this point this is all speculation :)
Amd are beating many to the ball with this tech and due to their apu , gpu and soc achievements they are certainly one to bet on imho.
These are the same people who name a Southern Islands GPU after a Northern hemisphere island archipelago (Cape Verde)
AMD ain't care. If this information gets traction, How many 7970/7990/7950's do you think AMD are going to sell ? You really think that a purported AMD slide (and lets face it, don't you find it just a little bit suspicious that this surfaces just as the GTX 700 hype machine gains momentum?) is going to headshot Nvidia's sales but leave AMD untouched ?
You do remember AMD Osborning themselves with the Trinity hype that resulted in the company taking a write down on Llano inventory recently ? AFAIA, GloFo's 20nm HKMG so far is of the low power (LPM) variety,20nm TSVisn't slated for volume production until 2014.
Still, the "late-2013" launch, could mean just about anything (and I'm presuming the original time frame originates from James Prior's addition at the end of this piece) - tape out, risk wafer debug/validation, volume wafer starts laid down, shipping for revenue, or even ....retail availability...but if it's on 20nm then the signs are that it would be TSMC.
Also if you are going to compare an ATI superclocked edition in a vs comparison, compare it to an NVIDIA superclocked edition.
Not an NVIDIA standard model..
Especially so because those serial processing modules have an uncanny resemblance to bulldozer/piledriver cores...
EDIT: OH WAIT this thread had a second page *derp*
EDIT #2: OH TEH WAIT AGAIN this much more resembles a ramped-up version of the PS4 :|
As for manufactured news I don't care bumped the stock up another $.25 :)
Bus change...256-bit bus width to 384
IP insertion...VLIW4 to GCN
Process node...40nm to 28nm
TSV....I was under the impression that TSMC had already demonstrated viability under 28nm (CoWoS) over a year ago. TSV looks to be an extrapolation of the process.
edit - also, take a look at the new play station hardware, kinda similar? i know little about programming but i imagine porting to a system that has cpu and system memory separated by a comparatively slow, high latency pcie bus could be a hassle. now if you had 8gb of memory and some cpu-like cores on the card?
:toast:
Im sorry though i should have said, I do agree with your original point that Amd will be waiting on 20nm finfet bulk process or Soi gate last I forget the actual process names.
If AMD did really "out'd-this" info this early they played it wrong! I would've waited the month when Nvidia started releasing card like the GTX780/GTX770 for review and had set MSRP's, then fired off with "technical white paper" on what they will have, that would've been the right timing.
Heck for all we know Nvidia out'd this information and some parts of it carry some truth. Better for them to get it out now than letting AMD pull it out and rain on their release. We just never know who behind the curtain and what the string actually mean.
doesn't seem right.