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
Sincerely, AlMostDead
even today, crysis 2 is still a great choice for power consumption testing
And I agree, Crysis 2 maxed out can still stress most hardware configurations out there, and make even the fastest system break a sweat, it's as good a test tool as any other game out there.
I think its quite clear to most that the pics of an Apu, so if this were Vi then amd will just be making scaleable Apus for everything and most situations, and its too soon for that imho but id welcome a mythical gpu like that pic x2, because damn them things would fold well even intels phi would look a bit weak on double precision compared to that spec of chip.
So with Crysis 2 the Ghz furnishes 9.8% more performance, while requiring 28% more watts than a GTX 680. If several other titles had that same trend(s) it could be appreciated it as a veritable results, but one data point is not definitive proof.
While Titan has 35% performance increase while basically using the same power of GHz, (even with the one data point) Titan appears to have some determinate efficiency. Now, can the Titan LE (fusing off 2 SMX) provide something approaching that performance/watts?
A couple of observations:
1. How do you know that the maximum measurement is the peak of a millisecond?, and
2. Isn't it conceivable that the other cards are being measured the exact same way...so in fact the GTX 680's full load power measurement could also be a transient peak of a millisecond duration ? I was told by another guy who holds AMD to be the one true god, that they have it on good authority that the 7970GE was tested with Crysis 2, Metro 2033, BF3 multiplayer, Furmark, and 3DMark while the GTX 680 was tested with Solitaire, Minesweeper, Tetris, Farmville and The Lost Titans. If true- and I'm assured it is, that could account for the discrepancy. If so, then the world-wide conspiracy against AMD does indeed cover the entire planet!
Xbit ( difference of 78W in Metro 2033)
HT4U(difference of 76W - gaming benchmarks)
PCGH ( difference of 73W in Battlefield Bad Company 2)
Hardware.info(difference of 65W in Metro 2033)
Hexus (difference of 61W in Far Cry 3)
PC Perspective( difference of 61W in BF3)
SweClockers (difference of 50W - application not specified)
Lab501 (difference of 47W in Crysis 2)
Hardware Canucks(difference of45W in Ungine Valley bench)
TechPowerUp(difference of 43W in Crysis 2)
HotHardware( difference of 40W -application not specified)
TechSpot (difference of 38W in Crysis 3)
Hardware France(difference of 42W in Anno 2070 and 36W in BF3)
Bit-tech(difference of 35W in Unigine Heaven bench)
ComputerBase (difference of 35W in AC3)
Anandtech (difference of 24W in BF3)
HardwareLUXX (difference of 13W - application not specified)
Tech Report( difference of 8W - application not specified)
By my count, that takes in Northern, Central, Western, and Eastern Europe, North America, and Australia.
So the question is, what is faster my AMD Radeon HD 4870 512mb or the Intel HD 4600 on the 4770k?
Next question, as i am pretty sure Intel will not have managed to compress an entire high-end (ish) video card from 2008 into a few transistors on a CPU, can i still use the OpenCL power of the Haswell chip if i have my old discrete card plugged in? I remember on early Sandy Bridge reviews you could only use QuickSync if you had a monitor plugged into Intel's video outputs.
Discrete graphics won't choke in games like integrated. The Intel one does 4K video well but for games the AMD takes it. Can somebody else advise :D
Wouldn't it be more sensible to see some reviews BEFORE making that decision?
I use a hybrid physx card that needs to have a monitor attached btw I just use the second input on a single monitor, I can't see them both but dont use thr second input and it enables the features I want, the pjl136 dude could use this same tactic on his intel gfx output surely.
One day this phone will pay for all these messups.
Odd dp I edited sorry.
most editors use cheap killawatts that take a reading every 1-2 seconds. some even slower/people just look and memorize the highest number.
a handful of sites use proper measuring devices. i'm running at 12 samples per second, which in my opinion is a good compromise between accuracy and speed.
it could be interesting to look at power consumption with sub-microsecond resolution to observe the effects of power limiting systems, but spending a few k just for that doesn't seem to be worth it.