Bookmark and Share


The first actual specifications and performance estimations regarding next-generation mobile platform code-named Kaveri has leaked onto the Internet. Just as expected, the forthcoming chip from Advanced Micro Devices will bring a lot of innovations to the market. However, it is not projected that it will also bring breakthroughs in performance.

ExtremeSpec web-site has published what it claims to be a slide from an AMD presentation describing what is called “AMD Kaveri 2.0 platform”. Just like AMD said, the A-series “Kaveri” will feature two dual-core x86 Steamroller modules (two integer units per module), one FPU with two 128-bit FMACs, one MMX unit, Radeon R7 graphics processing unit with 512 stream processors (eight GCN compute units) as well as HUMA [heterogeneous unified memory architecture] shared memory controller.

What AMD has not revealed is that the SteamrollerB x86 cores will demonstrate around 20% performance improvement compared to Piledriver x86 cores in the same 15W – 35W power envelope. Graphics engine will provide another 30% performance boost. It is noteworthy that the estimates are “pre-silicon” projections and actual total compute performance of desktop AMD A10-7850K (3.7GHz x86 core clock-rate, 720MHz graphics engine frequency) “Kaveri” APU will be around 856GFLOPS, which is quite a lot.

What AMD has not yet announced is that the Kaveri platform will have new UVD 4.2 universal video decode engine, VCE 2.0 video compression engine, SAMU 2.1 secure asset management unit. Besides, it will feature PCIe 3.0 x16 interconnection for discrete graphics solutions, dedicated PCI Express interface for solid-state drives and so on. To make the new mobile platform even more competitive, it will feature AMD Start Now 3.0 technology with smart sleep capability (probably, it will support Connected Standby technology from Windows 8.1).

Desktop versions of AMD A-series “Kaveri” chips will be available starting from January 14, 2014. Mobile flavours will emerge at a later date.

AMD did not comment on the news-story.

Tags: AMD, Fusion, Kaveri, Radeon, GCN, HSA, TueAudio, Mantle, Steamroller


Comments currently: 23
Discussion started: 11/27/13 09:30:31 PM
Latest comment: 12/04/13 11:18:54 AM
Expand all threads | Collapse all threads


Very poor redacted article

What AMD has not revealed is that the

What AMD has not yet announced is that the

“Kaveri” APU will be around 856GFLOPS, which is quite a lot.
this one repeated twice
0 2 [Posted by: Nehemoth  | Date: 11/27/13 09:30:31 PM]
- collapse thread

Give Anton a break. Russian is his mother tongue, not English.
1 1 [Posted by: linuxlowdown  | Date: 11/28/13 02:48:37 AM]
This site is not "Shakespeare and you", this is a place where we come to read about breaking tech news, detailed hardware reviews and such.

I come here to read and debate about techs, not to read useless comments from a porn addict mutt.

Please accept my best regards!
1 1 [Posted by: MHudon  | Date: 11/28/13 01:22:05 PM]

For some time the 20% CPU and 30% GPU performance increases have been known and these are quite substantial improvements compared to Intels 5-10% with a new CPU. All the extra tech goodies are just the basis for much more love to come soon. It's all good for consumers and AMD.
2 0 [Posted by: beenthere  | Date: 11/27/13 09:44:36 PM]
- collapse thread

If (big if) it turns out to be true, its still not enough to catch up.
0 1 [Posted by: Chris Martinelli  | Date: 11/27/13 10:38:29 PM]
The difference is that Intel increases performance about 5% every single year whereas AMD has not delivered a substantial performance improvement in about five years.
1 3 [Posted by: lol123  | Date: 11/27/13 10:46:54 PM]
What I see is the performance gap closing and the value/$ getting higher and higher.

Seams like AMD will deliver! Hiiiiiihaaaa!
1 0 [Posted by: MHudon  | Date: 11/28/13 01:23:51 PM]
At what point will we be able to replace an 8core piledriver chip (8350) with a 4 core steamroller, such as this, and not get a perf drop? It looks like stand alone CPU chips are done for AMD.
0 0 [Posted by: KeyBoardG  | Date: 12/04/13 11:18:54 AM]

I'm a wait and see type of person because richalnd was suppose to do the same thing and it fell far short of what articles were proclaiming it would do. same is said about haswell over ivy bridge and the cpu performance over ivy bridge was hardly noticeable.
0 0 [Posted by: SteelCity1981  | Date: 11/27/13 09:47:25 PM]

show the post
1 4 [Posted by: lol123  | Date: 11/27/13 10:45:30 PM]
- collapse thread

This 20% is ultimate success for AMD
1 0 [Posted by: Tristan  | Date: 11/28/13 08:36:34 AM]
This is ultimate success for AMD
1 0 [Posted by: Tristan  | Date: 11/28/13 08:50:50 AM]
Pricewise, you're totally wrong! You're not even funny...

Please give a kiss to your wife and my childrens tonight will ya!
0 0 [Posted by: MHudon  | Date: 11/28/13 01:25:23 PM]

show the post
0 3 [Posted by: Tukee44  | Date: 11/27/13 11:05:42 PM]
- collapse thread

from what ivybridge to what haswell, with what display in each and how many cells in each case.
1 0 [Posted by: Yorgos  | Date: 11/28/13 08:33:39 AM]

Even in the techgeek community, AMD is not getting their message across. Differences between Intel's CPU architecture and Steamroller aside, the die area allocation strategy is also different. AMD has decided to allocate more silicon to GPU than Intel does. Not because AMD believes that everyone needs all those millions of transistors to play games, although that is a very clear advantage. No, the reason is for GPGPU compute or HSA. Instead of comparing AMD's Steamroller CPU with previous generations or with Intel's CPUs, you have to look at the whole die processing capability. For AMD there are two problems to overcome: firstly, for now at least, they need applications optimised to take advantage of GPU compute, and secondly, they have to re-educate the reviewers and consumers to compare the entire package as APUs rather than simply CPUs. With HUMA they hope to make it easier for software to take advantage of the GPU compute capability and over the next 2 or 3 iterations they will hopefully reach a point where all software can benefit seemlessly and transparently from the full compute performance of the APU. Of course, ARM etc are treading the same path, and Intel will inevitably follow along too.
4 2 [Posted by: mjv1121  | Date: 11/28/13 05:00:01 AM]
- collapse thread

Precisely, benchmarks such as x264 GPU compute optimised encoding should show AMD at least double the speed of Intel's best.
1 1 [Posted by: linuxlowdown  | Date: 11/28/13 10:15:12 AM]
Except x264 has limited use in GPU compute. Only intra frame prediction is used in GPU compute.

Not to mentioned that x264 is heavily integer based, something GPU sucks at doing. GPU is floating point based.

As for AMD being the double speed of Intel, unless AMD follows Intel step of adding dedicated H.264 ASIC encoder within its iGPU, it won't beat Intel Quick Sync.
0 0 [Posted by: trumpet-205  | Date: 11/28/13 11:22:45 AM]
You're probably half right. But could you explain this to me? -

"Several weeks ago AMD dropped a bombshell: x264 and Handbrake would both feature GPU acceleration, largely via OpenCL"

"GPU acceleration in Handbrake comes via three avenues: DXVA support for GPU accelerated video decode, OpenCL/GPU acceleration for video scaling and color space conversion, and OpenCL/GPU acceleration of the lookahead function of the x264 encoding process."

"A quad-core Sandy Bridge goes from being 73% faster down to 7% faster than Trinity. Ivy Bridge on the other hand goes from being 2.15x the speed of Trinity to a smaller but still pronounced 29.6% lead."

"This truly is the holy grail for what AMD is hoping to deliver with heterogeneous compute in the short term. The Sandy Bridge comparison is particularly telling. What once was a significant performance advantage for Intel, shrinks to something unnoticeable. If AMD could achieve similar gains in other key applications, I think more users would be just fine in ignoring the CPU deficit and would treat Trinity as a balanced alternative to Intel. The Ivy Bridge gap is still more significant but it's also a much more expensive chip, and likely won't appear at the same price points as AMD's A10 for a while."

I can't wait to see Kaveri vs Haswell Handbrake x264 benches.

Date: May 15, 2012.
0 1 [Posted by: linuxlowdown  | Date: 11/29/13 05:56:55 AM]

It says "Kaveri 2.0" and "Steamroller B".

Why is everyone assuming this is Kaveri launching January 2014?

Also.... unless AMD state it's 20% instruction per clock (IPC) improvement, the performance increase over Richland may largely just be a simple clock rate improvement provided by the 32nm - 28nm transition.
0 0 [Posted by: JBG  | Date: 11/28/13 05:23:45 AM]
- collapse thread

or perf/Watt, thx to 28nm. We know these type of improvements from Buldozer propaganda.
1 0 [Posted by: Tristan  | Date: 11/28/13 08:38:50 AM]
The frequency will be lower.(I think it was in the apu13 slides) Achieving a 20% boost with the lower frequency means that at equal frequency(and core count) this arch sould give like 30% IPC impprovement over richland(which fits all the rumors and leaked benchmarks).

Kaveri 2.0/ Steamroller B is Kaveri it's 2.0/B because when the new design team "came to power" they trashed the entire steamroller arch and redrew almost everything. From what I heard there were major flaws in the old desing that Raja and his team just threw right out the window, this also explains the delay in launching it. My impression is that if they would have gone with the previous designs they would again have been the laughing stock of the entire world...

The node reduction doesn't allways help, see intels transition to 22nm and the heat/scaling problems and laughable performance improvements(in the first generation). It's much better to use a fully mature node for big cores then to use an experimental/new one(that's why intel postponed 14nm shrink(twice already) they obvioulsy hit a snag there and it seems they will not have any 14nm big cores before 2015 or even later...

Apparently this chip will also have a new memory controller so these apu's might be the first to not be memory starved - I guess we need to wait and see...

I guess we'll just have to wait for january (I know I am, I want this thing in my next rig)
2 0 [Posted by: tcube  | Date: 11/28/13 12:56:59 PM]
Steamroller B is old news. Hmmm. I must of missed the memo:
0 0 [Posted by: JBG  | Date: 11/28/13 10:38:49 PM]


Add your Comment

Related news

Latest News

Wednesday, November 5, 2014

10:48 pm | LG’s Unique Ultra-Wide Curved 34” Display Finally Hits the Market. LG 34UC97 Available in the U.S. and the U.K.

Wednesday, October 8, 2014

12:52 pm | Lisa Su Appointed as New CEO of Advanced Micro Devices. Rory Read Steps Down, Lisa Su Becomes New CEO of AMD

Thursday, August 28, 2014

4:22 am | AMD Has No Plans to Reconsider Recommended Prices of Radeon R9 Graphics Cards. AMD Will Not Lower Recommended Prices of Radeon R9 Graphics Solutions

Wednesday, August 27, 2014

1:09 pm | Samsung Begins to Produce 2.13GHz 64GB DDR4 Memory Modules. Samsung Uses TSV DRAMs for 64GB DDR4 RDIMMs

Tuesday, August 26, 2014

10:41 am | AMD Quietly Reveals Third Iteration of GCN Architecture with Tonga GPU. AMD Unleashes Radeon R9 285 Graphics Cards, Tonga GPU, GCN 1.2 Architecture