Bookmark and Share


Advanced Micro Devices last week unveiled the first details about its next-generation Jaguar micro-architecture for low-power and ultra low-power devices. The new architecture promises to enable feature-rich multi-core x86 microprocessors with extremely low consumption of energy, but with improved performance compared to today's x86 micro-architectures. AMD hopes to enter new markets with ULP chips powered by Jaguar.

In order to significantly improve performance of Jaguar-based over the Bobcat, AMD decided to go into virtually all logical directions: increase the amount of cores, increase clock-speed, add support for modern instructions, increase amount of executed instructions per clock (IPC). AMD also decided to improve power efficiency thro clock gating and unit redesign in a bid to ensure lower idle power consumption compared to existing low-power designs. The first chips to utilize Jaguar are code-named Kabini and Temash that will be made using 28nm fabrication process, but going forward it is likely that future Jaguar-based products will be produced at 20nm nodes. 


The amount of cores in one Jaguar-powered accelerated processing unit can be up to four, as suggested by design. In this case, the 2MB shared cache unit (second level cache) will be shared between the cores, but will be divided into four 512 KB data banks, which can independently be switched off to save energy.

In order to boost clock-speed by 10% compared to today's Bobcat-powered chips, Jaguar micro-architecture features longer pipeline.

The third major improvements incorporated into the architecture is the up-to-date set of supported instructions, as reported, Jaguar features SSE4.1, SSE4.2, AES, PCLMUL, AVX, BMI, F16C as well as MOVBE.

In a bid to boost IPC by 15%, Jaguar introduces 128-bit floating point unit (FPU) with enhancements and double-pumping to support 256-bit AVX instructions as well as  an innovative integer unit with new hardware divider, larger schedulers and more out-of-order resources.

Finally, AMD added new technologies to trim power consumption; for example, AMD implemented a new CC6 state with even deeper energy economy, with each core able to go there independently.

Each Jaguar core implemented in 28nm process technology will be 3.1mm2 large, which is smaller compared to Bobcat core in 40nm, which is 4.9mm2. In order to improve performance, AMD will try to increase the amount of cores as well as clock-speed, whereas in order to offer something competitive for tablets (in Temash form), AMD will leave clock-speeds on current levels and will only put two cores into system-on-chip.

Tags: AMD, Jaguar, Bobcat, x86


Comments currently: 38
Discussion started: 09/05/12 12:33:51 AM
Latest comment: 01/16/14 04:09:30 PM
Expand all threads | Collapse all threads


show the post
0 5 [Posted by: keysplayer  | Date: 09/05/12 12:33:51 AM]
- collapse thread

Jaguar will not be found in any 40nm parts.
4 0 [Posted by: quasi_accurate  | Date: 09/05/12 12:56:22 AM]

show the post
0 5 [Posted by: eleman  | Date: 09/05/12 12:57:55 AM]
- collapse thread

By your logic the ultimate processor would have 1 stage long pipeline (ie no pipelining), but of course they don't. Its so much more complicated than your simple explanation. Longer pipelines reduces the amount of gates required to implement each stage (as its smaller) which reduces the propagation delay of each step which increases maximum frequency. This is independent of IPC.

The only area a longer pipeline reduces performance is in branch misprediction, as the pipeline needs to stall as it gets flushed. However, in this case and in many cases, you can get a sizeable increase in max frequency with only a minor increase in branch misprediction penalty. It says in the article that they are getting 10% higher frequency from one stage longer pipeline, so branch misprediction has gone from 13 to 14 cycles, or a 7.7% bigger penalty. Given a branch prediction rates are ~85-90% even in the Pentium days, that 7.7% increase only occurs on 10-15% of branches (which make up ~25-33% of average code).

Long story short, 10% higher frequency, smaller than 10% penalty = net gain in performance. Its about finding the optimal point between reducing stage propagation delays (the returns get smaller the more stages you add) and the increase in branch mis-prediction (which increases linearly with pipeline length). Prescott didn't work for many reasons, not just pipeline length (which has 31 stages long, far longer than Jaguar anyway). Just saying longer pipeline = no improvement without even a basic analysis of the architecture is just incorrect.
8 2 [Posted by: genie  | Date: 09/05/12 05:37:42 AM]
Hmm... Your explanation makes sense. Thank you.
1 0 [Posted by: eleman  | Date: 09/06/12 12:27:46 AM]
All good man, I think Intel's 'megahertz at all costs' strategy during the P4 days jaded most people regarding changing architecture to support higher freq's. But like everything, the reality is always more complicated than at first glance.
0 0 [Posted by: genie  | Date: 09/06/12 11:44:16 PM]
AMD's K8 had longer pipeline than AMD K7, yet it featured _much_better IPC.
5 0 [Posted by: Anton  | Date: 09/05/12 09:35:16 AM]

Where is a "revolution" Bulldozer module design?
Where is shared FP and other resources?
Dirk, Rory is betraying your legacy which you had been bearing so many years.
1 3 [Posted by: Azazel  | Date: 09/05/12 05:49:21 AM]
- collapse thread

Jaguar is an evolution of the Bobcat core, which itself was a completely different design from Bulldozer and was aimed at low-power applications.
4 1 [Posted by: quasi_accurate  | Date: 09/05/12 08:45:43 AM]
Do low-power applications need powerful FP and AVX?
Would has it not been better to use sharing for FP and AVX? Would it?
Shared Fetch and Decode would have been of some use to lower power consumption in low performance cpus. Wouldn't it?
1 2 [Posted by: Azazel  | Date: 09/05/12 10:25:53 AM]

Most improvements in tech/life are evolutionary, not revolutionary. Pipeline length by itself does not determine a CPU's overall performance and is only one important factor. Jaguar is for 28nm process not 40nm. It's reported that GloFo will produce these chips as TSMC still lacks sufficient 28nm capacity.
6 2 [Posted by: beenthere  | Date: 09/05/12 06:22:12 AM]

Whilst AMD are pretty much a niche CPU company compared to the glory days of Opteron these days, their APU's are great in the space they were designed for

ppl saying AMD is finished/die would have a very rude awakening should this ever happen because then we all would be paying a lot more for those precious intel CPU's
3 1 [Posted by: alpha0ne  | Date: 09/07/12 12:36:26 AM]

Not the most exciting product but at least there is progress. One can only imagine the performance increase once they rework the L2 cache. In Bobcat the L2 cache isn't unified nor does it work at the same clocks as the rest of the cores but instead runs at half the core clocks. The gains in Jaguar very easily suggest that the L2 hasn't changed in that regard. Pity that had the L2 cache been designed like it should have it would have easily compeated with higher end products.
0 0 [Posted by: nforce4max  | Date: 01/15/13 12:30:50 PM]

This Chip has won AMD the Playstation 4 and Xbox 720 contracts..

Both consoles will have AMD Jaguar Micro-Architecture CPU's and Graphics core next GPU's... looks like they pulled off one hell of a CPU

Ohh and the product they will be selling is a custome "8" core Jaguar CPU~
0 0 [Posted by: vid_ghost  | Date: 01/24/13 04:06:19 PM]

Sony and Microsoft both made a mistake on choice of hardware! What they should of used is:

Sony should of went with Nvidia Project Denver 20nm (ARMv8 + Maxwell GPU)The PS4 would of been 2x faster then AMD Jaguar crap! Right now Nvidia has the fastest GPU for PC on Earth! GeForce GTX TITAN.

Microsoft should of used Intel 14nm Broadwell (HD Graphics 5000) Eighth generation GPU! Also would have been 2x faster then what AMD has to offer MS Xbox!

AMD has nothing to compare with 14nm Broadwell (Intel 5th gen CPU+GPU.
0 0 [Posted by: ToxicTaZ  | Date: 02/21/13 02:38:51 PM]
- collapse thread

There is one small problem with your theory.THERE IS NO 20NM YET.
0 0 [Posted by: saneblane  | Date: 02/27/13 01:26:19 PM]
Derp is all I got from that. Dude for how many years has AMD and NVidia been trading blows in GPU's? and since when does Intel know how to make a gpu? Yes intel processors are faster but graphics suck balls dude always have. Check out some benchies with piledriver in multi thread compete with ivy bridge. Yes it uses more power but still competes. Before you go ohh intel is the best amd sucks think about where amd is competing in. Think about the support amd cpu/gpus are gonna get after this. think about the competition, the price drops etc. Think about if both consoles are using 8 core x86/x64 cpus think about finally having developers learning about and implementing true multithread in games. Think about the fact that they are running X86/X64 architecture, really think about it (don't just ohh amd is sh*t so it doesn't matter). Think about how easy it will be to port games over to PC. No more skyrims and cods and gta4s and sh*t like that. Think about the fact that this chip won (apparently) both contracts (ps4 def but only rumors on 720) and its in an APU. Gotta be a half decent chip. Stop being a troll and realise this is a great step for intel/NVidia users aswell as will help unify instructions on all platforms. All gaming done on one hardware set. Thankyou AMD you have done a great job for the gaming industry. I hope this works out well for you!
0 0 [Posted by: Corey  | Date: 03/03/13 08:17:55 AM]
Sometimes it is interesting to read all the comments regarding this CPU/GPU thing, whos bested who and soo on, depite my lazyness on reading. I myself was a fan of AMD but now accepted both developers as part of my dayly business needs. Sure some might say AMD craps or likewise but I guess all this might give an idea of understanding on how global market and demand works and to improvise current trend and model so there is choice and direction on where to look at.. All parties benefiting from eachothers... furthermore they all share the same piece of cake. And definately have better ideas on how to get their business running for years to come.
0 0 [Posted by: Norman78  | Date: 01/16/14 04:09:30 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