Jaguar (microarchitecture)

Jaguar - Family 16h

Produced	From Mid-2013 to present
Common manufacturer(s)	AMD
Min. feature size	28 nm
Instruction set	AMD64 (x86-64)
L1 cache	64 KB per core
L2 cache	1 MB to 2 MB shared
Socket(s)	Socket AM1 Socket FT3 (BGA-769)
Predecessor	Bobcat - Family 14h
Successor	Puma - Family 16h (2nd-gen)
Core name(s)	Kabini Temash

The AMD Jaguar Family 16h is a low-power SoC microarchitecture designed by AMD, succeeding the Bobcat Family microarchitecture in 2013 and being succeeded by AMD's Puma architecture in 2014. It is two-way superscalar and capable of out of order execution. It forms the basis for AMD's Semi-Custom Business Unit and four product families: Kabini aimed at notebooks and mini PCs, Temash aimed at tablets, Kyoto aimed at micro-servers, and the G-Series aimed at embedded applications. Both the PlayStation 4 and the Xbox One use chips based on the Jaguar microarchitecture.

Design

• 32 KiB instruction + 32 KiB data L1 cache per core, L1 cache includes parity error detection
• 16 way, 1-2 MiB unified L2 cache shared by two or four cores, L2 cache is protected from errors by the use of error correcting code
• Out-of-order execution and Speculative execution
• Integrated memory controller
• Two-way integer execution
• Two-way 128-bit wide floating-point and packed integer execution
• Integer hardware divider
• Consumer processors support 2 DDR3L DIMMs in one channel at frequencies up to 1600 MHz
• Server processors support 2 DDR3 DIMMS in one channel at frequencies up to 1600 MHz with ECC
• As a SoC (not just an APU) it integrates Fusion controller hub
• Jaguar does not feature clustered multi-thread (CMT), meaning that execution resources are not shared between cores

Instruction set support

The Jaguar core has support for the following instruction sets and instructions: MMX, SSE, SSE2, SSE3, SSSE3, SSE4a, SSE4.1, SSE4.2, AVX, F16C, CLMUL, AES, BMI1, MOVBE (Move Big-Endian instruction), XSAVE/XSAVEOPT, ABM (POPCNT/LZCNT), and AMD-V.

Improvements over Bobcat

• Over 10% increase in clock frequency
• Over 15% improvement in instructions per clock (IPC)
• Added support for SSE4.1, SSE4.2, AES, CLMUL, MOVBE, AVX, F16C, and BMI1
• Up to 4 CPU cores
• L2 cache is shared between cores
• FPU datapath width increased to 128 bit
• Added hardware integer divider
• Enhanced cache prefetchers
• Doubled bandwidth of load-store units
• C6 and CC6 low power states with lower entry and exit latency
• Smaller, 3.1 mm² area per core
• Integrated Fusion controller hub (FCH)
• Video Coding Engine

Processors

See also: List of AMD Accelerated Processing Unit microprocessors and List of AMD mobile microprocessors

Consoles

• ¹ Unified Shaders : Texture Mapping Units : Render Output Units
• ² Pixel fillrate is calculated as the number of ROPs multiplied by the base core clock speed.
• ³ Texture fillrate is calculated as the number of TMUs multiplied by the base core clock speed.

Device	CPU			GPU					Memory				Special features
	Cores	Frequency	L2 Cache	Cores1	Frequency	GFLOPS	Pixel Fillrate (GP/s)2	Texture Fillrate (GT/s)3	Amount	Bus Width (bit)	Bus Type	Bandwidth (GB/s)
Sony PS4	dual 4 core modules	1.6 GHz	2 x 2 MB	1152:72:32	800 MHz	1840	25.6	57.6	8 GB	256	GDDR5	176	8 ACEs in the GPU and additional modules
Microsoft Xbox One	dual 4 core modules	1.75 GHz	2 x 2 MB	768:48:16	853 MHz	1310	13.6	40.9	8 GB	256	DDR3	68	2 ACEs in the GPU and 32MB ESRAM 204 GB/s

Desktop

SoCs using Socket AM1:

Model	CPU			GPU			TDP	Memory	Socket
	Cores	Frequency	L2 Cache	Model	Cores	Frequency
Athlon 5350	4	2.05 GHz	2 MB	Radeon R3	128:?:?	600 MHz	25 W	DDR3-1600	AM1
Athlon 5150		1.6 GHz
Sempron 3850		1.3 GHz				450 MHz
Sempron 2650	2	1.45 GHz	1 MB			400 MHz		DDR3-1333

Desktop/Mobile

Target segment	Model	CPU				GPU				TDP	Memory	Turbo Core
		Cores	Frequency	Max. Turbo	L2 Cache	Model	Config.	Frequency	Turbo
Notebooks /Mini-PCs	A6-5200	4	2.0 GHz	N/A	2 MB	HD 8400	128:16:4	600 MHz	N/A	25 W	DDR3L-1600	No
	A4-5100		1.55 GHz			HD 8330		500 MHz		15 W
	A4-5000		1.5 GHz			HD 8330		500 MHz
Notebooks	E2-3000	2	1.65 GHz		1 MB	HD 8280		450 MHz
	E1-2500		1.4 GHz			HD 8240		400 MHz			DDR3L-1333
	E1-2100		1.0 GHz			HD 8210		300 MHz		09 W
Tablets	A6-1450	4		1.4 GHz	2 MB	HD 8250			400 MHz	08 W	DDR3L-1066	Yes
	A4-1350			N/A		HD 8210			N/A		DDR3-1066	No
	A4-1250	2			1 MB	HD 8210					DDR3L-1333
	A4-1200					HD 8180		225 MHz		03.9 W	DDR3L-1066

Server

Model	CPU			GPU		TDP	Memory
	Cores	Frequency	L2 Cache	Cores	Frequency
Opteron X2150	4	1.1–1.9 GHz	2 MB	128	266–600 MHz	11–22 W	DDR3-1600 ECC
Opteron X1150		1.0–2.0 GHz		N/A		09–17 W

^ CPU and GPU frequencies are adjustable in BIOS.

Embedded

Model	CPU			GPU			TDP	Memory
	Cores	Frequency	L2 Cache	Model	Config.	Frequency
GX-420CA	4	2.0 GHz	2 MB	HD 8400E	128:16:4	600 MHz	25 W	DDR3-1600 ECC
GX-416RA		1.6 GHz		N/A			15 W
GX-415GA		1.5 GHz		HD 8330E	128:16:4	500 MHz
GX-217GA	2	1.65 GHz	1 MB	HD 8280E		450 MHz
GX-210HA		1.0 GHz		HD 8210E		300 MHz	09 W	DDR3-1333 ECC
GX-210JA				HD 8180E		225 MHz	06 W	DDR3-1066 ECC

External links

• Software Optimization Guide for Family 16h Processors
• Jaguar AMD’s Next Generation Low Power x86 Core at Hot Chips 24
• Slides about the design of jaguar presented at ISSCC 2013
• Jaguar presentation (video) at ISSCC 2013
• Discussion initiated on RWT forums by Jeff Rupley, Chief Architect of the Jaguar core
• BKDG for Family 16h Models 00h-0Fh Processors
• Revision Guide for Family 16h Models 00h-0Fh Processors