COOL Interconnect COOL Interconnect Low Power Interconnection - - PowerPoint PPT Presentation

cool interconnect cool interconnect low power
SMART_READER_LITE
LIVE PREVIEW

COOL Interconnect COOL Interconnect Low Power Interconnection - - PowerPoint PPT Presentation

COOL Interconnect COOL Interconnect Low Power Interconnection Technology Low Power Interconnection Technology for Scalable 3D LSI Design for Scalable 3D LSI Design Marco Chacin, Hiroyuki Uchida, Michiya Hagimoto, Takashi Miyazaki, Takeshi


slide-1
SLIDE 1

COOL Chips XIV COOL Chips XIV

COOL Interconnect Low Power Interconnection Technology for Scalable 3D LSI Design COOL Interconnect Low Power Interconnection Technology for Scalable 3D LSI Design

Marco Chacin, Hiroyuki Uchida, Michiya Hagimoto, Takashi Miyazaki, Takeshi Ohkawa, Rimon Ikeno, Yukoh Matsumoto TOPS Systems Corporation Fumito Imura, Motohiro Suzuki, Katsuya Kikuchi, Hiroshi Nakagawa, Masahiro Aoyagi National Institute of Advanced Industrial Science and Technology

slide-2
SLIDE 2

COOL Chips XIV COOL Chips XIV

1. Background 2. 3D LSI

  • Current 3D LSI technology

3. Heterogeneous Multichip 3D LSI: COOL System 4. COOL Interconnect

  • Protocol
  • Interface Circuit Design

5. Test Chip 6. Summary 7. Future work

Outline Outline

slide-3
SLIDE 3

COOL Chips XIV COOL Chips XIV

Background Background

IT機器における国内総電力消費量と CO2排出量予測(2006~2050)

A breakthrough in the reduction of LSI energy consumption is needed The key is not in the miniaturization, is in the efficient use of its transistors

Reasons f Reasons for

  • r the

the increase increase of

  • f

energy consumptio energy consumption ・Very high number of units ・Increase in features ・Amount Amount of

  • f data an

data and d Communi mmunication s cation spee eeds ds

Utilization Wall

・the number of active transistors within any given semiconductor chip has exponentially decreased accordingly with its miniaturization <Multicore miniaturization>

Leakage current increase due to semiconductor miniaturization

Consumption in IT devices ・Appliances (IT) ・Network switches ・Data Center

Low power technology ・Clock gating ・Dual Vt ・Static back-gate bias ・Dynamic back-gate bias ・Power cut-off ・DVFS technology ・Low voltage

Consumption requirement in mobile devices

(800mW-1000mW)

Always increasing power consumption

Classical scaling Leakage- limited Scaling Transistors S2 S2 Frequency S S Consumption (Cap) 1/S 1/S Consumption (Vdd) 1/S2 ~1 Utilization 1 1/S2

4 Cores@2GHz 65nm 8 Cores@2GHz 32nm 4 Cores@4GHz 32nm

・Can’t increase freq ・Only 8 cores are used

・ ・

【Challenges】 ・Energy consumption limit ・Increase in leakage current ・Operation frequency limit Dark Silicon

slide-4
SLIDE 4

COOL Chips XIV COOL Chips XIV

Different technologies like common circuit logic, memory, etc. can be freely stacked to surpass Moore’s Law

More than simple integration, the processor architecture also change to mitigate the Utilization Wall, the Memory Wall, the Power Wall and the ILP Wall

【Advantages of a scalable 3D LSI】

-Low power consumption -High level integration -High performance -Short-to-market cycle -Flexibility -Low cost

Scalable 3D LSI

Through-silicon via (TSV)

Ideal

Stacked LSI based on application requirements I/O Memory Processor 129 chips stacked in a 3D LSI ISSCC2010 Keio U/ Prof. Kuroda 8Gb 3D DRAM ISSCC2009 Samsung 4 layer stacked memory cell ISSCC2010 Unity Semiconductor

Standardization of inter- chip communication ・Easy to design ・Easy to dest

Current state

3D LSI 3D LSI

slide-5
SLIDE 5

COOL Chips XIV COOL Chips XIV

Through-silicon via(TSV)

3D integration technology that uses microbump bonding and adhesive injection Main stacking methods:

  • Wafer on Wafer
  • Wafer on Chip
  • Chip on Chip

Memory I/O Processor

Current 3D LSI Technology Current 3D LSI Technology

slide-6
SLIDE 6

COOL Chips XIV COOL Chips XIV

Low power heterogeneous Multichip/Multicore Microprocessor architecture designed for 3D stack chips Distributed software technology that effectively drives the hardware Scalable interface technology for low power 3D LSI chips

Interface logic Memory

Distributed processing Heterogeneous Multichip/Multicore

Shared bus

Peripheral Chip

Heterogeneous Multichip 3D LSI: Heterogeneous Multichip 3D LSI: COOL System COOL System

slide-7
SLIDE 7

COOL Chips XIV COOL Chips XIV

COOL Interconnect COOL Interconnect

Physical layer Physical layer Circuit layer Circuit layer

Multilayer IP

Logic layer Logic layer

Fabrication Process of 1600 TSV in 40x40 array

D Q Q D Q D

Interface driver Scalable bus interface logic

COOL Interconnect Interface Chip ID Bus arbiter Data converter I/O (C) I/O (B) I/O (A)

COOL Interconnect Technology

slide-8
SLIDE 8

COOL Chips XIV COOL Chips XIV

Interface logic Shared bus (1600 TSV)

COOL Interconnect COOL Interconnect

COOL Interconnect Technology

slide-9
SLIDE 9

COOL Chips XIV COOL Chips XIV

Interconnection type:Shared bus Communication model:Multi-Master / Multi-Slave (No of Master chips:Max 16, Slaves:Max 16) Protocol:3-stage pipeline -S0 : Request phase -S1 : Address / command transmission phase -S2 : Data transfer phase COOL Interconnect Technology

Protocol (1/2) Protocol (1/2)

slide-10
SLIDE 10

COOL Chips XIV COOL Chips XIV

Data transfer throughput :1 cycle Data transfer latency :2 cycles Arbitration model: Dual distributed -Bus-Master -Bus-Slave Transfer mode:single, burst、lock

Through TSV’s these signals reach all chips within the 3D LSI stack and communicate using an implemented common interface logic

COOL Interconnect‘s data signals consist of 1024 out of 1600 TSV COOL Interconnect‘s data signals consist of 1024 out of 1600 TSV

COOL Interconnect Technology

Signal type Description bus_clock

  • Bus clock

bus_reset

  • Bus reset

bus_req_m(15:0) S0 Request Master bus_req_s(15:0) S0 Request Slave bus_s_stat(15:0) S0 Slave Status bus_command(3:0) S1 Command bus_address(31:0) S1 Address bus_destination(3:0) S1 Destination bus_lock S1 Bus lock bus_data(1023:0) S2 Data bus_data_valid(6:0) S2 Data Valid bus_ready S2 Data Ready bus_error S2 Data Error

Protocol (2/2) Protocol (2/2)

slide-11
SLIDE 11

COOL Chips XIV COOL Chips XIV

0.562 0.526 0.512 0.494 0.493 0.498 0.504 0.385 0.414 0.427 0.434 0.449 0.460 0.470 0.480 0.664 2.97 1.59 1.46 0.73 0.48 0.36 0.28 5.90

0.00 0.10 0.20 0.30 0.40 0.50 0.60 0.70 X1 X2 X3 X4 X8 X12 X16 X20

Driver Size Power Consumption [W@50MHz/1600pins]

0.0 2.0 4.0 6.0 8.0 10.0 12.0 14.0

Noise [VnS]

Noise Receiver cirucit consumption Driver and reciver circuit consumption D Q Q D Q D

x12 Interface driver

Cool Interconnect DDR‐2(1) Circ. 0.67 mm2 8.64mm2 Area I/O 4 mm2 4 mm2 Consumpt 246.5 mW (max)(2) 576 mW Bandwidth 6.4 GB/s 6.4 GB/s Freq 50 MHz 400 MHz

Driver circuit optimization through simulation Driver circuit optimization through simulation

(1) DDR-2 PC-6400 (2) Estimation

Comparison with current interconnection technology

COOL Interconnect Technology

Interface Circuit Design Interface Circuit Design

slide-12
SLIDE 12

COOL Chips XIV COOL Chips XIV

・ Bandwidth: 6.4GByte/sec @50MHz ・ Inter-chip connection: 1600 TSV ・ Model:Multi-Master / Multi-Slave Shared bus

Master(CPU): Up to 16 chips Slave(Memory/IO): Up to 16 Chips

Technology TSMC0.25um ・Interface are:4.67mm2

Interface circuit: 0.67mm2 (14%) TSV area: 4mm2 (86%) Logic TEG I/O pad TEG TEG TEG 8.3mm 6mm TSV array

2.16mm 2.16mm

Scalable 3D LSI chips interface verification and test

(a) Conventional I/O area (4mm2) (b) TSV area (4mm2) 5mm 5mm 0.2mm 5mm 5mm

2mm

Same area

Test Chip Test Chip

slide-13
SLIDE 13

COOL Chips XIV COOL Chips XIV

  • Scalability of 3D LSI designs will play a crucial role in

extending the silicon road-map for the next generation of microprocessor IPs

  • We presented a working model of a standard

interconnection bus for 3D LSI units that provides scalability as well as high performance and low power consumption

  • Implementing a standard way of inter-chip communication

common challenges can be avoided or completely solved.

  • COOL Interconnect provides the necessary versatility to

design 3D LSI efficiently made for the application with a short-to-market cycle and easy fabrication process

Summary Summary

slide-14
SLIDE 14

COOL Chips XIV COOL Chips XIV

  • Increased power densities that can result from

carelessly placing one computational block over another

  • Find ways to collaborate with open innovation

business partners to create a new standard

Future Work Future Work

slide-15
SLIDE 15

COOL Chips XIV COOL Chips XIV This research is supported by the New Energy and Industrial Technology Development Organization (NEDO)

Cool System Innovator Cool System Innovator