Day 26: November 11, 2011 Memory Overview

Slides:



Advertisements
Similar presentations
COEN 180 SRAM. High-speed Low capacity Expensive Large chip area. Continuous power use to maintain storage Technology used for making MM caches.
Advertisements

Semiconductor Memory Design. Organization of Memory Systems Driven only from outside Data flow in and out A cell is accessed for reading by selecting.
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 24: November 4, 2011 Synchronous Circuits.
COEN 180 DRAM. Dynamic Random Access Memory Dynamic: Periodically refresh information in a bit cell. Else it is lost. Small footprint: transistor + capacitor.
Introduction to CMOS VLSI Design Lecture 13: SRAM
11/29/2004EE 42 fall 2004 lecture 371 Lecture #37: Memory Last lecture: –Transmission line equations –Reflections and termination –High frequency measurements.
Introduction to CMOS VLSI Design SRAM/DRAM
S. Reda EN160 SP’07 Design and Implementation of VLSI Systems (EN0160) Lecture 31: Array Subsystems (SRAM) Prof. Sherief Reda Division of Engineering,
Modern VLSI Design 2e: Chapter 6 Copyright  1998 Prentice Hall PTR Topics n Memories: –ROM; –SRAM; –DRAM. n PLAs.
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 27: November 5, 2014 Dynamic Logic Midterm.
Lecture 19: SRAM.
Parts from Lecture 9: SRAM Parts from
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 28: November 15, 2013 Memory Periphery.
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 30: November 12, 2014 Memory Core: Part.
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 28: November 16, 2012 Memory Periphery.
FPGA-Based System Design: Chapter 3 Copyright  2004 Prentice Hall PTR Topics n Latches and flip-flops. n RAMs and ROMs.
Modern VLSI Design 4e: Chapter 6 Copyright  2008 Wayne Wolf Topics Memories: –ROM; –SRAM; –DRAM; –Flash. Image sensors. FPGAs. PLAs.
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 27: November 14, 2011 Memory Core.
Z. Feng MTU EE4800 CMOS Digital IC Design & Analysis 12.1 EE4800 CMOS Digital IC Design & Analysis Lecture 12 SRAM Zhuo Feng.
Advanced VLSI Design Unit 06: SRAM
CSE477 L24 RAM Cores.1Irwin&Vijay, PSU, 2002 CSE477 VLSI Digital Circuits Fall 2002 Lecture 24: RAM Cores Mary Jane Irwin ( )
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 24: November 5, 2010 Memory Overview.
 Seattle Pacific University EE Logic System DesignMemory-1 Memories Memories store large amounts of digital data Each bit represented by a single.
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 26: October 31, 2014 Synchronous Circuits.
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 22: November 1, 2010 Dynamic Logic.
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 28: November 16, 2011 Memory Periphery.
Washington State University
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 28: November 7, 2014 Memory Overview.
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 28: November 8, 2013 Memory Overview.
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 20: October 25, 2010 Pass Transistors.
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 27: November 14, 2012 Memory Core: Part.
Penn ESE370 Fall DeHon 1 ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 30: November 21, 2012 Crosstalk.
Day 3: September 10, 2012 Gates from Transistors
Memories.
ESE532: System-on-a-Chip Architecture
Lecture 19: SRAM.
Dynamic Random Access Memory (DRAM) Basic
Day 29: November 18, 2011 Dynamic Logic
Latches and Flip-flops
Hakim Weatherspoon CS 3410 Computer Science Cornell University
EE345: Introduction to Microcontrollers Memory
Prof. Kavita Bala and Prof. Hakim Weatherspoon
Day 33: November 19, 2014 Crosstalk
Day 22: October 23, 2013 Pass Transistor Logic
Day 22: October 31, 2011 Pass Transistor Logic
Day 19: October 24, 2011 Ratioed Logic
ESE534: Computer Organization
Day 31: November 23, 2011 Crosstalk
Day 26: November 1, 2013 Synchronous Circuits
Day 23: November 2, 2012 Pass Transistor Logic: part 2
Day 29: November 11, 2013 Memory Core: Part 1
Digital Logic & Design Dr. Waseem Ikram Lecture 40.
Day 27: November 6, 2013 Dynamic Logic
Day 30: November 13, 2013 Memory Core: Part 2
Day 25: November 7, 2011 Registers
Day 21: October 29, 2010 Registers Dynamic Logic
Day 3: September 4, 2013 Gates from Transistors
Day 5: September 17, 2010 Restoration
Day 14: October 8, 2010 Performance
Day 18: October 20, 2010 Ratioed Logic Pass Transistor Logic
Overview Last lecture Digital hardware systems Today
Day 15: October 13, 2010 Performance: Gates
Sequential Logic.
ESE534: Computer Organization
Day 29: November 10, 2014 Memory Core: Part 1
Levels in Processor Design
Day 16: October 17, 2011 Performance: Gates
Day 26: November 10, 2010 Memory Periphery
Day 25: November 8, 2010 Memory Core
ESE532: System-on-a-Chip Architecture
Presentation transcript:

Day 26: November 11, 2011 Memory Overview ESE370: Circuit-Level Modeling, Design, and Optimization for Digital Systems Day 26: November 11, 2011 Memory Overview Penn ESE370 Fall2011 -- DeHon

Today Memory Project 2 is on this (posted) Motivation Organization Basic components Optimization concerns Project 2 is on this (posted) Not looked at midterms  try for Monday Penn ESE370 Fall2011 -- DeHon

Know how to store state Penn ESE370 Fall2011 -- DeHon

Register Storage Could just put together a large number of registers Concerns? Penn ESE370 Fall2011 -- DeHon

Concerns? Large number of wires May want to store for many cycles Could determine area 5l wire pitch  how wide? May want to store for many cycles Penn ESE370 Fall2011 -- DeHon

Usage Scenario How many state values read on each cycle? Penn ESE370 Fall2011 -- DeHon

Concerns? Large number of wires May want to store for many cycles Could determine area May want to store for many cycles Not able to update all on every cycle Not able to use all on every cycle Penn ESE370 Fall2011 -- DeHon

Limited Data Use What else do we need to share the wires if can only use one register on each cycle? Use with shared data path Need to select the one output Can only update one Need to control which one gets written Penn ESE370 Fall2011 -- DeHon

Limited Data Use Add load enable to register Logic to enable one register on write Mux to select output Penn ESE370 Fall2011 -- DeHon

Good Solution? Could get away with just latch Not full register with master/slave latch Pay large amount for decode and mux Proportional to memory bits Penn ESE370 Fall2011 -- DeHon

Memory Idea Maximize storage density (bits/cm2) By minimizing the size/complexity of the repeated element Use shared periphery circuits to provide full functionality Trades off bandwidth (concurrent access) to save area Penn ESE370 Fall2011 -- DeHon

Memory Bank Penn ESE370 Fall2011 -- DeHon

Share Address Decode Word – group of bits read/written together All have same control Penn ESE370 Fall2011 -- DeHon

Share Address Decode Words Mux select bits (words) from row read When only want a subset Penn ESE370 Fall2011 -- DeHon

Share Address Decode Result: only spend N0.5 area (perimeter) on selecting rather than linear in bits Penn ESE370 Fall2011 -- DeHon

Gate Density When is 14n > 6n+32*sqrt(n) ? Penn ESE370 Fall2011 -- DeHon

Memory Row Use shared enable for wire economy Word line Penn ESE370 Fall2011 -- DeHon

Memory Column Use shared bus for area and wire economy Row enable selects the cells to read/write from bus Penn ESE370 Fall2011 -- DeHon

Memory Cell Hold data Conditionally drive onto output bus Conditionally overwritten with data from bus Penn ESE370 Fall2011 -- DeHon

SRAM Memory bit Penn ESE370 Fall2011 -- DeHon

SRAM Memory bit Core is back-to-back inverters for storage Like static latch Penn ESE370 Fall2011 -- DeHon

SRAM Memory bit Core is back-to-back inverters for storage Like static latch To minimize size, doesn’t include disable Penn ESE370 Fall2011 -- DeHon

SRAM Memory bit Pass gate mux for output to column Bit-Line (BL) Penn ESE370 Fall2011 -- DeHon

SRAM Memory bit How do we write into this cell? No directionality to pass gate If drive BL strong enough, can flip value in selected cell Ratioed operation Penn ESE370 Fall2011 -- DeHon

Column Capacitance What is capacitance of bit line (column)? Waccess (M5,M6) – transistor width of column device d rows g=Cdiff/Cgate Penn ESE370 Fall2011 -- DeHon

Time Driving Bit Line In terms of Waccess, Wbuf (M1,M3), d For Waccess=Wbuf=1, d=512, g=0.5 Penn ESE370 Fall2011 -- DeHon

Column Capacitance Consequence Want Waccess, Wbuf small to keep memory cell small Increasing Waccess, also increases Cbl Don’t really win by sizing up Driving bit line will be slow Penn ESE370 Fall2011 -- DeHon

Column Sensing Speedup read time by sensing limited swing Sense circuit detects small change in bit line voltage(s) Precharge to intermediate voltage BL and /BL swing opposite directions Amplifies for output Penn ESE370 Fall2011 -- DeHon

Output Amps Bottom of array includes Sense Amplifiers from bit lines to output Penn ESE370 Fall2011 -- DeHon

Column Write Writes driven from outside array Use large driver Strong enough to flip memory bit Strong so can charge column quickly Disable when not write Be careful on your project2 Could overwrite wrong row Penn ESE370 Fall2011 -- DeHon

Complete Memory Bank Penn ESE370 Fall2011 -- DeHon

Admin Project 2 out Andre away on Tuesday Due November 23 Note recommend milestones Andre away on Tuesday No office hour Tuesday Will be back for Wednesday lecture Penn ESE370 Fall2011 -- DeHon

Idea Memory for compact state storage Share circuitry across many bits Minimize area per bit  maximize density Aggressively use: Pass transistors, Ratioing Precharge, Amplifiers to keep area down Penn ESE370 Fall2011 -- DeHon