Computer Organization and Design The Hardware/Software Interface		
		Chapter 5	
		Large and Fast: Exploiting Memory Hierarchy	
Memory Technology
Static RAM (SRAM)
- 0.5ns - 2.5ns, $2000 - $5000 per GB
Dynamic RAM (DRAM)
- 50ns - 70ns, $20 - $75 per GB
Magnetic disk
- 5ms - 20ms, $0.20 - $2 per GB
Ideal memory
■ Access time of SRAM
■ Capacity and cost/GB of disk
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy —
Principle of Locality
Programs access a small proportion of their address space at any time
Temporal locality
■ Items accessed recently are likely to be accessed again soon
■ e.g., instructions in a loop, induction variables
Spatial locality
■ Items near those accessed recently are likely to be accessed soon
■ E.g., sequential instruction access, array data
®
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 3
Taking Advantage of Locality
Memory hierarchy
Store everything on disk
Copy recently accessed (and nearby) items from disk to smaller DRAM memory
■ Main memory
Copy more recently accessed (and nearby) items from DRAM to smaller SRAM memory
■ Cache memory attached to CPU
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 4
Memory Hierarchy Levels
Processor
Data is transferred
Block (aka line): unit of copying
■ May be multiple words
If accessed data is present in upper level
■ Hit: access satisfied by upper level
Hit ratio: hits/accesses
If accessed data is absent
■ Miss: block copied from lower level
Time taken: miss penalty
Miss ratio: misses/accesses = 1 - hit ratio
■ Then accessed data supplied from upper level
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 5
Cache Memory
■ Cache memory
■ The level of the memory hierarchy closest to the CPU
■ Given accesses X1t    Xn_-,, Xn
X,
"n-2
X
n-1
X5
X,
X,
V1-2
X
n- 1
X,
How do we know if the data is present?
Where do we look?
a. Before the reference to X„    b. After the reference to Xn
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 6
Direct Mapped Cache
Location determined by address Direct mapped: only one choice
■ (Block address) modulo (#Blocks in cache)
Cache
0- i-Oi-O-i-O-^
OOt-t-OOt-i-
00001- t-t-1-
#Blocks is a power of 2
Use low-order address bits
00001     00101     01001     01101     10001     10101     11001 11101
Memory
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 7
Tags and Valid Bits
How do we know which particular block is stored in a cache location?
■ Store block address as well as the data
■ Actually, only need the high-order bits
■ Called the tag
What if there is no data in a location?
■ Valid bit: 1 = present, 0 = not present
■ Initially 0
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 8
Cache Example
8-blocks, 1 word/block, direct mapped Initial state
Index	V	Tag	Data
000	N		
001	N		
010	N		
011	N		
100	N		
101	N		
110	N		
111	N		
<
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 9
Cache Example
Word addr	Binary addr	Hit/miss	Cache block
22	10 110	Miss	110
Index	V	Tag	Data
000	N		
001	N		
010	N		
011	N		
100	N		
101	N		
			
111	N		
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 10
Cache Example
Word addr	Binary addr	Hit/miss	Cache block
26	11 010	Miss	010
Index	V	Tag	Data
000	N		
001	N		
			
011	N		
100	N		
101	N		
110	Y	10	Mem[10110]
111	N		
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 11
Cache Example
Word addr	Binary addr	Hit/miss	Cache block
22	10 110	Hit	110
26	11 010	Hit	010
Index	V	Tag	Data
000	N		
001	N		
010	Y	11	Mem[11010]
011	N		
100	N		
101	N		
110	Y	10	Mem[10110]
111	N		
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 12
Cache Example
Word addr	Binary addr	Hit/miss	Cache block
16	10 000	Miss	000
3	00 011	Miss	011
16	10 000	Hit	000
Index	V	Tag	Data
			
001	N		
010	Y	11	Mem[11010]
		00	
100	N		
101	N		
110	Y	10	Mem[10110]
111	N		
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 13
Cache Example
Word addr	Binary addr	Hit/miss	Cache block
18	10010	Miss	010
Index	V	Tag	Data
000	Y	10	Mem[10000]
001	N		
			
011	Y	00	Mem[00011]
100	N		
101	N		
110	Y	10	Mem[10110]
111	N		
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 14
Address Subdivision
<
Hit
Address (showing bit positions)
31 30 •	■ 13 12 11 ■		■2	1 0
				Byte offset
Tag
20
Index
Index
0
1
2
1021
1022
1023
20
9
10
Valid Tag Data
32
Data
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 15
Example: Larger Block Size
64 blocks, 16 bytes/block
■ To what block number does address 1200 map?
Block address = Li 200/16j = 75 Block number = 75 modulo 64 = 11
31
10 9
4 3
0
Tag
Index Offset
22 bits
6 bits
4 bits
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 16
Block Size Considerations
Larger blocks should reduce miss rate
■ Due to spatial locality
But in a fixed-sized cache
■ Larger blocks =^> fewer of them
More competition => increased miss rate
■ Larger blocks => pollution
Larger miss penalty
■ Can override benefit of reduced miss rate
■ Early restart and critical-word-first can help
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 17
Cache Misses
On cache hit, CPU proceeds normally On cache miss
- Stall the CPU pipeline
■ Fetch block from next level of hierarchy
■ Instruction cache miss
Restart instruction fetch
■ Data cache miss
Complete data access
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 18
Write-Through
On data-write hit, could just update the block in cache
■ But then cache and memory would be inconsistent
Write through: also update memory But makes writes take longer
■ e.g., if base CPI = 1, 10% of instructions are stores, write to memory takes 100 cycles
Effective CPI = 1 +0.1x100 = 11
Solution: write buffer
■ Holds data waiting to be written to memory
■ CPU continues immediately
Only stalls on write if write buffer is already full
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 19
Write-Back
Alternative: On data-write hit, just update the block in cache
■ Keep track of whether each block is dirty
When a dirty block is replaced
■ Write it back to memory
■ Can use a write buffer to allow replacing block to be read first
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 20
Write Allocation
What should happen on a write miss? Alternatives for write-through
■ Allocate on miss: fetch the block
■ Write around: don't fetch the block
Since programs often write a whole block before reading it (e.g., initialization)
For write-back
■ Usually fetch the block
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 21
Example: Intrinsity FastMATH
Embedded MIPS processor
■ 12-stage pipeline
■ Instruction and data access on each cycle
Split cache: separate l-cache and D-cache
- Each 16KB: 256 blocks * 16 words/block
■ D-cache: write-through or write-back
SPEC2000 miss rates
■ l-cache: 0.4%
■ D-cache: 11.4%
■ Weighted average: 3.2%
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 22
Example: Intrinsity FastMATH
Hit
Tag
18 bits V Tag
18
4
Address (showing bit positions)
31    •••    14 13—65 — 21 0
18
Index
8
4 Byte offset
512 bits
Data
32
32
Mux
32
Data
Block offset
256 entries
32
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 23
Main Memory Supporting Caches
Use DRAMs for main memory
■ Fixed width (e.g., 1 word)
■ Connected by fixed-width clocked bus
Bus clock is typically slower than CPU clock
Example cache block read
■ 1 bus cycle for address transfer
■ 15 bus cycles per DRAM access
■ 1 bus cycle per data transfer
For 4-word block, 1-word-wide DRAM
■ Miss penalty = 1 + 4x15 + 4x1 = 65 bus cycles
■ Bandwidth = 16 bytes / 65 cycles = 0.25 B/cycle
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 24
Increasing Memory Bandwidth
a. One-word-wide memory organization
Processor
Processor
^^lOiultiplexor'
Cache
Bus
Memory
Memory	Memory	Memory	Memory
bankO	bank 1	bank 2	bank 3
b. Wider memory organization
c. Interleaved memory organization
4-word wide memory
■ Miss penalty = 1 +15 + 1 =17 bus cycles
■ Bandwidth = 16 bytes /17 cycles = 0.94 B/cycle
4-bank interleaved memory
■ Miss penalty = 1 + 15 + 4x1 =20 bus cycles
. Bandwidth = 16 bytes / 20 cycles = 0.8 B/cycle
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 25
Advanced DRAM Organization
Bits in a DRAM are organized as a rectangular array
■ DRAM accesses an entire row
■ Burst mode: supply successive words from a row with reduced latency
Double data rate (DDR) DRAM
■ Transfer on rising and falling clock edges
Quad data rate (QDR) DRAM
■ Separate DDR inputs and outputs
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 26
DRAM Generations
Year	Capacity	$/GB
1980	64Kbit	$1500000
1983	256Kbit	$500000
1985	1Mbit	$200000
1989	4Mbit	$50000
1992	16Mbit	$15000
1996	64Mbit	$10000
1998	128Mbit	$4000
2000	256Mbit	$1000
2004	512Mbit	$250
2007	1Gbit	$50
300 250 +-200 150 100 50 0
i      i      i      i      i     i      i      i r
'80 '83 '85 '89 '92 '96 '98 '00 '04 '07
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 27
Measuring Cache Performance
Components of CPU time
■ Program execution cycles
Includes cache hit time
■ Memory stall cycles
Mainly from cache misses
With simplifying assumptions:
Memory stal I cycles
Memory accesses Program
x Miss rate x Miss penalty
Instructions    Misses    ... u =-x-x Miss penalty
Program Instruction
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy
Cache Performance Example
Given
■ l-cache miss rate = 2%
■ D-cache miss rate = 4%
■ Miss penalty = 100 cycles
- Base CPI (ideal cache) = 2
■ Load & stores are 36% of instructions
Miss cycles per instruction
- l-cache: 0.02 x 100 = 2
- D-cache: 0.36 x 0.04 x 100 = 1.44 Actual CPI = 2 + 2 + 1.44 = 5.44
- Ideal CPU is 5.44/2 =2.72 times faster
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 29
Average Access Time
Hit time is also important for performance Average memory access time (AMAT)
■ AMAT = Hit time + Miss rate x Miss penalty
Example
■ CPU with 1 ns clock, hit time = 1 cycle, miss penalty = 20 cycles, l-cache miss rate = 5%
. AMAT = 1 + 0.05 x 20 = 2ns 2 cycles per instruction
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 30
Performance Summary
When CPU performance increased
■ Miss penalty becomes more significant
Decreasing base CPI
■ Greater proportion of time spent on memory stalls
Increasing clock rate
■ Memory stalls account for more CPU cycles
Can't neglect cache behavior when evaluating system performance
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 31
Associative Caches
Fully associative
■ Allow a given block to go in any cache entry
■ Requires all entries to be searched at once
■ Comparator per entry (expensive)
n-way set associative
■ Each set contains n entries
■ Block number determines which set
(Block number) modulo (#Sets in cache)
■ Search all entries in a given set at once
■ n comparators (less expensive)
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 32
Associative Cache Example
Direct mapped
Block # 01234567
Data
Tag Search
Set associative
Set#   0    1    2 3
Fully associative
Data
Data
Tag Search
Tag Search
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 33
Spectrum of Associativity
For a cache with 8 entries
One-way set associative (direct mapped)
Block   Tag Data
0
1
2
3
4
5
6
7
Two-way set associative
Set    Tag  Data Tag Data
0
1
2
3
Four-way set associative
Set    Tag Data Tag Data Tag Data Tag Data
0 |       |       |       |       |       |       | |
1
Eight-way set associative (fully associative)
Tag Data Tag Data Tag Data Tag Data Tag Data Tag Data Tag Data Tag Data
M 14
Chapter 5 —
Large and Fast: Exploiting Memory Hierarchy
Associativity Example
Compare 4-block caches
■ Direct mapped, 2-way set associative, fully associative
■ Block access sequence: 0, 8, 0, 6, 8
Direct mapped
Block address	Cache index	Hit/miss	Cache content after access			
			0	1	2	3
0	0	miss				
8	0	miss				
0	0	miss				
6	2	miss	Mem[0]			
8	0	miss			Mem[6]	
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 35
Associativity Example
2-way set associative
Block	Cache	Hit/miss	Cache content after access			
address	index		SetO		Set 1	
0	0	miss				
8	0	miss	Mem[0]			
0	0	hit	Mem[0]	Mem[8]		
6	0	miss	Mem[0]			
8	0	miss		Mem[6]		
Fully associative
Block address		Hit/miss	Cache content after access			
0		miss				
8		miss	Mem[0]			
0		hit	Mem[0]	Mem[8]		
6		miss	Mem[0]	Mem[8]		
8		hit	Mem[0]	Mem[8]	Mem[6]	
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 36
How Much Associativity
Increased associativity decreases miss rate
■ But with diminishing returns
Simulation of a system with 64KB D-cache, 16-word blocks, SPEC2000
- 1-way: 10.3%
- 2-way: 8.6%
- 4-way: 8.3%
- 8-way: 8.1%
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 37
Set Associative Cache Organization
Address
31 30-- 12 11 1098---32 1 0
Tag
22
Index
Index   V Tag Data 0
1
2
253
254
255
<
V Tag    Data       V Tag    Data       V Tag Data
Ö
Hit
22
32
^-to-1 multiplexor)
Data
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 38
Replacement Policy
Direct mapped: no choice Set associative
■ Prefer non-valid entry, if there is one
■ Otherwise, choose among entries in the set
Least-recently used (LRU)
■ Choose the one unused for the longest time
Simple for 2-way, manageable for 4-way, too hard beyond that
Random
■ Gives approximately the same performance as LRU for high associativity
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 39
Multilevel Caches
Primary cache attached to CPU
■ Small, but fast
Level-2 cache services misses from primary cache
■ Larger, slower, but still faster than main memory
Main memory services L-2 cache misses Some high-end systems include L-3 cache
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 40
Multilevel Cache Example
Given
- CPU base CPI = 1, clock rate = 4GHz
■ Miss rate/instruction = 2%
■ Main memory access time = 100ns
With just primary cache
■ Miss penalty = 100ns/0.25ns = 400 cycles . Effective CPI = 1 + 0.02 x 400 = 9
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 41
Example (cont.)
Now add L-2 cache
■ Access time = 5ns
■ Global miss rate to main memory = 0.5%
Primary miss with L-2 hit
■ Penalty = 5ns/0.25ns = 20 cycles
Primary miss with L-2 miss
■ Extra penalty = 400 cycles
CPI = 1 + 0.02 x 20 + 0.005 x 400 = 3.4 Performance ratio = 9/3.4 = 2.6
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 42
Multilevel Cache Considerations
Primary cache
■ Focus on minimal hit time
L-2 cache
■ Focus on low miss rate to avoid main memory access
■ Hit time has less overall impact Results
■ L-1 cache usually smaller than a single cache
■ L-1 block size smaller than L-2 block size
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 43
Interactions with Advanced CPUs
Out-of-order CPUs can execute instructions during cache miss
■ Pending store stays in load/store unit
■ Dependent instructions wait in reservation stations
Independent instructions continue
Effect of miss depends on program data flow
■ Much harder to analyse
■ Use system simulation
®
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 44
Interactions with Software
Misses depend on memory access patterns
■ Algorithm behavior
■ Compiler optimization for memory access
<
Radix Sort
16     32     64    128   256   512   1024 2048 4096 Size (K items to sort)
1600
16     32     64    128   256   512   1024 2048 4096 Size (K items to sort)
8! 3-
£
0) 2
16     32     64    128   256   512   1024 2048 4096 Size (K items to sort)
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 45
Virtual Memory
Use main memory as a "cache" for secondary (disk) storage
■ Managed jointly by CPU hardware and the operating system (OS)
Programs share main memory
■ Each gets a private virtual address space holding its frequently used code and data
■ Protected from other programs
CPU and OS translate virtual addresses to physical addresses
■ VM "block" is called a page
■ VM translation "miss" is called a page fault
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 46
Address Translation
Fixed-size pages (e.g., 4K)
Virtual addresses
Virtual address
Physical addresses 31 30 29 28 27 ...................... 15 14 13 12 11 10 9 8 ........... 32 1 0
Virtual page number
Page offset
c
Translation
Disk addresses
29 28 27 .........:............ 15 141312 11 1098
32 10
Physical page number
Page offset
Physical address
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 47
Page Fault Penalty
On page fault, the page must be fetched from disk
■ Takes millions of clock cycles
■ Handled by OS code
Try to minimize page fault rate
Fully associative placement
■ Smart replacement algorithms
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 48
Page Tables
Stores placement information
■ Array of page table entries, indexed by virtual page number
■ Page table register in CPU points to page table in physical memory
If page is present in memory
■ PTE stores the physical page number
■ Plus other status bits (referenced, dirty, ...)
If page is not present
■ PTE can refer to location in swap space on disk
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 49
Translation Using a Page Table
Page table register
Virtual address
31  30 29 28 27........................15 14 1 3 12 1 1  10 9 8 ........3 2 1 0
Virtual page number
Valid
Page table
Page offset
20
If 0 then page is not present in memory
29 28 2 7.........
Physical page number
18
■1 5 1 4 1 3 1 2 1 1  1 0 9 8-
12
■3210
Physical page number
Page offset
<
Physical address
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 50
Mapping Pages to Storage
Virtual page number
Page table
Physical page or Valid   disk address
Physical memory
Disk storage
<
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 51
Replacement and Writes
To reduce page fault rate, prefer least-recently used (LRU) replacement
■ Reference bit (aka use bit) in PTE set to 1 on access to page
■ Periodically cleared to 0 by OS
■ A page with reference bit = 0 has not been used recently
Disk writes take millions of cycles
■ Block at once, not individual locations
■ Write through is impractical
■ Use write-back
■ Dirty bit in PTE set when page is written
®
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 52
Fast Translation Using a TLB
Address translation would appear to require extra memory references
■ One to access the PTE
■ Then the actual memory access
But access to page tables has good locality
■ So use a fast cache of PTEs within the CPU
■ Called a Translation Look-aside Buffer (TLB)
. Typical: 16-512 PTEs, 0.5-1 cycle for hit, 10-100 cycles for miss, 0.01 %-1% miss rate
■ Misses could be handled by hardware or software
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 53
Fast Translation Using a TLB
TLB
Virtual page                                    Physical page number Valid Dirty Ref_Tag_address
1  1 1 7
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 54
TLB Misses
If page is in memory
■ Load the PTE from memory and retry
■ Could be handled in hardware
Can get complex for more complicated page table structures
■ Or in software
Raise a special exception, with optimized handler
If page is not in memory (page fault)
■ OS handles fetching the page and updating the page table
■ Then restart the faulting instruction
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 55
TLB Miss Handler
TLB miss indicates
■ Page present, but PTE not in TLB
■ Page not preset
Must recognize TLB miss before destination register overwritten
■ Raise exception
Handler copies PTE from memory to TLB
■ Then restarts instruction
■ If page not present, page fault will occur
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 56
Page Fault Handler
Use faulting virtual address to find PTE Locate page on disk Choose page to replace
■ If dirty, write to disk first
Read page into memory and update page table
Make process runnable again
■ Restart from faulting instruction
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 57
TLB and Cache Interaction
Virtual address
TLB
		31 30 29		.............................14	13 12	11 10 9.......	■3210			
			Virtual page number			Page offset				
			*■	20				J 2		
ilid Dirty				Tag		Physical page number				
										
		© —								
		©—								
		© —								
		© —								
		©—1								
■»							20			
Physical page number
-Physical address-
Page offset
Physical address tag
Cache index 8
Block offset
Byte
offsel
Valid_Tag
Cache
Cache hit
12
Data
32
If cache tag uses physical address
■ Need to translate before cache lookup
Alternative: use virtual address tag
■ Complications due to aliasing
Different virtual addresses for shared physical address
Data
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 58
Memory Protection
Different tasks can share parts of their virtual address spaces
■ But need to protect against errant access
■ Requires OS assistance
Hardware support for OS protection
■ Privileged supervisor mode (aka kernel mode)
■ Privileged instructions
■ Page tables and other state information only accessible in supervisor mode
■ System call exception (e.g., syscall in MIPS)
®
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 59
The Memory Hierarchy
The BIG Picture
Common principles apply at all levels of the memory hierarchy
■ Based on notions of caching
At each level in the hierarchy
■ Block placement
■ Finding a block
■ Replacement on a miss
■ Write policy
®
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy —
Block Placement
Determined by associativity
■ Direct mapped (1-way associative)
One choice for placement
■ n-way set associative
n choices within a set
■ Fully associative
Any location
Higher associativity reduces miss rate
■ Increases complexity, cost, and access time
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 61
Finding a Block
Associativity	Location method	Tag comparisons
Direct mapped	Index	1
n-way set associative	Set index, then search entries within the set	n
Fully associative	Search all entries	#entries
	Full lookup table	0
Hardware caches
■ Reduce comparisons to reduce cost
Virtual memory
■ Full table lookup makes full associativity feasible
■ Benefit in reduced miss rate
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 62
Replacement
Choice of entry to replace on a miss
■ Least recently used (LRU)
Complex and costly hardware for high associativity
■ Random
Close to LRU, easier to implement
Virtual memory
■ LRU approximation with hardware support
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 63
Write Policy
Write-through
■ Update both upper and lower levels
■ Simplifies replacement, but may require write buffer
Write-back
■ Update upper level only
■ Update lower level when block is replaced
■ Need to keep more state
Virtual memory
■ Only write-back is feasible, given disk write latency
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 64
Sources of Misses
Compulsory misses (aka cold start misses)
■ First access to a block
Capacity misses
■ Due to finite cache size
■ A replaced block is later accessed again
Conflict misses (aka collision misses)
■ In a non-fully associative cache
■ Due to competition for entries in a set
■ Would not occur in a fully associative cache of the same total size
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 65
Cache Design Trade-offs
Design change	Effect on miss rate	Negative performance effect
Increase cache size	Decrease capacity misses	May increase access time
Increase associativity	Decrease conflict misses	May increase access time
Increase block size	Decrease compulsory misses	Increases miss penalty. For very large block size, may increase miss rate due to pollution.
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 66
Virtual Machines
Host computer emulates guest operating system and machine resources
Improved isolation of multiple guests
■ Avoids security and reliability problems
■ Aids sharing of resources
Virtualization has some performance impact
■ Feasible with modern high-performance comptuers
Examples
. IBM VM/370 (1970s technology!) . VMWare
■ Microsoft Virtual PC
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 67
Virtual Machine Monitor
Maps virtual resources to physical resources
■ Memory, I/O devices, CPUs
Guest code runs on native machine in user mode
■ Traps to VMM on privileged instructions and access to protected resources
Guest OS may be different from host OS VMM handles real I/O devices
■ Emulates generic virtual I/O devices for guest
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 68
Example: Timer Virtualization
In native machine, on timer interrupt
■ OS suspends current process, handles interrupt, selects and resumes next process
With Virtual Machine Monitor
■ VMM suspends current VM, handles interrupt, selects and resumes next VM
If a VM requires timer interrupts
■ VMM emulates a virtual timer
■ Emulates interrupt for VM when physical timer
interrupt occurs
®
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 69
Instruction Set Support
User and System modes
Privileged instructions only available in system mode
■ Trap to system if executed in user mode
All physical resources only accessible using privileged instructions
■ Including page tables, interrupt controls, I/O registers
Renaissance of virtualization support
■ Current ISAs (e.g., x86) adapting
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 70
Cache Control
Example cache characteristics
■ Direct-mapped, write-back, write allocate
■ Block size: 4 words (16 bytes)
. Cache size: 16 KB (1024 blocks)
■ 32-bit byte addresses
■ Valid bit and dirty bit per block
■ Blocking cache
CPU waits until access is complete 31 10 9 4  3 0
Tag Index Offset
<
18 bits 10 bits        4 bits
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy —
Interface Signals
CPU	Read/Write	Cache	Read/Write	Memory
	Valid		Valid	
	32 Address        ✓ „		32 Address        , ,	
	Write Data    3? „		Write Data    12? ,	
	(   Read Data 3?		t   Read Data 12?	
	t Readv		„ Ready	
				
Multiple cycles per access
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 72
Finite State Machines
Use an FSM to sequence control steps
Set of states, transition on each clock edge
■ State values are binary encoded
■ Current state stored in a register
■ Next state
= fn (current state,
current inputs)
Control output signals = f0 (current state)
14
		
	Combinational control logic	
	Outputs X	
	Inputs A	v.
t	•n	
Inputs from cache datapath
Datapath control outputs
State register
Next state
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 73
Cache Controller FSM
Could partition into separate states to reduce clock cycle time
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 74
Cache Coherence Problem
Suppose two CPU cores share a physical address space
■ Write-through caches
Time step	Event	CPU A's cache	CPU B's cache	Memory
0				0
1	CPU A reads X	0		0
2	CPU B reads X	0	0	0
3	CPU A writes 1 to X	1		1
M 14
Chapter 5 —
Large and Fast: Exploiting Memory Hierarchy —
Coherence Defined
Informally: Reads return most recently written value
Formally:
■ P writes X; P reads X (no intervening writes) => read returns written value
■ Pi writes X; P2 reads X (sufficiently later) => read returns written value
c.f. CPU B reading X after step 3 in example
■ P1 writes X, P2 writes X
=> all processors see writes in the same order
End up with the same final value for X
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 76
Cache Coherence Protocols
Operations performed by caches in multiprocessors to ensure coherence
■ Migration of data to local caches
Reduces bandwidth for shared memory
■ Replication of read-shared data
Reduces contention for access
Snooping protocols
■ Each cache monitors bus reads/writes
Directory-based protocols
■ Caches and memory record sharing status of blocks in a directory
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 77
Invalidating Snooping Protocols
Cache gets exclusive access to a block when it is to be written
■ Broadcasts an invalidate message on the bus
■ Subsequent read in another cache misses
Owning cache supplies updated value
CPU activity	Bus activity	CPU A's cache	CPU B's cache	Memory
				0
CPU A reads X	Cache miss for X	0		0
CPU B reads X	Cache miss for X	0	0	0
CPU A writes 1 to X	Invalidate for X	1		0
CPU B read X	Cache miss for X	1	1	1
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 78
Memory Consistency
When are writes seen by other processors
■ "Seen" means a read returns the written value
■ Can't be instantaneously
Assumptions
■ A write completes only when all processors have seen it
■ A processor does not reorder writes with other accesses
Consequence
■ P writes X then writes Y
=^> all processors that see new Y also see new X
■ Processors can reorder reads, but not writes
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 79
Multilevel On-Chip Caches
Intel Nehalem 4-core processor
Two channel (128 bit) rjiemory interface
Per core: 32KB L1 l-cache, 32KB L1 D-cache, 512KB L2 cache
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 80
2-Level TLB Organization
	Intel Nehalem	AMD Opteron X4
Virtual addr	48 bits	48 bits
Physical addr	44 bits	48 bits
Page size	4KB, 2/4MB	4KB, 2/4MB
L1 TLB (per core)	L1 l-TLB: 128 entries for small pages, 7 per thread (2x) for large pages L1 D-TLB: 64 entries for small pages, 32 for large pages Both 4-way, LRU replacement	L1 l-TLB: 48 entries L1 D-TLB: 48 entries Both fully associative, LRU replacement
L2 TLB (per core)	Single L2 TLB: 512 entries 4-way, LRU replacement	L2 l-TLB: 512 entries L2 D-TLB: 512 entries Both 4-way, round-robin LRU
TLB misses	Handled in hardware	Handled in hardware
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 81
3-Level Cache Organization
	Intel Nehalem	AMD Opteron X4
L1 caches (per core)	L1 l-cache: 32KB, 64-byte blocks, 4-way, approx LRU replacement, hit time n/a L1 D-cache: 32KB, 64-byte blocks, 8-way, approx LRU replacement, writeback/allocate, hit time n/a	L1 l-cache: 32KB, 64-byte blocks, 2-way, LRU replacement, hit time 3 cycles L1 D-cache: 32KB, 64-byte blocks, 2-way, LRU replacement, writeback/allocate, hit time 9 cycles
L2 unified cache (per core)	256KB, 64-byte blocks, 8-way, approx LRU replacement, writeback/allocate, hit time n/a	512KB, 64-byte blocks, 16-way, approx LRU replacement, writeback/allocate, hit time n/a
L3 unified cache (shared)	8MB, 64-byte blocks, 16-way, replacement n/a, writeback/allocate, hit time n/a	2MB, 64-byte blocks, 32-way, replace block shared by fewest cores, write-back/allocate, hit time 32 cycles
n/a: data not available
<
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 82
Mis Penalty Reduction
Return requested word first
■ Then back-fill rest of block
Non-blocking miss processing
■ Hit under miss: allow hits to proceed
■ Mis under miss: allow multiple outstanding misses
Hardware prefetch: instructions and data Opteron X4: bank interleaved L1 D-cache
■ Two concurrent accesses per cycle
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 83
Pitfalls
Byte vs. word addressing
■ Example: 32-byte direct-mapped cache, 4-byte blocks
Byte 36 maps to block 1 Word 36 maps to block 4
Ignoring memory system effects when writing or generating code
■ Example: iterating over rows vs. columns of arrays
■ Large strides result in poor locality
M 14
Chapter 5 —
Large and Fast: Exploiting Memory Hierarchy
Pitfalls
In multiprocessor with shared L2 or L3 cache
■ Less associativity than cores results in conflict misses
■ More cores => need to increase associativity
Using AMAT to evaluate performance of out-of-order processors
■ Ignores effect of non-blocked accesses
■ Instead, evaluate performance by simulation
M 14
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 85
Pitfalls
Extending address range using segments
- E.g., Intel 80286
■ But a segment is not always big enough
■ Makes address arithmetic complicated
Implementing a VMM on an ISA not designed for virtualization
■ E.g., non-privileged instructions accessing hardware resources
■ Either extend ISA, or require guest OS not to
use problematic instructions
®
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy — 86
Concluding Remarks
Fast memories are small, large memories are slow
■ We really want fast, large memories ©
■ Caching gives this illusion ©
Principle of locality
■ Programs use a small part of their memory space frequently
Memory hierarchy
■ L1 cache o L2 cache o ... o DRAM memory ^ disk
Memory system design is critical for multiprocessors
Chapter 5 — Large and Fast: Exploiting Memory Hierarchy
Exercises
Answer the following exercises, and send your answers as a PDF attachment to the email address listed below
xamiri@fi.muni.cz I     Leave body of the email blank Deadline is May 19th Exercises
5.3.2(b), 5.4.1(b), 5.4.2(b), 5.7.3(a), 5.8.1(b), 5.8.4(b), 5.8.5(b), 5.10.1(a), 5.11.1(a)
M 14
Chapter 1 — Computer Abstractions and Technology — 88