Lecture 6.1

Chapter 4

Shared Memory Architecture

Advanced Computer Architecture and Parallel Processing Hesham El-Rewini & Mostafa Abd-El-Barr

4.1 Classification of Shared
Memory Systems
• Shared memory systems are multi-port
and categorized as follows:

– Uniform memory Access (UMA)
– Non-uniform Memory Access (NUMA)
– Cache Only Memory Architecture (COMA)
M

P1 P2


Memory Systems
• Uniform Memory Access (UMA)
– Shared memory is accessible by all processors through an
interconnection network in the same way a single processor
accesses its memory.
– All processors have equal access time to any memory location.
– Because access to the shared memory is balanced, these
systems are called SMP (symmetric multiprocessor)

M M M M

Bus

C C C C

P P P P


Memory Systems
• Non Uniform Memory Access (NUMA)
– Each processor has part of the shared memory attached.
– The access time to modules depend on the distance to the
processor, which results a non-uniform memory access time.

Interconnection Network

P P P P

M M M M


Memory Systems
• Cache-Only Memory Architecture (COMA)
– Like NUMA, each processor has part of the shared memory in
COMA.
– COMA requires the data to be migrated to the processor
requesting it.


D D D D

C C C C

P P P P


4.2 Bus-Based Symmetric
Multiprocessors
• Shared memory systems can be designed using
bus-based or switch-based interconnection
networks.

Global Memory P C

P C

C C C P C

P P P
P C

M M M M


4.3 Basic Cache Coherency
Methods
• Cache-memory coherence using two
policies:
– Write-Through:
• The memory is updated every time the cache is
updated.
– Write-Back:
• The memory is updated only when the block in the
cache is being replaced.


Methods
• Cache-memory coherence using two
policies:

x Memory x’ Memory x Memory

x Cache x’ Cache x’ Cache

P P P

Before Write through Write back

Methods
• Cache-cache coherence using two
policies:
– Write-Invalidate:
• Maintains consistency by reading from local
caches until a write occurs.
– Write Update:
• Maintains consistency by immediately updating all
copies in all caches.


Methods
• Write-Invalidate:
x x’ x

x x x’ I x’ I

P1 P2 P3 P1 P2 P3 P1 P2 P3

Before Write Through Write back


Methods
• Write-Update:
x x’ x

x x x’ x’ x’ x’

P1 P2 P3 P1 P2 P3 P1 P2 P3

Before Write Through Write back


Methods
• Shared Memory System Coherence :

– Four combinations to maintain coherence
among all caches and global memory:

• Write-Update and Write-Through.
• Write-Update and Write-Back.
• Write-Invalidate and Write-Through.
• Write-Invalidate and Write-Back.


4.4 Snooping Protocols
• Based on watching bus activities and carry out the
appropriate coherency commands when necessary.

• Global memory is moved in blocks, and each block has a
state associated with it, which determines what happens
to the entire contents of the block.

• The state of a block might change as a result of the
operations Read-Miss, Read-Hit, Write-Miss, and Write-
Hit.


• Write-Invalidate and Write-Through

– Multiple processors can read block copies from main
memory safely until one processor updates its copy.

– At this time, all cache copies are invalidated and the
memory is updated to remain consistent.



State Description

Valid The copy is consistent with global memory
[VALID]

Invalid The copy is inconsistent
[INV]


Event Actions

Read Hit Use the local copy from the cache.

Read Miss Fetch a copy from global memory. Set the state of this
copy to Valid.
Write Hit Perform the write locally. Broadcast an Invalid
command to all caches. Update the global memory.

Write Get a copy from global memory. Broadcast an invalid
Miss command to all caches. Update the global memory.
Update the local copy and set its state to Valid.
Replace Since memory is always consistent, no write back is
needed when a block is replaced.


• Write-Invalidate and Write-Back (Ownership
Protocol)

– A valid block can be owned by memory and shared in multiple
caches that can contain only the shared copies of the block.

– Multiple processors can safely read these blocks from their
caches until one processor updates its copy.

– At this time, the writer becomes the only owner of the valid block
and all other copies are invalidated.


• Write-Invalidate and Write-Back
(Ownership Protocol)
State Description

Shared Data is valid and can be read safely. Multiple copies
(Read-Only) can be in this state
[RO]
Exclusive Only one valid cache copy exists and can be
(Read- read from and written to safely. Copies in other
Write) [RW] caches are invalid

Invalid The copy is inconsistent
[INV]


Event Action


Read Miss: If no Exclusive (Read-Write) copy exists, then
supply a copy from global memory. Set the state of
this copy to Shared (Read-Only). If an Exclusive
(Read-Write) copy exists, make a copy from the
cache that set the state to Exclusive (Read-Write),
update global memory and local cache with the copy.
Set the state to Shared (Read-Only) in both caches.

Write Hit If the copy is Exclusive (Read-Write), perform the write
locally. If the state is Shared (Read-Only), then
broadcast an Invalid to all caches. Set the state to
Exclusive (Read-Write).

Write Miss Get a copy from either a cache with an Exclusive (Read-
Write) copy, or from global memory itself. Broadcast an
Invalid command to all caches. Update the local copy
and set its state to Exclusive (Read-Write).

Block If a copy is in an Exclusive (Read-Write) state, it has to
Replacement be written back to main memory if the block is being
replaced. If the copy is in Invalid or Shared (Read-Only)
states, no write back is needed when a block is replaced.


• Write-Once

– This write-invalidate protocol, which was proposed by
Goodman in 1983 uses a combination of write-through
and write-back.

– Write-through is used the very first time a block is
written. Subsequent writes are performed using write
back.


• Write-Once
State Description

Invalid The copy is inconsistent.
[INV]

Valid The copy is consistent with global memory.
[VALID]

Reserved Data has been written exactly once and the copy is consistent with
[RES] global memory. There is only one copy of the global memory block
in one local cache.
Dirty Data has been updated more than once and there is only one copy
[DIRTY] in one local cache. When a copy is dirty, it must be written back to
global memory

• Write-Once

Event Actions


Read If no Dirty copy exists, then supply a copy from global
Miss memory. Set the state of this copy to Valid. If a dirty copy
exists, make a copy from the cache that set the state to
Dirty, update global memory and local cache with the
copy. Set the state to VALID in both caches.


• Write-Once
Write Hit If the copy is Dirty or Reserved, perform the write locally,
and set the state to Dirty. If the state is Valid, then broadcast
an Invalid command to all caches. Update the global memory
and set the state to Reserved.

Write Get a copy from either a cache with a Dirty copy or from
Miss global memory itself. Broadcast an Invalid command to all
caches. Update the local copy and set its state to Dirty.

Block If a copy is in a Dirty state, it has to be written back to main
Replace memory if the block is being replaced. If the copy is in Valid,
ment Reserved, or Invalid states, no write back is needed when a
block is replaced.


• Write-Update and Partial Write-Through

– In this protocol an update to one cache is written to memory at
the same time it is broadcast to other caches sharing the
updated block.

– These caches snoop on the bus and perform updates to their
local copies.

– There is also a special bus line, which is asserted to indicate that
at least one other cache is sharing the block.


State Description

Valid This is the only cache copy and is consistent
Exclusive with global memory
[VAL-X]

Shared There are multiple caches copies shared. All copies
[SHARE] are consistent with memory

Dirty This copy is not shared by other caches and
[DIRTY] has been updated. It is not consistent with
global memory. (Copy ownership)

Event Action

Read Hit Use the local copy from the cache. State does not change

Read Miss: If no other cache copy exists, then supply a copy from
global memory. Set the state of this copy to Valid
Exclusive. If a cache copy exists, make a copy from the
cache. Set the state to Shared in both caches. If the cache
copy was in a Dirty state, the value must also be written
to memory.


Write Hit Perform the write locally and set the state to Dirty. If the state
is Shared, then broadcast data to memory and to all caches and
set the state to Shared. If other caches no longer share the
block, the state changes from Shared to Valid Exclusion.

Write Miss The block copy comes from either another cache or from global
memory. If the block comes from another cache, perform the
update and update all other caches that share the block and
global memory. Set the state to Shared. If the copy comes from
memory, perform the write and set the state to Dirty.
Block If a copy is in a Dirty state, it has to be written back to main
Replacement memory if the block is being replaced. If the copy is in Valid
Exclusive or Shared states, no write back is needed when a
block is replaced.


• Write-Update and Write-Back

– This protocol is similar to the previous one
except that instead of writing through to the
memory whenever a shared block is updated,
memory updates are done only when the
block is being replaced.


State Description

Valid This is the only cache copy and is consistent with
Exclusive global memory
[VAL-X]
Shared There are multiple caches copies shared.
Clean
[SH-CLN]
Shared There are multiple shared caches copies. This is the last
Dirty one being updated. (Ownership)
[SH-DRT]
Dirty This copy is not shared by other caches and has
[DIRTY] been updated. It is not consistent with global
memory. (Ownership)

Event Action

Read Hit Use the local copy from the cache. State does not change

Read Miss: If no other cache copy exists, then supply a copy from global
memory. Set the state of this copy to Valid Exclusive. If a cache
copy exists, make a copy from the cache. Set the state to Shared
Clean. If the supplying cache copy was in a Valid Exclusion or
Shared Clean, its new state becomes Shared Clean. If the
supplying cache copy was in a Dirty or Shared Dirty state, its
new state becomes Shared Dirty.


Write Hit If the sate was Valid Exclusive or Dirty, Perform the write locally
and set the state to Dirty. If the state is Shared Clean or Shared
Dirty, perform update and change state to Shared Dirty.
Broadcast the updated block to all other caches. These caches
snoop the bus and update their copies and set their state to
Shared Clean.
Write Miss The block copy comes from either another cache or from global
memory. If the block comes from another cache, perform the
update, set the state to Shared Dirty, and broadcast the updated
block to all other caches. Other caches snoop the bus, update
their copies, and change their state to Shared Clean. If the copy
comes from memory, perform the write and set the state to Dirty.
Block If a copy is in a Dirty or Shared Dirty state, it has to be written
Replacement back to main memory if the block is being replaced. If the copy is
in Valid Exclusive, no write back is needed when a block is
replaced.

4.5 Directory Based Protocols
– Due to the nature of some interconnection networks
and the size of the shared memory system, updating
or invalidating caches using snoopy protocols might
become unpractical.

– Cache coherence protocols that somehow store
information on where copies of blocks reside are
called directory schemes.


– A directory is a data structure that maintains
information on the processors that share a memory
block and on its state.
– The information maintained in the directory could be
either centralized or distributed.

– A Central directory maintains information about all
blocks in a central data structure.

– The same information can be handled in a distributed
fashion by allowing each memory module to maintain
a separate directory.


• Protocol Categorization:

– Full Map Directories.

– Limited Directories.

– Chained Directories.


– Full-map Directories:

• Each directory entry contains N pointers, where N
is the number of processors.

• There could be N cached copies of a particular
block shared by all processors.

• For every memory block, an N bit vector is
maintained, where N equals the number of
processors in the shared memory system. Each bit
in the vector corresponds to one processor.


– Full-map Directories:

Memory Directory

X: 1 0 1 0 Data


X: Data X: Data

Cache C0 Cache C1 Cache C2 Cache C3


– Limited Directories:

• Fixed number of pointers per directory entry
regardless of the number of processors.

• Restricting the number of simultaneously cached
copies of any block should solve the directory size
problem that might exist in full-map directories.


– Limited Directories:

Memory Directory
X: C0 C2 Data


X: Data X: Data



– Chained Directories:

• Chained directories emulate full-map by
distributing the directory among the caches.

• Solving the directory size problem without
restricting the number of shared block copies.

• Chained directories keep track of shared copies of
a particular block by maintaining a chain of
directory pointers.


– Chained Directories:

Memory
Directory
X: C2 Data


X: CT Data X: C0 Data



• Invalidate Protocols:

– Centralized directory invalidate.
– Scalable Coherent Interface (SCI).
– Stanford Distributed Directory (SDD).


• Centralized Directory Invalidate

– Invalidating signals and a pointer to the requesting
processor are forwarded to all processors that have a
copy of the block.

– Each invalidated cache sends an acknowledgment to
the requesting processor.

– After the invalidation is complete, only the writing
processor will have a cache with a copy of the block.


• Centralized Directory Invalidate – Write by
P3
Memory Directory
Write-reply
X: 1 0 1 0 Data
Invalidate &
requester
Invalidate & Write
requester inv-ack

X: Data X: Data


inv-ack


• Scalable Coherent Interface (SCI)

– Doubly linked list of distributed directories.

– Each cached block is entered into a list of processors
sharing that block.

– For every block address, the memory and cache
entries have additional tag bits. Part of the memory
tag identifies the first processor in the sharing list (the
head). Part of each cache tag identifies the previous
and following sharing list entries.



– Initially memory is in the uncached state and cached
copies are invalid.

– A read request is directed from a processor to the
memory controller. The requested data is returned to
the requester’s cache and its entry state is changed
from invalid to the head state.

– This changes the memory state from uncached to
cached.



– When a new requester directs its read request to
memory, the memory returns a pointer to the head.

– A cache-to-cache read request (called Prepend) is
sent from the requester to the head cache.

– On receiving the request, the head cache sets its
backward pointer to point to the requester’s cache.

– The requested data is returned to the requester’s
cache and its entry state is changed to the head state.


• Scalable Coherent Interface (SCI)-Sharing
list addition

Memory Before After
Memory

1) read

Cache C0 Cache C2
Cache C0 Cache C2
(head) (Invalid) (middle) (head)

2) prepend



– The head of the list has the authority to purge other
entries in the list to obtain an exclusive (read-write)
entry.


• Scalable Coherent Interface (SCI)-Head
purging other entries.

Memory

Cache C0 Cache C2 Cache C3
(tail) (middle) (head)
Purge
Purge


• Scalable Distributed Directory (SDD)

– A singly linked list of distributed directories.

– Similar to the SCI protocol, memory points to the
head of the sharing list.

– Each processor points only to its predecessor.

– The sharing list additions and removals are handled
different from the SCI protocol .



– On a read miss, a new requester sends a read-miss
message to memory.
– The memory updates its head pointers to point to the
requester and send a read-miss-forward signal to the
old head.
– On receiving the request, the old head returns the
requested data along with its address as a read-miss-
reply.
– When the reply is received, at the requester’s cache,
the data is copied and the pointer is made to point to
the old head.


• Scalable Distributed Directory (SDD)-List
addition
Before After
Memory Memory

1) read
2) read- miss-
forward

(head) Cache C2
(Invalid) (middle) (head)

3) read-miss-reply



– On a write miss, a requester sends a write-
miss message to memory.
– The memory updates its head pointers to
point to the requester and sends a write-miss-
forward signal to the old head.
– The old head invalidates itself, returns the
requested data as a write-miss-reply-data
signal, and send a write-miss-forward to the
next cache in the list.


– When the next cache receives the write-miss-forward
signal, it invalidates itself and sends a write-miss-
forward to the next cache in the list.

– When the write-miss-forward signal is received by the
tail or by a cache that no longer has copy of the
block, a write-miss-reply is sent to the requester.

– The write is complete when the requester receives
both write-miss-reply-data and write-miss-reply.


• Scalable Distributed Directory (SDD)- Write
Miss List Removal

Before After
Memory
Memory
1) write

2) write miss-forward

3) write miss-reply-data
(tail) (head) (invalid)
Cache C3
(exclusive)
3) write miss-forward
4) write miss-reply


4.6 Shared Memory Programming
• Task Creation
Supervisor fork

… Workers

Supervisor join

… Workers

Supervisor
… Workers


• Communication – Serial versus Parallel
Code
Code
Private Data
Data

Shared Data

Stack Private Stack

• Communication – Via shared data

Code Access Access Code

Private Data Private Data

Shared Data

Private Stack Private Stack

Process 1 Process 2


• Synchronization - Locks
P1 P2 P3
Lock
…..
wait
…..
unlock wait
Lock
…..
…..
unlock
Lock
…..
…..
Locks unlock

• Synchronization - Barriers

T0 T1 T2

Barrier

Barrier
wait
wait Barrier

proceed proceed proceed
Synchronization
Point


4.7 Summary
• A shared memory system is made of multiple processors
and memory modules that are connected via some
interconnection network.
• Shared memory multiprocessors are usually bus-based
or switch-based.
• Communication among processors is achieved by writing
to and reading from memory.
• Synchronization among processors is achieved using
locks and barriers.
• Performance of a shared memory system becomes an
issue when the interconnection network connecting the
processors to global memory becomes a bottleneck.


4.7 Summary
• Local caches are used to alleviate the bottleneck
problem.
• But the introduction of caches created a consistency
problem among caches and between memory and
caches.
• Cache coherence schemes can be categorized into 2
main categories:
– Snooping protocols.
– Directory-based protocols.
• The programming model in shared memory systems has
proven to be easier to program compared to message
passing systems.


Lecture 6.1

More Related Content

What's hot

Similar to Lecture 6.1

More from Mr SMAK

Recently uploaded

Lecture 6.1