• Home

  • Documents

  • Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time...
42
Scalable NUMA-aware Blocking Synchronization Primitives Sanidhya Kashyap, Changwoo Min, Taesoo Kim

Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

  • Upload
    others

  • View
    4

  • Download
    0

Embed Size (px)

Citation preview

Page 1: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Scalable NUMA-awareBlocking Synchronization Primitives

Sanidhya Kashyap, Changwoo Min, Taesoo Kim

Page 2: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

'i

The rise of big NUMA machines

Page 3: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

'i

The rise of big NUMA machines

Page 4: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

'i

The rise of big NUMA machines

Page 5: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Importance of NUMA awarenessNUMA node 1 NUMA node 2

W1

W2

W3

W4

W5W6

L

File

W1 W2 W3 W4 W6 W5

NUMA oblivious

Page 6: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Importance of NUMA awarenessNUMA node 1 NUMA node 2

W1

W2

W3

W4

W5W6

L

File

W1 W2 W3 W4 W6 W5

NUMA oblivious

W1 W6 W2 W3 W4 W5

NUMA aware/hierarchical

Page 7: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Importance of NUMA awarenessNUMA node 1 NUMA node 2

W1

W2

W3

W4

W5W6

L

File

Idea:Make synchronization primitives NUMA aware!

W1 W2 W3 W4 W6 W5

NUMA oblivious

W1 W6 W2 W3 W4 W5

NUMA aware/hierarchical

Page 8: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Lock's research efforts and their useLinux kernel lock

adoption / modificationDekker's algorithm (1962)Semaphore (1965)

Lamport's bakery algorithm (1974)Backoff lock (1989)Ticket lock (1991)MCS lock (1991)

Hierarchical lock – HCLH (2006)Flat combining NUMA lock (2011)

Remote Core locking (2012)Cohort lock (2012)

RW cohort lock (2013)Malthusian lock (2014)

HMCS lock (2015)AHMCS lock(2016)

HBO lock (2003)

Lock's research efforts

Page 9: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Lock's research efforts and their useLinux kernel lock

adoption / modificationDekker's algorithm (1962)Semaphore (1965)

Lamport's bakery algorithm (1974)Backoff lock (1989)Ticket lock (1991)MCS lock (1991)

Hierarchical lock – HCLH (2006)Flat combining NUMA lock (2011)

Remote Core locking (2012)Cohort lock (2012)

RW cohort lock (2013)Malthusian lock (2014)

HMCS lock (2015)AHMCS lock(2016)

HBO lock (2003)

NUMA-awarelocks

Lock's research efforts

Page 10: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Lock's research efforts and their useLinux kernel lock

adoption / modificationDekker's algorithm (1962)Semaphore (1965)

Lamport's bakery algorithm (1974)Backoff lock (1989)Ticket lock (1991)MCS lock (1991)

Hierarchical lock – HCLH (2006)Flat combining NUMA lock (2011)

Remote Core locking (2012)Cohort lock (2012)

RW cohort lock (2013)Malthusian lock (2014)

HMCS lock (2015)AHMCS lock(2016)

HBO lock (2003)

NUMA-awarelocks

Spinlock TTAS→

Semaphore TTAS + block→

Rwsem TTAS + block→

Spinlock ticket→

Mutex TTAS + spin + block→

Rwsem TTAS + spin + block→

Spinlock ticket→

Mutex TTAS + block→

Rwsem TTAS + block→

Spinlock qspinlock→

Mutex TTAS + spin + block→

Rwsem TTAS + spin + block→

Lock's research efforts

1990s

2011

2014

2016

Page 11: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Lock's research efforts and their useLinux kernel lock

adoption / modificationDekker's algorithm (1962)Semaphore (1965)

Lamport's bakery algorithm (1974)Backoff lock (1989)Ticket lock (1991)MCS lock (1991)

Hierarchical lock – HCLH (2006)Flat combining NUMA lock (2011)

Remote Core locking (2012)Cohort lock (2012)

RW cohort lock (2013)Malthusian lock (2014)

HMCS lock (2015)AHMCS lock(2016)

HBO lock (2003)

NUMA-awarelocks

Spinlock TTAS→

Semaphore TTAS + block→

Rwsem TTAS + block→

Spinlock ticket→

Mutex TTAS + spin + block→

Rwsem TTAS + spin + block→

Spinlock ticket→

Mutex TTAS + block→

Rwsem TTAS + block→

Spinlock qspinlock→

Mutex TTAS + spin + block→

Rwsem TTAS + spin + block→

Lock's research efforts

1990s

2011

2014

2016

Adopting NUMA aware locks is not easy

Page 12: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Issues with NUMA-aware primitives

● Memory footprint overhead– Cohort lock single instance: 1600 bytes– Example: 1–4 GB of lock space vs 38 MB of Linux’s

lock for 10 M inodes

● Does not support blocking/parking behavior

Page 13: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Blocking/parking approach● Under subscription

– #threads <= #cores

● Over subscription– #threads > #cores

● Spin-then-park strategy1) Spin for a certain duration

2) Add to a parking list

3) Schedule out (park/block)

under-subscription

Lock

thro

ughp

ut

→

#thread →

Page 14: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Blocking/parking approach● Under subscription

– #threads <= #cores

● Over subscription– #threads > #cores

● Spin-then-park strategy1) Spin for a certain duration

2) Add to a parking list

3) Schedule out (park/block)

under-subscription

Lock

thro

ughp

ut

→

#thread →

over-subscription

Page 15: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Blocking/parking approach● Under subscription

– #threads <= #cores

● Over subscription– #threads > #cores

● Spin-then-park strategy1) Spin for a certain duration

2) Add to a parking list

3) Schedule out (park/block)

under-subscription

Lock

thro

ughp

ut

→

#thread →

over-subscription

Spinning

Page 16: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Blocking/parking approach● Under subscription

– #threads <= #cores

● Over subscription– #threads > #cores

● Spin-then-park strategy1) Spin for a certain duration

2) Add to a parking list

3) Schedule out (park/block)

under-subscription

Lock

thro

ughp

ut

→

#thread →

over-subscription

Spinning

Parking

Page 17: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Blocking/parking approach● Under subscription

– #threads <= #cores

● Over subscription– #threads > #cores

● Spin-then-park strategy1) Spin for a certain duration

2) Add to a parking list

3) Schedule out (park/block)

under-subscription

Lock

thro

ughp

ut

→

#thread →

over-subscription

Spinning

Parking

Spin + park

Page 18: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Issues with blocking synchronization primitives

● High memory footprint for NUMA-aware locks● Inefficient blocking strategy

– Scheduling overhead in the critical path– Cache-line contention while scheduling out

Page 19: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

CST lock

● NUMA-aware lock● Low memory footprint

– Allocate socket specific data structure when used– 1.5–10X memory less memory consumption

● Efficient parking/wake-up strategy– Limit the spinning up to a waiter’s time quantum– Pass the lock to an active waiter– Improves scalability by 1.2–4.7X

Page 20: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

CST lock design

● NUMA-aware lock➢ Cohort lock principle

+ Mitigates cache-line contention and bouncing

● Memory efficient data structure➢ Allocate socket structure (snode) when used➢ Snodes are active until the life-cycle of the lock

+ Does not stress the memory allocator

Page 21: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

CST lock design

● NUMA-aware parking list➢ Maintain separate per-socket parking lists for

readers and writers

+ Mitigates cache-line contention in over-subscribed scenario

+ Allows distributed wake-up of parked readers

Page 22: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

CST lock design

● Remove scheduler intervention➢ Pass the lock to a spinning waiter➢ Waiters park themselves if more than one tasks are

running on a CPU (system load)

+ Scheduler not involved in the critical path

+ Guarantees forward progress of the system

Page 23: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Lock instantiation

socket_listsocket_list

global_tailglobal_tail

Threads: ● Initially no snodes are allocated

● Thread in a particular socket initiates an allocation

Page 24: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Lock instantiation

socket_listsocket_list

global_tailglobal_tail

T1/S1T1/S1Threads:

Socket 1

socket_list [S1]socket_list [S1]

● Initially no snodes are allocated

● Thread in a particular socket initiates an allocation

Page 25: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Lock instantiation

socket_listsocket_list

global_tailglobal_tail

T1/S1T1/S1Threads:

Socket 1

T2/S1T2/S1 T3/S1T3/S1

socket_list [S1]socket_list [S1]

● Initially no snodes are allocated

● Thread in a particular socket initiates an allocation

Page 26: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Lock instantiation

socket_listsocket_list

global_tailglobal_tail

T1/S1T1/S1Threads:

Socket 1

Socket 2

T2/S1T2/S1 T3/S1T3/S1 T4/S2T4/S2

socket_list [S1]socket_list [S1]socket_list [S1, S2]socket_list [S1, S2]

● Initially no snodes are allocated

● Thread in a particular socket initiates an allocation

Page 27: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

CST lock phaseCST lock instance

socket_listsocket_list

Threads:

global_tailglobal_tail

● Allocate thread specific structure on the stack● Three states for each node

– L locked→– UW unparked/spinning waiter→– PW parked / blocked / scheduled out waiter→

Page 28: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

CST lock phaseCST lock instance T1/S1T1/S1

socket_listsocket_list

waiting_tailwaiting_tail

parking_tailparking_tail

UWUW

snode_nextsnode_next

LL

Threads:

Socket 1

socket_list [S1]socket_list [S1]

global_tailglobal_tail

● Allocate thread specific structure on the stack● Three states for each node

– L locked→– UW unparked/spinning waiter→– PW parked / blocked / scheduled out waiter→

Page 29: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

CST lock phaseCST lock instance T1/S1T1/S1

socket_listsocket_list

waiting_tailwaiting_tail

parking_tailparking_tail

UWUW

snode_nextsnode_next

nextnext

p_nextp_next

LLLL

Threads:

Socket 1

socket_list [S1]socket_list [S1]

global_tailglobal_tail

T1

● Allocate thread specific structure on the stack● Three states for each node

– L locked→

– UW unparked/spinning waiter→

– PW parked / blocked / scheduled out waiter→

Page 30: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

CST lock phaseCST lock instance T1/S1T1/S1

socket_listsocket_list

waiting_tailwaiting_tail

parking_tailparking_tail

UWUW

snode_nextsnode_next

nextnext

p_nextp_next

LL

Acquire global lock

LL

Threads:

Socket 1

socket_list [S1]socket_list [S1]

global_tailglobal_tail

T1

● Allocate thread specific structure on the stack● Three states for each node

– L locked→

– UW unparked/spinning waiter→

– PW parked / blocked / scheduled out waiter→

Page 31: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

CST lock phaseCST lock instance T1/S1T1/S1

socket_listsocket_list

waiting_tailwaiting_tail

parking_tailparking_tail

UWUW

snode_nextsnode_next

nextnext

p_nextp_next

LLLL

Threads:

nextnext

p_nextp_next

UWUW

T2/S1T2/S1 T3/S1T3/S1

Socket 1

socket_list [S1]socket_list [S1]

global_tailglobal_tail nextnext

p_nextp_next

UWUWT1 T2 T3

● Allocate thread specific structure on the stack● Three states for each node

– L locked→

– UW unparked/spinning waiter→

– PW parked / blocked / scheduled out waiter→

Page 32: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

CST lock phase

nextnext

p_nextp_next

LL

CST lock instance T1/S1T1/S1

waiting_tailwaiting_tail

parking_tailparking_tail

UWUW

snode_nextsnode_next

socket_listsocket_list

waiting_tailwaiting_tail

parking_tailparking_tail

UWUW

snode_nextsnode_next

nextnext

p_nextp_next

LLLL

Threads:

nextnext

p_nextp_next

UWUW

T2/S1T2/S1 T3/S1T3/S1

Socket 1

Socket 2

socket_list [S1]socket_list [S1]socket_list [S1, S2]socket_list [S1, S2]

global_tailglobal_tail

T4/S2T4/S2

nextnext

p_nextp_next

UWUWT1 T2 T3

T4

● Allocate thread specific structure on the stack● Three states for each node

– L locked→

– UW unparked/spinning waiter→

– PW parked / blocked / scheduled out waiter→

Page 33: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

CST lock phase: blocking/parking

nextnext

p_nextp_next

LL

CST lock instance T1/S1T1/S1

waiting_tailwaiting_tail

parking_tailparking_tail

UWUW

snode_nextsnode_next

socket_listCV

socket_listCV

waiting_tailwaiting_tail

parking_tailparking_tail

UWUW

snode_nextsnode_next

nextnext

p_nextp_next

LLLL

Threads:

nextnext

p_nextp_next

UWUW

T2/S1T2/S1 T3/S1T3/S1

Socket 1

Socket 2

PWPWsocket_list [S1]CV

socket_list [S1]CV

socket_list [S1, S2]CV

socket_list [S1, S2]CV

global_tailglobal_tail

T4/S2T4/S2

nextnext

p_nextp_next

UWUW

T2/S1T2/S1

● Before scheduling out, waiters atomically– Update the status from UW to PW– Add themselves to the parking list

T1 T2 T3

T4

Page 34: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

CST unlock phaseCST lock instance

T1/S1T1/S1

socket_list [S1]socket_list [S1]

global_tailglobal_tailwaiting_tailwaiting_tail

parking_tailparking_tail

UWUW

snode_nextsnode_next

nextnext

p_nextp_next

LLLL

Threads:

nextnext

p_nextp_next

UWUW

T2/S1T2/S1

Socket 1

T1 T2

T3/S1T3/S1

nextnext

p_nextp_next

UWUWT3

Pass the lock to a spinning waiter

Page 35: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

CST unlock phaseCST lock instance

T1/S1T1/S1

socket_list [S1]socket_list [S1]

global_tailglobal_tailwaiting_tailwaiting_tail

parking_tailparking_tail

UWUW

snode_nextsnode_next

nextnext

p_nextp_next

LLLL

Threads:

nextnext

p_nextp_next

UWUW

T2/S1T2/S1

Socket 1

T1 T2

T3/S1T3/S1

nextnext

p_nextp_next

UWUWT3

Pass the lock to a spinning waiter

T2/S1T2/S1

Page 36: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

CST unlock phaseCST lock instance

T1/S1T1/S1

socket_list [S1]socket_list [S1]

global_tailglobal_tailwaiting_tailwaiting_tail

parking_tailparking_tail

UWUW

snode_nextsnode_next

nextnext

p_nextp_next

LLLL

Threads:

nextnext

p_nextp_next

UWUW

T2/S1T2/S1

Socket 1

T1 T2

T3/S1T3/S1

nextnext

p_nextp_next

UWUWT3

socket_list [S1]socket_list [S1]

global_tailglobal_tailwaiting_tailwaiting_tail

parking_tailparking_tail

UWUW

snode_nextsnode_next

nextnext

p_nextp_next

LLLLnextnext

p_nextp_next

UWUW

Socket 1

PWPWT1 T2nextnext

p_nextp_next

UWUWT3

Pass the lock to a spinning waiter

T2/S1T2/S1

Page 37: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

CST unlock phaseCST lock instance

T1/S1T1/S1

socket_list [S1]socket_list [S1]

global_tailglobal_tailwaiting_tailwaiting_tail

parking_tailparking_tail

UWUW

snode_nextsnode_next

nextnext

p_nextp_next

LLLL

Threads:

nextnext

p_nextp_next

UWUW

T2/S1T2/S1

Socket 1

T1 T2

T3/S1T3/S1

nextnext

p_nextp_next

UWUWT3

socket_list [S1]socket_list [S1]

global_tailglobal_tailwaiting_tailwaiting_tail

parking_tailparking_tail

UWUW

snode_nextsnode_next

nextnext

p_nextp_next

LLLLnextnext

p_nextp_next

UWUW

Socket 1

PWPWT1 T2nextnext

p_nextp_next

UWUWT3

socket_list [S1]socket_list [S1]

global_tailglobal_tailwaiting_tailwaiting_tail

parking_tailparking_tail

UWUW

snode_nextsnode_next

LLnextnext

p_nextp_next

UWUW

Socket 1

PWPWT2nextnext

p_nextp_next

LLT3

Pass the lock to a spinning waiter

T2/S1T2/S1

Page 38: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Implementation

● Implemented in the Linux kernel● Structures modified

– File system: inode– Memory management: mmap_sem

● Please see our paper– Read-write lock– Pseudo code

https://github.com/sslab-gatech/cst-locks

Page 39: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Evaluation

● Performance of locks in terms of scalability and memory footprint?

● Blocking/parking strategy effectiveness?

● Setup: 8-socket, 120-core NUMA machine

Page 40: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Case study: PsearchyJo

bs/h

our

● Overcomes memory footprint and scheduling overhead● Uses 1.5–9.1X less memory than the Cohort lock● Improves throughput by 1.4–1.6X

Mem

ory

foot

prin

t (M

B)

0

40

80

120

160

0 20 40 60 80 100 120#thread

Memory utilization

0

40

80

120

160

200

240

0 20 40 60 80 100 120#thread

Vanilla

Cohort

CST

Throughput

Page 41: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Effective parking strategy

● Better performance for both under- and over-subscribed scenario

● Improves scalability by 1.3–3.7X

0

40

80

120

160

200

240

1 2 4 8 16 32 64 128 256#thread

Enumerate a directory (rwsem)

0

0.1

0.2

0.3

0.4

0.5

1 2 4 8 16 32 64 128 256#thread

File creation (mutex)

Vanilla

Cohort

CST

M o

ps /

sec

Page 42: Scalable NUMA-aware Blocking Synchronization …...– Limit the spinning up to a waiter’s time quantum – Pass the lock to an active waiter – Improves scalability by 1.2–4.7X

Conclusion

● Two blocking synchronization primitives– NUMA-aware mutex and read-write semaphore

● Dynamically allocated data structure– Resolve NUMA-aware lock’s footprint issue

● Efficient spin-then-park strategy– Scheduling-aware parking/wake-up strategy– Mitigate scheduler interaction

Thank you!


Waiter Presentation PDF

Waiter Presentation PDF

Documents
WAITER/ WAITRESS

WAITER/ WAITRESS

Documents
Robotic Waiter

Robotic Waiter

Documents
PMQ0400-A 5-08 Dumbwaiters Waupaca Elevator Planning Guide · 2014-09-14 · CABLE-WAITER™ & PACA-WAITER® PLANNING GUIDE PMQ0400.4 Mechanical Features Cable-Waiter Paca-Waiter

PMQ0400-A 5-08 Dumbwaiters Waupaca Elevator Planning Guide · 2014-09-14 · CABLE-WAITER™ & PACA-WAITER® PLANNING GUIDE PMQ0400.4 Mechanical Features Cable-Waiter Paca-Waiter

Documents
The Dumb Waiter Script

The Dumb Waiter Script

Documents
ROBOTIC WAITER FOR RESTAURANTS

ROBOTIC WAITER FOR RESTAURANTS

Documents
Corkscrews for Sale Waiter’s Friends 3 · WF103 Waiter’s friend marked GERMANY. Ernst Steinfeld’s April 10, 1899 German Registered Design No. 116521 for the addition of a wire

Corkscrews for Sale Waiter’s Friends 3 · WF103 Waiter’s friend marked GERMANY. Ernst Steinfeld’s April 10, 1899 German Registered Design No. 116521 for the addition of a wire

Documents
Melissa - Kindergarten teacher and Waiter

Melissa - Kindergarten teacher and Waiter

Documents
Lev Aloha Against Waiter Scams

Lev Aloha Against Waiter Scams

Documents
The Dumb Waiter

The Dumb Waiter

Documents
PUIGPULL AND PULLPLUS WAITER’S CORKSCREWS...ASI 55450 5302 5306 5301 PullPlus Waiter’s Corkscrew, Enameled Steel Handle; 5302 PullPlus Vintage Waiter’s Corkscrew, Stainless Steel

PUIGPULL AND PULLPLUS WAITER’S CORKSCREWS...ASI 55450 5302 5306 5301 PullPlus Waiter’s Corkscrew, Enameled Steel Handle; 5302 PullPlus Vintage Waiter’s Corkscrew, Stainless Steel

Documents
[XLS]stepindia.orgstepindia.org/docs/ap-mot-placement.xls · Web viewRJY-H.K.U-10 RJY-WAITER-10 RJY-WAITER-09 ELURU-H.K.U-09 ELURU-WAITER-09 RJY-WAITER-08 ELURU-H.K.U-08 ELURU-WAITER-08

[XLS]stepindia.orgstepindia.org/docs/ap-mot-placement.xls · Web viewRJY-H.K.U-10 RJY-WAITER-10 RJY-WAITER-09 ELURU-H.K.U-09 ELURU-WAITER-09 RJY-WAITER-08 ELURU-H.K.U-08 ELURU-WAITER-08

Documents
WAITER MANUAL IRCV1 Trainer

WAITER MANUAL IRCV1 Trainer

Documents
CMi5586DXLC133, CMi5486DXLC100HR, … · USA CMi5586DXLC133, CMi5486DXLC100HR, CMi5486DXLC66HR cpuModuleTM User’s Manual BIOS Versions 4.7x Publication CMi5586 2000.05.23

CMi5586DXLC133, CMi5486DXLC100HR, … · USA CMi5586DXLC133, CMi5486DXLC100HR, CMi5486DXLC66HR cpuModuleTM User’s Manual BIOS Versions 4.7x Publication CMi5586 2000.05.23

Documents
Primetray-A smart Waiter

Primetray-A smart Waiter

Business
Corkscrews for Sale Waiter’s Friends - Bullworks.Net · Corkscrews for Sale Waiter’s Friends Don Bull, P. O. Box 596, Wirtz, VA 24184 USA email: corkscrew@bullworks.net (Please

Corkscrews for Sale Waiter’s Friends - Bullworks.Net · Corkscrews for Sale Waiter’s Friends Don Bull, P. O. Box 596, Wirtz, VA 24184 USA email: [email protected] (Please

Documents
Cruise Shipboard Property Management SystemSelect the Keeper, Assistant Keeper, Head Waiter, Waiter, and Assistant Waiter from the drop-down list. 6. Click Apply to save the Housekeeping

Cruise Shipboard Property Management SystemSelect the Keeper, Assistant Keeper, Head Waiter, Waiter, and Assistant Waiter from the drop-down list. 6. Click Apply to save the Housekeeping

Documents
Attributes of Waiter

Attributes of Waiter

Documents
WAITER MANUAL IRCV1 _Trainer_.pdf

WAITER MANUAL IRCV1 _Trainer_.pdf

Documents
Multiculturalism in Zadie Smith’s Short Story “The Waiter’s Wife”

Multiculturalism in Zadie Smith’s Short Story “The Waiter’s Wife”

Documents
Thank you for purchasing our Dual Hinge Waiter’s Corkscrew!ML.pdf · Our Dual Hinge Waiter’s Corkscrew offers the simplest and most basic way to open a bottle of wine — and

Thank you for purchasing our Dual Hinge Waiter’s Corkscrew!ML.pdf · Our Dual Hinge Waiter’s Corkscrew offers the simplest and most basic way to open a bottle of wine — and

Documents
19 Waiter & Waitress Hotel Restaurant

19 Waiter & Waitress Hotel Restaurant

Documents
Waiter Training

Waiter Training

Documents
Waiter Training Manual Preview Copy

Waiter Training Manual Preview Copy

Documents
Corkscrews for Sale Waiter’s Friends 4WF177 Waiter’s friend marked ED. WÜSTHOF SOLINGEN with trident trademark and Andres Kim. The Trident trademark was registered May 9, 1895

Corkscrews for Sale Waiter’s Friends 4WF177 Waiter’s friend marked ED. WÜSTHOF SOLINGEN with trident trademark and Andres Kim. The Trident trademark was registered May 9, 1895

Documents
Importance of A Topless Waiter

Importance of A Topless Waiter

Entertainment & Humor
Waiter-Learner Manual (ENGLISH).indd

Waiter-Learner Manual (ENGLISH).indd

Documents
Waiter on Wheels

Waiter on Wheels

Documents
Waiter & Waitress Book

Waiter & Waitress Book

Documents
ITFT - waiter etiquettes

ITFT - waiter etiquettes

Education