1. Automatic Performance Setting for Dynamic Voltage Scaling
Krisztián Flautner
Steve Reinhardt
Trevor Mudge
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

2. Overview A mechanism for quantifying the user experience.
Metric: response time.
Automatic, no user program modifications required.
Run-time feedback to the kernel.
Guiding performance setting of DVS processors.
For interactive episodes: slow down processor to save energy when response times are fast enough.
For periodic events: track periodicity, utilization and inter-task communication to establish necessary performance.
Simulated and experimental results.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

3. Dynamic Voltage Scaling
Execute only as fast as necessary to meet deadlines.
Running fast and idling is not energy efficient.
Power = Capacitance • voltage2 • frequency
Voltage is proportional to the frequency.
Reduce f and v to match performance demands.
Reduced frequency implies longer execution time.
Energy ~ voltage2
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

4. Why bother? ? Higher performance = increased power consumption. 100
386
486
Pentium(R)
MMX
Pentium Pro
(R)
Pentium II (R) 1 10
100
1.5m 1m
0.8m
0.6m
0.35m
0.25m
0.18m
0.13m
Max Power (Watts) ?
Source: Intel
Note the logarithmic scale on the y axis.
Higher performance = increased power consumption.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

5. Power Density! Rocket Nozzle Sun’s Nuclear Reactor Surface ? Hot plate
Source: Intel
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

6. Small performance reduction = big energy savings
Graph based on
Intel XScale data
20% performance reduction = 32% energy reduction
40% performance reduction = 55% energy reduction
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

7. Processors supporting DVS
lpARM
Intel SA-1100
Transmeta Crusoe 5600
Intel XScale
Intel XScale Demo
Min.
8Mhz
1.1V
1.8mW
59Mhz
0.79V
106mW
500Mhz
1.2V
~1W
150Mhz
0.75V
40mW
Max.
100Mhz
3.3V
220mW
251Mhz
1.65V
964mW
700Mhz
1.6V
~2W
800Mhz
1.5V
900mW
1000Mhz
1.75V
1.45W
Process
0.6
0.35
0.18
Max/min energy 9
4.4
1.8 4
5.4
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

8. Some recent desktop processors
Intel Pentium IV
Intel Pentium III
AMD Athlon
Model 4
MPC 7450
Core
1.7V
1.35V
1.65V
1.75V
1.75V
1.8V
1.8V
I/O
400Mhz
100Mhz, 133Mhz
3.3V
200Mhz, 266Mhz
1.6V
133Mhz
1.8V-2.5V
Process
0.18
Max. Power
66.3W
12W
19.1W
38W
66W
17W
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

9. Performance setting algorithms
Programmer specified
Works well but requires explicit specification of deadlines.
Interval based algorithms
Use the ratio of idle to busy time to guide DVS.
Only work well if processor utilization is regular.
No service quality guarantees.
Ours: episode classification based
Find important execution episodes – predict their performance.
Works with existing user programs.
Works well with irregular workloads.
Uses information in kernel to derive deadlines automatically.
Impact on response time is automatically quantified.
Performance can be adapted to the user’s preference.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

10. Episode classification
Interactive episodes
When the user is waiting for the computer to respond.
Periodic episodes
Producer (e.g. MP3 player).
Consumer (e.g. sound daemon).
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

11. A utilization trace
Each horizontal quantum is a millisecond, height corresponds to the utilization in that quantum.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

12. Episode classification
Interactive (Acrobat Reader), Producer (MP3 playback), and Consumer (esd sound daemon) episodes.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

13. Mouse movement
X server updates screen every ~10ms. Update takes ~0.25ms.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

14. Interactive episodes
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

15. Interactive episodes can include idle time
Waiting for data from the network during a run of Netscape. Page rendering starts after 250ms.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

16. Finding interactive episodes
One way: mouse click indicates start, idle time indicates end.
Inaccurate, latency in finding the end of the episode.
Our approach: track inter-task communication.
Start of an interactive episode:
X server sends a message to another task.
During interactive episode:
Keep track of communicating tasks (episode’s task set).
Compute desired metrics.
Conditions for ending the episode (applied to tasks in task set):
No tasks are executing.
Data written by the tasks have been consumed.
No task was preempted the last time it ran.
No tasks are blocked on I/O.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

17. Characteristics of Interactive Episodes
Faster is not necessarily better.
Human perception has finite resolution.
Perception threshold is ~50ms.
The goal is to run fast enough to meet the perception threshold, no point to running any faster.
Many interactive episodes are already fast enough.
More will be imperceptible in the near future.
200ms perception threshold today estimates work done during 50ms 3 years from now.
Slow down the processor!
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

18. Time above the perception threshold
200ms perception threshold estimates 50ms perception threshold on a processor 3 years from now (performance doubles every ~1.5 years).
Time above the perception threshold is given as a percentage of time spent in all interactive episodes.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

19. The key: performance-setting algorithm
Use episode detection and classification.
Interactive episodes.
Periodic episodes (producer and consumer).
Performance-setting on a per episode basis.
Stretch episodes to their deadlines.
Interactive episode: perception threshold.
Stretch producer to consumer.
No modification of existing programs needed.
Works with irregular processor utilization and multiprogramming.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

20. Cumulative interactive episode length distribution
Minimum performance level sufficient
Max. performance
Notice: most time is spent in very few of the episodes.
For most of the episodes the minimum performance level (assuming 5x performance scaling) is sufficient.
FrameMaker
Cumulative number
Cumulative time
Episode length (sec)
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

21. Performance-setting strategy for interactive episodes
Predict the performance factor that would be correct most of the time (not for most events).
Based on past optimal performance factors.
Limit worst case impact on response time.
Run at full performance after PanicThreshold is reached.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

22. Performance-setting for interactive episodes
At the beginning of the episode
Wait 5ms before transition to ignore short episodes
Switch to predicted performance level.
During the episode
If episode duration reaches PanicThreshold, switch to maximum performance.
At the end of the episode
Estimate full performance episode duration.
Compute optimum performance level for past episode.
Compute new prediction based on optimum settings.
PanicThreshold = PerceptionThreshold(1 + PerformanceFactor)
Predicted PerformanceFactor is the average of past optimum settings, weighted by the corresponding episode lengths.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

23. Performance-setting algorithm
Periodic activity detected
Enter period-sampling mode.
Switch to maximum performance.
Establish base performance level.
Exit period-sampling mode.
Start of interactive episode
If not in period-sampling mode, apply interactive episode performance-setting policy.
End of interactive episode
Update interactive episode statistics.
Switch to base performance level, if there is periodic activity on the machine.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

24. Performance-setting during the Acrobat Reader benchmark (200ms p.t.)
Performance factor
Transitions to maximum performance level are due to reaching the PanicThreshold
Time (sec)
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

25. Performance-setting during the Acrobat Reader + MP3 benchmark (200ms p
Performance-setting during the Acrobat Reader + MP3 benchmark (200ms p.t.)
Full performance for periodic activity.
Transitions due to PanicThreshold
Performance factor
Time (sec)
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

26. Hardware assumptions Minimum performance 150Mhz @ 0.75V
Maximum performance
1.75V
PLL resynch time (stalls execution)
0.02ms
Voltage transition time
1ms
Assumptions based on Intel Xscale.
We assume that processor switches to sleep mode when it is not executing an episode.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

27. Energy factors (no MP3)
This is the projected energy savings assuming that processor goes into idle mode whenever it is not active.
This is a conservative assumption. If processor runs more, then more energy savings due to voltage scaling.
Boxed area shows the energy factor range that we can expect on today’s high-end processors.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

28. Energy factors with MP3 playback
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

29. Changes in cumulative episode lengths as the result of performance scaling (Xemacs 50ms p.t. )
Cumulative percentage of time
Before performance scaling
After performance scaling
Episode length (sec)
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

30. Vertigo A DVS implementation for Linux 2.4 kernel.
Currently runs on Transmeta Crusoe.
Test machine: Sony PictureBook (PCG-C1VN) using TM5600 processor (300Mhz-600Mhz).
Goals:
Robust implementation.
Evaluate our algorithms on computers with DVS.
Contrast with conventional DVS algorithm (LongRun).
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

31. Vertigo vs. LongRun LongRun: implemented as part of the processor.
Interval based algorithm (guided by busy vs. idle time).
Min. and max. range is controllable in software.
Vertigo: implemented in OS kernel.
Classification based algorithm.
Distinguishes important from unimportant parts of execution.
Takes the quality of the user experience into account.
Qualitative comparison on following graphs.
The two runs of the benchmarks are close but not identical.
Human repeated the runs of the benchmark.
Transitions to sleep are not shown.
Same perceived interactive performance.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

32. No user activity LongRun Vertigo Performance level Time (s)
Frequency range of the TM5600 processor.
50% = 1.3V
100% = 1.6V
Max. energy savings that should be
expected on this processor is ~34%.
Performance level
Vertigo
Time (s)
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

33. Emacs LongRun Vertigo Performance level Time (s) Performance level
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

34. Acrobat Reader LongRun Vertigo Performance level Time (s)
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

35. Acrobat Reader with sleep transitions
Performance level
LongRun
Frequent transitions to/from sleep mode.
Longer durations without sleeping.
Time (s)
Performance level
Vertigo
Time (s)
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

36. Desired improvements Processor parameters are good enough.
Faster voltage transitions would help a little.
As peak performance gets higher, lower minimum performance is desirable.
More sophisticated prediction algorithms.
Distinguish between episode instances, not just episode types.
Larger performance range for DVS processor.
Puts more pressure on performance-setting algorithm.
More opportunity for energy savings.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

37. Conclusions Many interactive episodes are already fast enough.
More will be fast enough in the near future.
Use Dynamic Voltage Scaling to save energy.
Episode classification based on inter-task communication.
Fast, accurate, no user program modifications required.
Performance-setting based on episode classification.
Works well with multiprogramming, irregular processor utilization.
Ensures high quality interactive performance.
Significant energy savings (10%-80%).
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

38. Future work Evaluate our algorithms on real hardware.
Processors are slowly becoming available.
Impact on interactive performance.
An API to specify episodes.
Light-weight: specify hints, not complete information.
Works in concert with existing detection mechanism.
Apply episode detection to other problems.
Scheduler: can real-time deadlines be detected automatically?
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

39. fin. Automatic Performance Setting for Dynamic Voltage Scaling
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

40. Response time
The time it takes for the computer to respond to user initiated events.
Faster is not always better.
Fundamental limit to what is perceptible to humans.
Movies: frames per second.
Perceptual causality: 50ms-100ms.
Dragging objects on screen: 200ms.
Non-continuous operation: 1-2sec.
The goal is to run fast enough to meet the perception threshold, no point to running any faster.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

41. The performance gap
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

42. Cumulative interactive episode length distribution
Minimum performance level sufficient
Max. performance
Xemacs
Cumulative number
Cumulative time
Episode length (sec)
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

43. Communication between tasks
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling

44. Producer and consumer episodes
Example: MP3 playback through esd sound daemon.
Monitor communications to/from sound daemon.
Distance between producer and consumer episodes determines necessary performance level.
Krisztián Flautner -
Automatic Performance Setting for Dynamic Voltage Scaling