Efficient Parallel Refinement for Hierarchical Radiosity on a DSM computer

1
Efficient Parallel Refinementfor Hierarchical
Radiosity on a DSM computer

• François X. Sillion, Jean-Marc Hasenfratz
• iMAGIS

2
Radiosity
3
Hierarchical Radiosity

• Hierarchical representation (mesh)
• Interactions computed at appropriate level

4
Strategies for Hierarchical Radiosity

• Gathering
• memory consuming (store links)
• Easier dynamic modifications
• Shooting
• Memory efficient
• Requires heuristic to decide shooting level
• Links recomputed as needed

5
Parallel Approaches

• Two approaches
• data exchange via message-passing algorithms
• Shared memory
• Partial solutions possible if natural
  partitioning exists (e.g. inside buildings)
  Fun96,FY97
• Virtual interfaces are harder to handle RAPP97
• Load balancing problemCav99

6
Scheduler

• Force all link refinement operations through a
  scheduler object
• Natural place for
• Parallel synchronization
• Orientation and steering of calculation
• Advantages of using scheduler
• Global view of all pending task at any given time
• Task extraction can be made according to various
  selection criteria

7
Example (sequential) schedulers

• Stack scheduler (depth first refinement)
• Priority scheduler
• Use simple structure (heap)
• Hierarchical level (breadth first)
• Size, energy, error
• Interactive user control
• Random scheduler...

8
Architecture
Main / GUI
Solver
Scheduler
9
Synchronization

• Scheduler
• Single object talks to all refiners gt Danger!
• Use simple blocks of refinement jobs
• Hierarchical data structure
• Consistency of hierarchical scene structure
• Interactions
• Links or energy representations

10
Test scenes
VRLab - 51 182 polygons
Aircraft - 184 456 polygons
Office - 5 285 polygons
11
Measurements

• Hardware architecture
• ccNUMA SGI 2000 computer with 64 microprocessors
• Limit to 40 microprocessors R10000 at 195MHz

12
Measurements

• Time measurements
• Refinement times system call which return clock
  ticks
• Memory access, cache access perfex software
  tool which uses the 31 hard counters of R10,000

13
Results

• CPU Refinement time

14
Results
Speed-up
15
Results
Influence of the size of link blocks on overall
CPU time
16
Results
Memory used before and during the iterations
17
Conclusions

• Very simple atomic tasks
• Easily managed with a single scheduler structure
• Easily implemented on top of an existing
  radiosity simulation code
• Thread setup
• New link creation upon refinement decision

18
Future work

• Understanding the peculiar behaviour observed
  for the aircraft scene
• Dealing with graphics resources for optimized
  calculations using graphics hardware

19
Acknowledgements

• Peter Kipfer contributed to the design and early
  implementation of this work.
• Thanks to Centre Charles Hermite for providing
  access to its computational resources
• Laurent Alonso provided useful advice on
  performance questions.
• This work was supported in part by the European
  Unions ESPRIT project 24944, ARCADE (Making
  Radiosity Usable).