THQGas Powered Games Supreme Commander and Supreme Commander: Forged Alliance

1
THQ/Gas Powered GamesSupreme CommanderandSuprem
e CommanderForged Alliance

• Thread for Performance

2
Supreme Commander runsbest on 4 cores - lets
see how!

• Threading in mid-project can be done!
• Decoupled threads give great performance
• Memory management extends the gains
• Lessons learned

3
Threading was a mid-stream change

• Code was initially single-threaded
• Game demanded more performance
• Changed mid-project (6-12 months into
  development)
• Separate render/sim threads to run at different
  rates
• Support multiple cores
• Limited architecture choices due to existing code
• Using Boost thread library
• Portable, open-source thread library

4
Render split is essential to speed

• Lots of little threads sound, loading, etc.
• Sim thread All simulation
• Render thread Full speed, lt10x per sim tick
• Sync phase Once frame is ready to render
• Sync render and sim
• Fully queued in and out of sim
• Fast

5
Decoupled architecture is built for speed
Issue

• Ready to start a frame and a simulation tick

6
Decoupled architecture is built for speed
Issue
Simulation
Sim Thread Interface
Render
Run decoupled sim and render Fully buffered input
to sim, call via Sim Thread Interface
7
Decoupled architecture is built for speed
Issue

• Render can run repeatedly
• Depends on sim duration

Simulation
Render
Render

Up to 10x per sim tick
8
Decoupled architecture is built for speed

• Fully decoupled? No.
• A few low level systems have locks.
• No major performance impact!

Issue
Simulation
Locks
Render
Render

Up to 10x per sim tick
9
Decoupled architecture is built for speed
Issue
Issue
Simulation
Sim Thread Interface
Render
Render
Render

• Sync sim thread out to render thread,
• via STI again

Up to 10x per sim tick
10
Decoupled architecture is built for speed
Issue
Issue
Simulation
Sim Thread Interface
Sim Thread Interface

• Multiplayer Record everything going through STI
• Send over network

Render
Render
Render

Up to 10x per sim tick
11
Decoupled architecture is built for speed
Issue
Issue
Issue
Simulation
Sim
Render
Render
Render
Render

• And so on

Up to 10x per sim tick
12
Thread model adapts to varying loads

• Architecture scales well with loads
• Render load will often dominate
• Re-render to keep frame rates up
• Sim-heavy map will try to be sim-dominated

13
Displaying frame times cool!

• Thread stats in real time

14
Sometimes, theres more to render

• Render
• Runs as fast as possible
• Simulation
• Sim/render sync
• Both threads synced, fully queued in and out of
  sim

15
Other times, theres more to simulate

• Sim runs across many rendered frames

16
A little sync doesnt slow this code down
Threads are busy most of the time!
Frame n
Frame n1
Sync
Waiting
Busy
Mostly waiting
17
Memory manager gives an additional boost

• Memory If youre not careful in a threaded game
• Memory use can thrash cache but not a problem
  here!
• Memory alloc/free can be slow
• Suspected memory management was problem
• Doing lots of small allocations
• Built code to make it easy to switch mem managers
• Custom mem manager outperforms default
  malloc/free
• Can cause some debugging questions
• Purchased commercial one for Supreme Commander
• Wrote new one for Forged Alliance

18
What are some current bottlenecks?

• Multiplayer all sims run concurrently
• Limited by least-common-denominator machine
• Thats the RTS way
• Monolithic render thread
• Multiple monitors, typically different views
• Possibly split off top part of render for second
  monitor?
• Too expensive/complex for niche feature

19
This was a great learning experience!

• Good intermediate step
• Especially for threading mid-project
• Would do it differently if doing it from scratch
• Target more processor cores
• General worker threads w/dispatch system
• Templates to define an interface to common
  semantics
• Directed work graph/node graph (hard to express)
• Or ?

• The engine is so good, itll be back in Demigod!
• Demigod team using modified Supreme Commander
  engine

20
We learned some DOs and DONTs

• Do
• Architect for threading from the start, if you
  can
• Thread single-threaded code, if you must
• Decouple threads where possible
• Dont
• Be afraid to thread single-threaded code

21
Supreme Commander runsbest on 4 cores thats
how!

• Threading in mid-project can be done!
• Decoupled threads give great performance
• Memory management extends the gains
• Lessons learned

22
So, what do you think?

• Have you tried something like this?
• Successes?
• Failures?
• Have you rejected trying something like this?
• Why?

23
(No Transcript)