Personal tools
The Open Lighting Project has moved!

We've launched our new site at www.openlighting.org. This wiki will remain and be updated with more technical information.

Difference between revisions of "OLA Performance Stats"

From wiki.openlighting.org

Jump to: navigation, search
(Created page with 'This is a brain dump of random OLA performance metrics. Unless otherwise mentioned these were performed on a MacBook Pro 2.4GHz Intel Core 2 Duo === Idle Test === Receiving…')
 
 
(5 intermediate revisions by 2 users not shown)
Line 1: Line 1:
This is a brain dump of random [[OLA]] performance metrics. Unless otherwise mentioned these were performed on a MacBook Pro 2.4GHz Intel Core 2 Duo
+
This is a brain dump of random [[OLA]] performance metrics. Unless otherwise mentioned these were performed on a MacBook Pro 2.4GHz Intel Core 2 Duo . More general information is at [[Pipe vs Unix Socket Performance]]
 +
 
 +
Data can be gathered with [[ola_mon]]
  
 
=== Idle Test ===
 
=== Idle Test ===
Line 12: Line 14:
  
 
Patching an E1.31 Output Port results in a loop time of 125uS  @ 200fps. OLA can sustain at least 2200 frames per second in this mode, where the loop time drops to 70uS. This was tested with a single universe and 4 universes.
 
Patching an E1.31 Output Port results in a loop time of 125uS  @ 200fps. OLA can sustain at least 2200 frames per second in this mode, where the loop time drops to 70uS. This was tested with a single universe and 4 universes.
 +
 +
=== Arduino RGB Mixer ===
 +
 +
This can only do about 22 full frames per second. The loop time increases to 2000uS.
 +
  
 
=== RPC Throughput ===
 
=== RPC Throughput ===
  
 +
The maximum with no ports/clients patched is around 20k rpcs/sec. At this point olad isn't reading the data fast enough and the socket buffers fill up and we losing framing.
  
  

Latest revision as of 18:17, 28 February 2012

This is a brain dump of random OLA performance metrics. Unless otherwise mentioned these were performed on a MacBook Pro 2.4GHz Intel Core 2 Duo . More general information is at Pipe vs Unix Socket Performance

Data can be gathered with ola_mon

Idle Test

Receiving an Streaming DMX update for a non-active universe with 512 channels is 40-50uS. This represents the best-case turn around time for an RPC.

UsbPro Test

Patching the Usb Pro Output port causes the average loop time to increase to 150uS for a 512 channel packet. 200 frames/second is about the upper limit the Usb Pro can support.

E1.31 Test

Patching an E1.31 Output Port results in a loop time of 125uS @ 200fps. OLA can sustain at least 2200 frames per second in this mode, where the loop time drops to 70uS. This was tested with a single universe and 4 universes.

Arduino RGB Mixer

This can only do about 22 full frames per second. The loop time increases to 2000uS.


RPC Throughput

The maximum with no ports/clients patched is around 20k rpcs/sec. At this point olad isn't reading the data fast enough and the socket buffers fill up and we losing framing.


TODO

The average isn't the best measurement, we should use the working time for just dmx queries.