Noack[1] has compared two compact XML
encodings, namely Fast Infoset (FI) and Efficient XML
Interchange (EXI), with the
STRING
-encoded XML specified in Subclause 6.3 of the IEC
61499 Compliance Profile for Feasibility Demonstrations. Noack's
results indicate that EXI is the most efficient, both in terms of
message size and parsing time, than either of the other two encoding
methods.
The table below presents a rough comparison of the encoding and
decoding times using
String
-encoded XML versus MGT_REQ
encoding as implemented with the FBRT
runtime on several platforms. It can be seen that the
MGT_REQ
encoding has an advantage of approximately 10:1 for both encoding and
decoding.
| Platform | ID |
Time, µs | |||
|---|---|---|---|---|---|
MGT_REQ |
XML STRING |
||||
| Encode | Decode | Encode | Decode | ||
| WIN10 | 1 | 0.7 | 1.1 | 6.6 | 10.2 |
| 2 | 0.4 | 0.7 | 10.1 | 18.6 | |
| 3 | 0.4 | 0.5 | 6.3 | 10.5 | |
| 4 | 0.4 | 0.8 | 3.8 | 4.4 | |
| 5 | 0.4 | 0.7 | 5.6 | 9.7 | |
| 6 | 0.4 | 0.5 | 6.1 | 11.8 | |
| 7 | 0.5 | 0.5 | 5.8 | 10.3 | |
| BODHI | 1 | 0.8 | 1.0 | 8.2 | 15.2 |
| 2 | 0.6 | 1.1 | 9.4 | 19.8 | |
| 3 | 0.6 | 0.9 | 7.7 | 15.4 | |
| 4 | 0.3 | 0.7 | 4.7 | 7.3 | |
| 5 | 0.6 | 0.9 | 7.6 | 16.5 | |
| 6 | 0.8 | 1.2 | 9.0 | 16.8 | |
| 7 | 0.5 | 0.9 | 8.1 | 16.1 | |
| RASPI | 1 | 3.1 | 4.3 | 28.4 | 64.0 |
| 2 | 3.1 | 4.3 | 33.1 | 75.7 | |
| 3 | 3.0 | 4.0 | 28.6 | 63.7 | |
| 4 | 2.1 | 3.3 | 14.81 | 29.0 | |
| 5 | 2.9 | 4.0 | 28.5 | 63.7 | |
| 6 | 2.9 | 4.0 | 30.3 | 68.0 | |
| 7 | 2.8 | 3.9 | 29.4 | 66.6 | |
| WIN10(2) | 1 | 0.48 | 0.46 | 4.77 | 4.92 |
| 2 | 0.42 | 0.31 | 6.14 | 6.43 | |
| 3 | 0.16 | 0.17 | 1.52 | 2.99 | |
| 4 | 0.22 | 0.29 | 0.93 | 1.43 | |
| 5 | 0.19 | 0.19 | 1.44 | 2.86 | |
| 6 | 0.17 | 0.17 | 1.73 | 3.09 | |
| 7 | 0.16 | 0.17 | 1.54 | 3.07 | |