e86205a308
Fix FFmpeg latency test code
2020-05-15 08:16:06 +03:00
a2d68b6d15
Fix parameters of parse_csv()
2020-05-15 07:53:43 +03:00
c0cb1a7cf2
Deallocate frames in example code
2020-05-15 07:53:15 +03:00
d287972f70
Fix latency test for uvgRTP
2020-05-15 07:40:12 +03:00
7b73aede33
Add pull_frame(timeout)
...
Instead of blocking until the library is destroyed, add ability
to block at most N milliseconds before returning.
The code returns early if a frame is received within the specified
timeframe but will block at most "timeout "milliseconds
2020-05-11 17:36:13 +03:00
037ea2b146
Fix best effort FPS calculations
2020-05-11 11:19:23 +03:00
4bae67527c
Set FFmpeg log level to panic
2020-05-11 11:19:14 +03:00
180071db52
Add --unit parameter
2020-05-11 11:18:53 +03:00
cbb19076cb
Calculate goodput estimate for receiver
2020-05-07 08:41:45 +03:00
d40c071167
Add ability to provide a list of executables
...
--exec can be used to provide a list of executables that should be
used to run the benchmarks
2020-05-06 15:22:25 +03:00
d3b2fcc32f
Create strict and best effort FPS modes
...
Strict mode will stop the benchmark script if the frame is not sent
within the timeframe specified by the FPS value.
Best-effort mode tries to keep the FPS steady but if sending the
frame takes longer than it should, it will not exit but keeps
sending until every frame has been sent
2020-05-06 11:30:46 +03:00
7fd1ffba54
Add default address and port to benchmark.pl
2020-05-06 11:22:37 +03:00
867aa02dbd
Rename debug() to uvgrtp_debug()
...
Prevent name clahes with other software, debug is after all
quite a common symbol name
2020-05-06 06:46:13 +03:00
20dec060da
Set default addresses for FFmpeg/uvgRTP
2020-05-05 20:28:24 +03:00
ac66b7ca7a
Use second to last frame to calculate time estimate
2020-05-05 19:37:28 +03:00
e62a7bda3a
Decrease probesize to 32
2020-05-05 19:33:37 +03:00
0a52fb3d26
Improve the FPS parameter
...
Make it possible to provide a list of FPS values that should be tested
and add step parameter that tells how the FPS value should be increased
2020-05-05 19:24:40 +03:00
3624456ae2
Fix FFmpeg's and uvgRTP's FPS calculations
...
If sending takes too long time, the benchmark is stopped
2020-05-05 13:02:12 +03:00
e75f0cb6d7
Output everything that was measured to CSV file
2020-05-05 12:47:50 +03:00
ec6e847310
Update TOTAL_BYTES
2020-05-05 12:47:34 +03:00
e7bcd377d3
Try to infer iter from file name in parse_csv()
2020-05-05 12:11:53 +03:00
977569edb1
Add parse option "all"
...
Similar to best but instead of printing only the best configuration,
print all configuration that meet the criteria
2020-05-04 12:05:06 +03:00
425d257758
Give frames per second
2020-05-04 11:26:00 +03:00
bda503d6cf
Give both single and total goodput
2020-05-04 09:23:42 +03:00
af5f3130ce
Add function for outputting a CSV file of the results
...
First line has the percentage of frames received and the second
line has the goodput of a single thread for that thread configuration
in megabytes
2020-05-02 21:38:35 +03:00
a273763060
Add code for testing latency
2020-05-02 10:59:38 +03:00
0b7d769c61
Improve argument parsing
2020-05-02 10:39:47 +03:00
64ac3d6d59
Add more examples to benchmarks/README.md
2020-05-02 10:39:43 +03:00
81d7f23942
Improve argument parsing
2020-05-01 19:59:34 +03:00
0e1804d8ee
Try to infer iter from file names in parse_all()
2020-05-01 19:33:24 +03:00
f104127d0f
Convert tabs to spaces
2020-05-01 19:32:46 +03:00
01b0240923
Add the number of iterations to file name
2020-05-01 18:26:40 +03:00
737693efd7
Infer role and number of threads from filename
2020-05-01 18:26:12 +03:00
f29918602c
Try to infer library from path
2020-05-01 15:21:38 +03:00
997924c111
Separate frame loss and packet loss
2020-05-01 12:58:06 +03:00
eac06c34cb
Add ability to find the best configuration of FPS and threads
...
By default, the script requires that 99% of the frames/packets have
been received but the limit is adjustable with --threshold parameter
2020-04-30 09:14:07 +03:00
daa8b9246a
Remove find_best.sh
...
The same functionality can be found from parse.pl
2020-04-30 07:10:11 +03:00
f01b0d3dfd
Ignore invalid benchmark runs
2020-04-30 06:58:58 +03:00
3525ee0e3e
Fix SRTP example with user-managed keys
2020-04-30 01:49:45 +03:00
c2438e3017
Add support for user-managed SRTP session
...
Because the architecture has changed a bit, the context is initialized
only after add_srtp_ctx() has been called to prevent the user from
sending unencrypted messages.
2020-04-30 01:45:28 +03:00
527b5e5c09
Set correct FPS in FFmpeg send benchmark
2020-04-29 20:47:23 +03:00
a41e8b93f3
Fix FFmpeg receiver code
...
The code bailed out too early
2020-04-29 20:47:10 +03:00
4f021a287b
Use iter instead of lines
...
They're the same thing and as iter is a mandatory parameter,
it's better to use it instead
2020-04-29 20:06:27 +03:00
ff8ef9bdaf
Add code for parsing FFmpeg's recv benchmark output
2020-04-29 20:06:00 +03:00
aca3932cb9
Add code for parsing FFmpeg's send benchmark output
2020-04-29 19:53:35 +03:00
2f8db432be
Use the arrival time of last succesful frame
2020-04-29 14:12:25 +03:00
e77dde544d
Add missing my keyword to generic receiver
2020-04-29 13:39:16 +03:00
a40f72ce6a
Fix total goodput calculation for sender
2020-04-28 11:38:13 +03:00
7e28da0cbb
Calculate average time for the benchmark run
2020-04-28 11:26:33 +03:00
1a24079a4a
Print estimate of time even if all frames were not received
2020-04-28 11:19:40 +03:00
Aaro Altonen