XerxesProfileResults-01

XerxesProfileResults.txt

Same use case of decode a 1080p video 1000 times.

This was done using the "Perf" tool

Compiler settings
- opt level: -Og
- -g



------------------------------------------------------------------

Run #1:

Most expensive functions:
     7.74%  h264dec  h264dec                   [.] WelsDec::DecodeCurrentAccessUnit
     7.40%  h264dec  h264dec                   [.] WelsDec::WelsResidualBlockCavlc


WelsDec::DecodeCurrentAccessUnit - Expensive lines

 32.05 │        rep    movsq %ds:(%rsi),%es:(%rdi)



WelsDec::WelsResidualBlockCavlc - Expensive lines

  7.08 │       movzbl (%r9,%rdx,1),%edx 



------------------------------------------------------------------

Run #2:

Most expensive functions:

     7.62%  h264dec  h264dec                   [.] WelsDec::DecodeCurrentAccessUnit
     7.38%  h264dec  h264dec                   [.] WelsDec::WelsResidualBlockCavlc

WelsDec::DecodeCurrentAccessUnit - Expensive lines

 32.72 │        rep    movsq %ds:(%rsi),%es:(%rdi)


WelsDec::WelsResidualBlockCavlc - Expensive lines

  6.83 │       movzbl (%r9,%rdx,1),%edx


------------------------------------------------------------------