Re: profiling emacs-23.1 vs emacs-22.3

[Dan Nicolaescu]

Kenichi Handa <address@hidden> writes:

  > In article <address@hidden>, Dan Nicolaescu <address@hidden> writes:
  > 
  > > It can be seen that 23.1 is quite a bit slower, and that it has a lot of
  > > extra calls to char_table_ref.  
  > > Are those calls necessary?
  > 
  > I found that the syntax of C is mostly defined in the parent of
  > CURRENT_SYNTAX_TABLE, and thus, in the call of SYNTAX_ENTRY (C),
  > the optimization for ASCII in this code (in lisp.h) doesn't work.
  > 
  > /* Almost equivalent to Faref (CT, IDX) with optimization for ASCII
  >    characters.  Do not check validity of CT.  */
  > #define CHAR_TABLE_REF(CT, IDX)                                             
 \
  >   ((ASCII_CHAR_P (IDX)                                                      
 \
  >     && SUB_CHAR_TABLE_P (XCHAR_TABLE (CT)->ascii)                    \
  >     && !NILP (XSUB_CHAR_TABLE (XCHAR_TABLE (CT)->ascii)->contents[IDX])) \
  >    ? XSUB_CHAR_TABLE (XCHAR_TABLE (CT)->ascii)->contents[IDX]               
 \
  >    : char_table_ref ((CT), (IDX)))
  > 
  > Could you try the attached patch?  If it improves the

It does improve performance:

Each sample counts as 0.01 seconds.
  %   cumulative   self              self     total           
 time   seconds   seconds    calls   s/call   s/call  name    
 41.83      8.32     8.32  4864344     0.00     0.00  forw_comment
 16.89     11.68     3.36    32800     0.00     0.00  scan_sexps_forward
  7.99     13.27     1.59    17296     0.00     0.00  scan_lists
  7.44     14.75     1.48    52222     0.00     0.00  re_search_2
  4.63     15.67     0.92 33372393     0.00     0.00  lookup_char_property
  2.82     16.23     0.56 18834514     0.00     0.00  mark_object
  2.46     16.72     0.49 18493110     0.00     0.00  next_interval
  1.51     17.02     0.30  2155558     0.00     0.00  update_syntax_table
  1.51     17.32     0.30 14379550     0.00     0.00  previous_interval
  1.31     17.58     0.26  3314939     0.00     0.00  re_match_2_internal
  1.26     17.83     0.25 23694027     0.00     0.00  Fassq
  1.11     18.05     0.22   202235     0.00     0.00  Fbyte_code
  0.96     18.24     0.19    13460     0.00     0.00  
Fprevious_single_property_change
  0.70     18.38     0.14 57214048     0.00     0.00  Fcdr
  0.70     18.52     0.14       43     0.00     0.02  Fgarbage_collect
  0.70     18.66     0.14  1975818     0.00     0.00  update_interval
  0.55     18.77     0.11   103433     0.00     0.00  skip_chars
  0.35     18.84     0.07  3606562     0.00     0.00  buf_bytepos_to_charpos
  0.35     18.91     0.07   605138     0.00     0.00  Ffuncall
  0.30     18.97     0.06     3267     0.00     0.00  Frassq
  0.30     19.03     0.06     1262     0.00     0.00  Fsetcar

It's still slower than 22.3 though.

One big difference is then time/number of calls to mark_objects 
129733 vs 18834514, so 145 times more calls to mark_object.
Do you know where do those come from?

The number of Fgarbage_collect calls does not increase that much: 
from 37 (for 22.3) to  43 (for 23.1).