jsc.SamplingProfile

interface SamplingProfile

• bytecodes: string

  A formatted breakdown of the samples by execution tier, followed by the individual bytecode locations that were sampled most often. Useful for seeing which tier hot code is running in: time spent in LLInt or Baseline means the code has not been optimized (yet).

  Example output:

  Tier breakdown:
  -----------------------------------
  LLInt:                   106  (1.545640%)
  Baseline:               2355  (34.339458%)
  DFG:                    3290  (47.973170%)
  FTL:                     833  (12.146398%)
  js builtin:              132  (1.924759%)
  Wasm:                      0  (0.000000%)
  Host:                    111  (1.618548%)
  RegExp:                   15  (0.218723%)
  C/C++:                     0  (0.000000%)
  Unknown Executable:      148  (2.158064%)
  
  Hottest bytecodes as <numSamples   'functionName#hash:JITType:bytecodeIndex'>
  273    'visit#<nil>:DFG:bc#63'
  121    'walk#<nil>:DFG:bc#7'
  119    '#<nil>:Baseline:bc#1'
  82    'Function#<nil>:None:<nil>'
  66    '#<nil>:DFG:bc#11'
  65    '#<nil>:DFG:bc#33'
  58    '#<nil>:Baseline:bc#7'
  53    '#<nil>:Baseline:bc#23'
  50    'forEach#<nil>:DFG:bc#83'
  49    'pop#<nil>:FTL:bc#65'
  47    '#<nil>:DFG:bc#99'
  45    '#<nil>:DFG:bc#16'
  44    '#<nil>:DFG:bc#7'
  44    '#<nil>:Baseline:bc#30'
  44    'push#<nil>:FTL:bc#214'
  41    '#<nil>:DFG:bc#50'
  39    'get#<nil>:DFG:bc#27'
  39    '#<nil>:Baseline:bc#0'
  36    '#<nil>:DFG:bc#27'
  36    'Dictionary#<nil>:DFG:bc#41'
  36    'visit#<nil>:DFG:bc#81'
  36    'get#<nil>:FTL:bc#11'
  32    'push#<nil>:FTL:bc#49'
  31    '#<nil>:DFG:bc#76'
  31    '#<nil>:DFG:bc#10'
  31    '#<nil>:DFG:bc#73'
  29    'set#<nil>:DFG:bc#28'
  28    'in_boolean_context#<nil>:DFG:bc#104'
  28    '#<nil>:Baseline:<nil>'
  28    'regExpSplitFast#<nil>:None:<nil>'
  26    'visit#<nil>:DFG:bc#95'
  26    'pop#<nil>:FTL:bc#120'
  25    '#<nil>:DFG:bc#23'
  25    'push#<nil>:FTL:bc#152'
  24    'push#<nil>:FTL:bc#262'
  24    '#<nil>:FTL:bc#10'
  23    'is_identifier_char#<nil>:DFG:bc#22'
  23    'visit#<nil>:DFG:bc#22'
  22    '#<nil>:FTL:bc#27'
  22    'indexOf#<nil>:None:<nil>'

• functions: string

  A formatted summary of the functions that were most often at the top of the stack when a sample was taken (that is, ranked by self time). Each entry shows the number of samples, the function name, its code hash, and the ID of its source file.

  Example output:

  
  Sampling rate: 100.000000 microseconds. Total samples: 6858
  Top functions as <numSamples  'functionName#hash:sourceID'>
  2948    '#<nil>:8'
  393    'visit#<nil>:8'
  263    'push#<nil>:8'
  164    'scan_ref_scoped#<nil>:8'
  164    'walk#<nil>:8'
  144    'pop#<nil>:8'
  107    'extract_candidates#<nil>:8'
   94    'get#<nil>:8'
   82    'Function#<nil>:4294967295'
   79    'set#<nil>:8'
   67    'forEach#<nil>:5'
   58    'collapse#<nil>:8'

• stackTraces: SamplingProfileStackTraces

  The raw samples: the full stack captured at each sampling interval, which is what functions and bytecodes are computed from. Use this to build your own reports, such as a flame graph or inclusive timings.