Add asyncio and threading performance tests for resampling operations by shauneccles · Pull Request #32 · tuxu/python-samplerate

shauneccles · 2025-11-19T05:02:38Z

Introduce performance tests to verify that the Global Interpreter Lock (GIL) is released during resampling operations, enabling parallel execution in multi-threaded applications. The tests assess the performance improvements and output quality when using different converter types, thread counts and event loops.

shauneccles · 2025-11-19T19:42:43Z

For reference this is current output on my machine
asyncio

PS C:\Users\shaun\python-samplerate-ledfx> uv run pytest tests/test_asyncio_performance.py -s
Uninstalled 1 package in 7ms
Installed 1 package in 23ms
================================================================ test session starts =================================================================
platform win32 -- Python 3.11.9, pytest-9.0.1, pluggy-1.6.0
rootdir: C:\Users\shaun\python-samplerate-ledfx
configfile: pyproject.toml
plugins: asyncio-1.3.0
asyncio: mode=Mode.STRICT, debug=False, asyncio_default_fixture_loop_scope=None, asyncio_default_test_loop_scope=function
collected 28 items                                                                                                                                    

tests\test_asyncio_performance.py
default loop - sinc_fastest async with ThreadPoolExecutor (2 concurrent):
  Sequential: 0.0499s
  Parallel: 0.0527s
  Speedup: 0.95x
  Platform: AMD64
  ⚠️  WARNING: Speedup 0.95x is below expected 1.1x
      This may be due to CI load or platform-specific threading overhead.
.
default loop - sinc_fastest async with ThreadPoolExecutor (4 concurrent):
  Sequential: 0.1063s
  Parallel: 0.1013s
  Speedup: 1.05x
  Platform: AMD64
  ⚠️  WARNING: Speedup 1.05x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
default loop - sinc_fastest async with ThreadPoolExecutor (8 concurrent):
  Sequential: 0.2039s
  Parallel: 0.2019s
  Speedup: 1.01x
  Platform: AMD64
  ⚠️  WARNING: Speedup 1.01x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
default loop - sinc_medium async with ThreadPoolExecutor (2 concurrent):
  Sequential: 0.0980s
  Parallel: 0.0984s
  Speedup: 1.00x
  Platform: AMD64
  ⚠️  WARNING: Speedup 1.00x is below expected 1.1x
      This may be due to CI load or platform-specific threading overhead.
.
default loop - sinc_medium async with ThreadPoolExecutor (4 concurrent):
  Sequential: 0.1956s
  Parallel: 0.1965s
  Speedup: 1.00x
default loop - sinc_medium async with ThreadPoolExecutor (8 concurrent):
  Sequential: 0.4028s
  Parallel: 0.3928s
  Speedup: 1.03x
  Platform: AMD64
  ⚠️  WARNING: Speedup 1.03x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
default loop - sinc_best async with ThreadPoolExecutor (2 concurrent):
  Sequential: 0.3047s
  Parallel: 0.3044s
  Speedup: 1.00x
  Platform: AMD64
  ⚠️  WARNING: Speedup 1.00x is below expected 1.1x
      This may be due to CI load or platform-specific threading overhead.
.
default loop - sinc_best async with ThreadPoolExecutor (4 concurrent):
  Sequential: 0.6116s
  Parallel: 0.6100s
  Speedup: 1.00x
  Platform: AMD64
  ⚠️  WARNING: Speedup 1.00x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
default loop - sinc_best async with ThreadPoolExecutor (8 concurrent):
  Sequential: 1.2147s
  Parallel: 1.2215s
  Speedup: 0.99x
  Platform: AMD64
  ⚠️  WARNING: Speedup 0.99x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
winloop loop - sinc_fastest async with ThreadPoolExecutor (2 concurrent):
  Sequential: 0.0500s
  Parallel: 0.0502s
  Speedup: 1.00x
  Platform: AMD64
  ⚠️  WARNING: Speedup 1.00x is below expected 1.1x
      This may be due to CI load or platform-specific threading overhead.
.
winloop loop - sinc_fastest async with ThreadPoolExecutor (4 concurrent):
  Sequential: 0.0994s
  Parallel: 0.0998s
  Speedup: 1.00x
  Platform: AMD64
  ⚠️  WARNING: Speedup 1.00x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
winloop loop - sinc_fastest async with ThreadPoolExecutor (8 concurrent):
  Sequential: 0.1986s
  Parallel: 0.2010s
  Speedup: 0.99x
  Platform: AMD64
  ⚠️  WARNING: Speedup 0.99x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
winloop loop - sinc_medium async with ThreadPoolExecutor (2 concurrent):
  Sequential: 0.0979s
  Parallel: 0.0984s
  Speedup: 0.99x
  Platform: AMD64
  ⚠️  WARNING: Speedup 0.99x is below expected 1.1x
      This may be due to CI load or platform-specific threading overhead.
.
winloop loop - sinc_medium async with ThreadPoolExecutor (4 concurrent):
  Sequential: 0.1959s
  Parallel: 0.1966s
  Speedup: 1.00x
  Platform: AMD64
  ⚠️  WARNING: Speedup 1.00x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
winloop loop - sinc_medium async with ThreadPoolExecutor (8 concurrent):
  Sequential: 0.3913s
  Parallel: 0.3931s
  Speedup: 1.00x
  Platform: AMD64
  ⚠️  WARNING: Speedup 1.00x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
winloop loop - sinc_best async with ThreadPoolExecutor (2 concurrent):
  Sequential: 0.3049s
  Parallel: 0.3045s
  Speedup: 1.00x
  Platform: AMD64
  ⚠️  WARNING: Speedup 1.00x is below expected 1.1x
      This may be due to CI load or platform-specific threading overhead.
.
winloop loop - sinc_best async with ThreadPoolExecutor (4 concurrent):
  Sequential: 0.6086s
  Parallel: 0.6119s
  Speedup: 0.99x
  Platform: AMD64
  ⚠️  WARNING: Speedup 0.99x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
winloop loop - sinc_best async with ThreadPoolExecutor (8 concurrent):
  Sequential: 1.2158s
  Parallel: 1.2171s
  Speedup: 1.00x
  Platform: AMD64
  ⚠️  WARNING: Speedup 1.00x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
default loop - sinc_fastest blocking vs executor:
  Without executor (blocks loop): 0.0100s
  With ThreadPoolExecutor: 0.0107s
  Improvement: 0.94x
  ⚠️  WARNING: Executor not significantly faster than blocking
      Expected executor < 0.0077s, got 0.0107s
      This may be due to CI load or platform-specific overhead.
.
winloop loop - sinc_fastest blocking vs executor:
  Without executor (blocks loop): 0.0099s
  With ThreadPoolExecutor: 0.0104s
  Improvement: 0.96x
  ⚠️  WARNING: Executor not significantly faster than blocking
      Expected executor < 0.0077s, got 0.0104s
      This may be due to CI load or platform-specific overhead.
.
default loop - 2 concurrent tasks - ThreadPool vs ProcessPool:
  ThreadPoolExecutor: 0.0204s
  ProcessPoolExecutor: 0.1688s
  Ratio: 8.29x
  → ThreadPool is faster
    (GIL release makes ThreadPool competitive with ProcessPool)
.
default loop - 4 concurrent tasks - ThreadPool vs ProcessPool:
  ThreadPoolExecutor: 0.0406s
  ProcessPoolExecutor: 0.1846s
  Ratio: 4.55x
  → ThreadPool is faster
    (GIL release makes ThreadPool competitive with ProcessPool)
.
winloop loop - 2 concurrent tasks - ThreadPool vs ProcessPool:
  ThreadPoolExecutor: 0.0202s
  ProcessPoolExecutor: 0.1658s
  Ratio: 8.21x
  → ThreadPool is faster
    (GIL release makes ThreadPool competitive with ProcessPool)
.
winloop loop - 4 concurrent tasks - ThreadPool vs ProcessPool:
  ThreadPoolExecutor: 0.0404s
  ProcessPoolExecutor: 0.1845s
  Ratio: 4.57x
  → ThreadPool is faster
    (GIL release makes ThreadPool competitive with ProcessPool)
.
default loop - Mixed I/O and CPU workload:
  Total time: 0.2063s
  Tasks completed: 5
  ✓ Performance meets expectations (< 0.35s)
.
winloop loop - Mixed I/O and CPU workload:
  Total time: 0.2075s
  Tasks completed: 5
  ✓ Performance meets expectations (< 0.35s)
.
======================================================================
Asyncio Performance Report
======================================================================

Test Configuration:
  Sample rate: 44100 Hz
  Duration: 5.0 seconds (220500 samples)
  Conversion ratio: 2.0x
  Executor: ThreadPoolExecutor

----------------------------------------------------------------------
Converter: sinc_fastest
----------------------------------------------------------------------
  1 concurrent task (baseline):
    Execution time: 0.0252s
  2 concurrent tasks:
    Parallel execution time: 0.0507s
    Equivalent sequential time: 0.0504s (2 × 0.0252s)
    Speedup: 0.99x
    Parallel efficiency: 49.6%
  4 concurrent tasks:
    Parallel execution time: 0.1002s
    Equivalent sequential time: 0.1007s (4 × 0.0252s)
    Speedup: 1.00x
    Parallel efficiency: 25.1%

----------------------------------------------------------------------
Converter: sinc_medium
----------------------------------------------------------------------
  1 concurrent task (baseline):
    Execution time: 0.0497s
  2 concurrent tasks:
    Parallel execution time: 0.0986s
    Equivalent sequential time: 0.0995s (2 × 0.0497s)
    Speedup: 1.01x
    Parallel efficiency: 50.4%
  4 concurrent tasks:
    Parallel execution time: 0.1967s
    Equivalent sequential time: 0.1990s (4 × 0.0497s)
    Speedup: 1.01x
    Parallel efficiency: 25.3%

----------------------------------------------------------------------
Converter: sinc_best
----------------------------------------------------------------------
  1 concurrent task (baseline):
    Execution time: 0.1532s
  2 concurrent tasks:
    Parallel execution time: 0.3157s
    Equivalent sequential time: 0.3064s (2 × 0.1532s)
    Speedup: 0.97x
    Parallel efficiency: 48.5%
  4 concurrent tasks:
    Parallel execution time: 0.6162s
    Equivalent sequential time: 0.6127s (4 × 0.1532s)
    Speedup: 0.99x
    Parallel efficiency: 24.9%
.
======================================================================
Asyncio Performance Report
======================================================================

Test Configuration:
  Sample rate: 44100 Hz
  Duration: 5.0 seconds (220500 samples)
  Conversion ratio: 2.0x
  Executor: ThreadPoolExecutor

----------------------------------------------------------------------
Converter: sinc_fastest
----------------------------------------------------------------------
  1 concurrent task (baseline):
    Execution time: 0.0251s
  2 concurrent tasks:
    Parallel execution time: 0.0505s
    Equivalent sequential time: 0.0502s (2 × 0.0251s)
    Speedup: 1.00x
    Parallel efficiency: 49.8%
  4 concurrent tasks:
    Parallel execution time: 0.1011s
    Equivalent sequential time: 0.1005s (4 × 0.0251s)
    Speedup: 0.99x
    Parallel efficiency: 24.8%

----------------------------------------------------------------------
Converter: sinc_medium
----------------------------------------------------------------------
  1 concurrent task (baseline):
    Execution time: 0.0496s
  2 concurrent tasks:
    Parallel execution time: 0.0986s
    Equivalent sequential time: 0.0993s (2 × 0.0496s)
    Speedup: 1.01x
    Parallel efficiency: 50.4%
  4 concurrent tasks:
    Parallel execution time: 0.1966s
    Equivalent sequential time: 0.1986s (4 × 0.0496s)
    Speedup: 1.01x
    Parallel efficiency: 25.3%

----------------------------------------------------------------------
Converter: sinc_best
----------------------------------------------------------------------
  1 concurrent task (baseline):
    Execution time: 0.1528s
  2 concurrent tasks:
    Parallel execution time: 0.3054s
    Equivalent sequential time: 0.3057s (2 × 0.1528s)
    Speedup: 1.00x
    Parallel efficiency: 50.0%
  4 concurrent tasks:
    Parallel execution time: 0.6147s
    Equivalent sequential time: 0.6113s (4 × 0.1528s)
    Speedup: 0.99x
    Parallel efficiency: 24.9%

threading

PS C:\Users\shaun\python-samplerate-ledfx> uv run pytest .\tests\test_threading_performance.py -s       
Uninstalled 1 package in 8ms
Installed 1 package in 10ms
================================================================ test session starts =================================================================
platform win32 -- Python 3.11.9, pytest-9.0.1, pluggy-1.6.0
rootdir: C:\Users\shaun\python-samplerate-ledfx
configfile: pyproject.toml
plugins: asyncio-1.3.0
asyncio: mode=Mode.STRICT, debug=False, asyncio_default_fixture_loop_scope=None, asyncio_default_test_loop_scope=function
collected 38 items                                                                                                                                    

tests\test_threading_performance.py
sinc_fastest with 2 threads:
  Sequential: 0.0499s
  Parallel: 0.0505s
  Speedup: 0.99x
  Platform: AMD64
  Individual thread times: ['0.0250s', '0.0248s']
  ⚠️  WARNING: Speedup 0.99x is below expected 1.2x
      Expected: 1.2x, Got: 0.99x
      (sequential=0.0499s, parallel=0.0505s)
      This may be due to CI load or platform-specific threading overhead.
.
sinc_fastest with 4 threads:
  Sequential: 0.1058s
  Parallel: 0.1010s
  Speedup: 1.05x
  Platform: AMD64
  Individual thread times: ['0.0251s', '0.0251s', '0.0250s', '0.0250s']
  ⚠️  WARNING: Speedup 1.05x is below expected 1.35x
      Expected: 1.35x, Got: 1.05x
      (sequential=0.1058s, parallel=0.1010s)
      This may be due to CI load or platform-specific threading overhead.
.
sinc_fastest with 6 threads:
  Sequential: 0.1491s
  Parallel: 0.1509s
  Speedup: 0.99x
  Platform: AMD64
  Individual thread times: ['0.0250s', '0.0249s', '0.0251s', '0.0250s', '0.0249s', '0.0249s']
  ⚠️  WARNING: Speedup 0.99x is below expected 1.35x
      Expected: 1.35x, Got: 0.99x
      (sequential=0.1491s, parallel=0.1509s)
      This may be due to CI load or platform-specific threading overhead.
.
sinc_fastest with 8 threads:
  Sequential: 0.1994s
  Parallel: 0.2010s
  Speedup: 0.99x
  Platform: AMD64
  Individual thread times: ['0.0249s', '0.0254s', '0.0249s', '0.0249s', '0.0250s', '0.0249s', '0.0248s', '0.0248s']
  ⚠️  WARNING: Speedup 0.99x is below expected 1.35x
      Expected: 1.35x, Got: 0.99x
      (sequential=0.1994s, parallel=0.2010s)
      This may be due to CI load or platform-specific threading overhead.
.
sinc_medium with 2 threads:
  Sequential: 0.0980s
  Parallel: 0.0982s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.0489s', '0.0489s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.2x
      Expected: 1.2x, Got: 1.00x
      (sequential=0.0980s, parallel=0.0982s)
      This may be due to CI load or platform-specific threading overhead.
.
sinc_medium with 4 threads:
  Sequential: 0.1961s
  Parallel: 0.1983s
  Speedup: 0.99x
  Platform: AMD64
  Individual thread times: ['0.0489s', '0.0490s', '0.0501s', '0.0496s']
  ⚠️  WARNING: Speedup 0.99x is below expected 1.35x
      Expected: 1.35x, Got: 0.99x
      (sequential=0.1961s, parallel=0.1983s)
      This may be due to CI load or platform-specific threading overhead.
.
sinc_medium with 6 threads:
  Sequential: 0.2936s
  Parallel: 0.2966s
  Speedup: 0.99x
  Platform: AMD64
  Individual thread times: ['0.0489s', '0.0491s', '0.0490s', '0.0506s', '0.0490s', '0.0489s']
  ⚠️  WARNING: Speedup 0.99x is below expected 1.35x
      Expected: 1.35x, Got: 0.99x
      (sequential=0.2936s, parallel=0.2966s)
      This may be due to CI load or platform-specific threading overhead.
.
sinc_medium with 8 threads:
  Sequential: 0.3924s
  Parallel: 0.3926s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.0492s', '0.0492s', '0.0489s', '0.0489s', '0.0488s', '0.0488s', '0.0488s', '0.0488s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.35x
      Expected: 1.35x, Got: 1.00x
      (sequential=0.3924s, parallel=0.3926s)
      This may be due to CI load or platform-specific threading overhead.
.
sinc_best with 2 threads:
  Sequential: 0.3042s
  Parallel: 0.3048s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.1523s', '0.1521s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.2x
      Expected: 1.2x, Got: 1.00x
      (sequential=0.3042s, parallel=0.3048s)
      This may be due to CI load or platform-specific threading overhead.
.
sinc_best with 4 threads:
  Sequential: 0.6091s
  Parallel: 0.6084s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.1518s', '0.1517s', '0.1521s', '0.1521s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.35x
      Expected: 1.35x, Got: 1.00x
      (sequential=0.6091s, parallel=0.6084s)
      This may be due to CI load or platform-specific threading overhead.
.
sinc_best with 6 threads:
  Sequential: 0.9147s
  Parallel: 0.9118s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.1521s', '0.1519s', '0.1520s', '0.1517s', '0.1516s', '0.1515s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.35x
      Expected: 1.35x, Got: 1.00x
      (sequential=0.9147s, parallel=0.9118s)
      This may be due to CI load or platform-specific threading overhead.
.
sinc_best with 8 threads:
  Sequential: 1.2161s
  Parallel: 1.2192s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.1520s', '0.1536s', '0.1518s', '0.1518s', '0.1517s', '0.1527s', '0.1522s', '0.1522s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.35x
      Expected: 1.35x, Got: 1.00x
      (sequential=1.2161s, parallel=1.2192s)
      This may be due to CI load or platform-specific threading overhead.
.
sinc_fastest Resampler.process() with 2 threads:
  Sequential: 0.0501s
  Parallel: 0.0519s
  Speedup: 0.97x
  Platform: AMD64
  Individual thread times: ['0.0265s', '0.0248s']
  ⚠️  WARNING: Speedup 0.97x is below expected 1.1x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_fastest Resampler.process() with 4 threads:
  Sequential: 0.1036s
  Parallel: 0.1010s
  Speedup: 1.03x
  Platform: AMD64
  Individual thread times: ['0.0249s', '0.0251s', '0.0250s', '0.0249s']
  ⚠️  WARNING: Speedup 1.03x is below expected 1.25x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_fastest Resampler.process() with 6 threads:
  Sequential: 0.1497s
  Parallel: 0.1516s
  Speedup: 0.99x
  Platform: AMD64
  Individual thread times: ['0.0250s', '0.0255s', '0.0250s', '0.0250s', '0.0249s', '0.0250s']
  ⚠️  WARNING: Speedup 0.99x is below expected 1.25x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_fastest Resampler.process() with 8 threads:
  Sequential: 0.2064s
  Parallel: 0.2013s
  Speedup: 1.03x
  Platform: AMD64
  Individual thread times: ['0.0250s', '0.0251s', '0.0250s', '0.0249s', '0.0250s', '0.0250s', '0.0250s', '0.0249s']
  ⚠️  WARNING: Speedup 1.03x is below expected 1.25x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_medium Resampler.process() with 2 threads:
  Sequential: 0.0980s
  Parallel: 0.0984s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.0490s', '0.0488s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.1x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_medium Resampler.process() with 4 threads:
  Sequential: 0.1958s
  Parallel: 0.1961s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.0488s', '0.0489s', '0.0489s', '0.0488s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.25x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_medium Resampler.process() with 6 threads:
  Sequential: 0.2936s
  Parallel: 0.2942s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.0489s', '0.0489s', '0.0488s', '0.0490s', '0.0488s', '0.0488s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.25x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_medium Resampler.process() with 8 threads:
  Sequential: 0.3911s
  Parallel: 0.3932s
  Speedup: 0.99x
  Platform: AMD64
  Individual thread times: ['0.0491s', '0.0490s', '0.0490s', '0.0490s', '0.0490s', '0.0489s', '0.0490s', '0.0489s']
  ⚠️  WARNING: Speedup 0.99x is below expected 1.25x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_best Resampler.process() with 2 threads:
  Sequential: 0.3034s
  Parallel: 0.3042s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.1517s', '0.1519s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.1x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_best Resampler.process() with 4 threads:
  Sequential: 0.6090s
  Parallel: 0.6097s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.1518s', '0.1523s', '0.1527s', '0.1520s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.25x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_best Resampler.process() with 6 threads:
  Sequential: 0.9141s
  Parallel: 0.9120s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.1523s', '0.1519s', '0.1515s', '0.1519s', '0.1518s', '0.1516s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.25x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_best Resampler.process() with 8 threads:
  Sequential: 1.2158s
  Parallel: 1.2170s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.1516s', '0.1521s', '0.1516s', '0.1516s', '0.1517s', '0.1521s', '0.1524s', '0.1524s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.25x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_fastest CallbackResampler with 2 threads:
  Sequential: 0.0503s
  Parallel: 0.0517s
  Speedup: 0.97x
  Platform: AMD64
  Individual thread times: ['0.0509s', '0.0255s']
  ⚠️  WARNING: Speedup 0.97x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_fastest CallbackResampler with 4 threads:
  Sequential: 0.1006s
  Parallel: 0.1013s
  Speedup: 0.99x
  Platform: AMD64
  Individual thread times: ['0.0499s', '0.0500s', '0.0504s', '0.0256s']
  ⚠️  WARNING: Speedup 0.99x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_fastest CallbackResampler with 6 threads:
  Sequential: 0.1515s
  Parallel: 0.1519s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.0501s', '0.0504s', '0.0502s', '0.0503s', '0.0501s', '0.0255s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_fastest CallbackResampler with 8 threads:
  Sequential: 0.2028s
  Parallel: 0.2021s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.0501s', '0.0503s', '0.0503s', '0.0501s', '0.0501s', '0.0502s', '0.0501s', '0.0254s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_medium CallbackResampler with 2 threads:
  Sequential: 0.0985s
  Parallel: 0.0989s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.0981s', '0.0496s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_medium CallbackResampler with 4 threads:
  Sequential: 0.1971s
  Parallel: 0.1971s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.0979s', '0.0981s', '0.0981s', '0.0495s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_medium CallbackResampler with 6 threads:
  Sequential: 0.2951s
  Parallel: 0.2960s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.0980s', '0.0984s', '0.0983s', '0.0983s', '0.0983s', '0.0495s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_medium CallbackResampler with 8 threads:
  Sequential: 0.3936s
  Parallel: 0.3941s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.0983s', '0.0981s', '0.0981s', '0.0982s', '0.0981s', '0.0981s', '0.0981s', '0.0494s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_best CallbackResampler with 2 threads:
  Sequential: 0.3052s
  Parallel: 0.3046s
  Speedup: 1.00x
  Platform: AMD64
  Individual thread times: ['0.3036s', '0.1523s']
  ⚠️  WARNING: Speedup 1.00x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_best CallbackResampler with 4 threads:
  Sequential: 0.6113s
  Parallel: 0.6576s
  Speedup: 0.93x
  Platform: AMD64
  Individual thread times: ['0.3029s', '0.3105s', '0.3532s', '0.1948s']
  ⚠️  WARNING: Speedup 0.93x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_best CallbackResampler with 6 threads:
  Sequential: 1.0773s
  Parallel: 1.0299s
  Speedup: 1.05x
  Platform: AMD64
  Individual thread times: ['0.3172s', '0.3112s', '0.3624s', '0.4016s', '0.3482s', '0.1543s']
  ⚠️  WARNING: Speedup 1.05x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
.
sinc_best CallbackResampler with 8 threads:
  Sequential: 1.4173s
  Parallel: 1.2259s
  Speedup: 1.16x
  Platform: AMD64
  Individual thread times: ['0.3069s', '0.3045s', '0.3059s', '0.3069s', '0.3054s', '0.3051s', '0.3054s', '0.1532s']
  ⚠️  WARNING: Speedup 1.16x is below expected 1.2x
      This may be due to CI load or platform-specific threading overhead.
..
======================================================================
GIL Release Performance Report
======================================================================

Test Configuration:
  Sample rate: 44100 Hz
  Duration: 5.0 seconds (220500 samples)
  Conversion ratio: 2.0x

----------------------------------------------------------------------
Converter: sinc_fastest
----------------------------------------------------------------------
  1 thread (baseline):
    Execution time: 0.0250s
  2 threads (parallel):
    Parallel execution time: 0.0503s
    Equivalent sequential time: 0.0499s (2 × 0.0250s)
    Speedup: 0.99x
    Parallel efficiency: 49.7%
    Avg thread time: 0.0249s
  4 threads (parallel):
    Parallel execution time: 0.1006s
    Equivalent sequential time: 0.0998s (4 × 0.0250s)
    Speedup: 0.99x
    Parallel efficiency: 24.8%
    Avg thread time: 0.0250s

----------------------------------------------------------------------
Converter: sinc_medium
----------------------------------------------------------------------
  1 thread (baseline):
    Execution time: 0.0492s
  2 threads (parallel):
    Parallel execution time: 0.0985s
    Equivalent sequential time: 0.0984s (2 × 0.0492s)
    Speedup: 1.00x
    Parallel efficiency: 49.9%
    Avg thread time: 0.0491s
  4 threads (parallel):
    Parallel execution time: 0.1976s
    Equivalent sequential time: 0.1968s (4 × 0.0492s)
    Speedup: 1.00x
    Parallel efficiency: 24.9%
    Avg thread time: 0.0493s

----------------------------------------------------------------------
Converter: sinc_best
----------------------------------------------------------------------
  1 thread (baseline):
    Execution time: 0.1518s
  2 threads (parallel):
    Parallel execution time: 0.3054s
    Equivalent sequential time: 0.3036s (2 × 0.1518s)
    Speedup: 0.99x
    Parallel efficiency: 49.7%
    Avg thread time: 0.1525s
  4 threads (parallel):
    Parallel execution time: 0.6081s
    Equivalent sequential time: 0.6073s (4 × 0.1518s)
    Speedup: 1.00x
    Parallel efficiency: 25.0%
    Avg thread time: 0.1519s

shauneccles added 2 commits November 19, 2025 16:02

Add asyncio and threading performance tests for resampling operations

46c1fd4

Filter ARM Mac test and change to warnings

40366fb

fakufaku self-assigned this Nov 19, 2025

Add pytest-asyncio dep

0973472

shauneccles mentioned this pull request Nov 19, 2025

Enhance GIL management and error handling in resampling #31

Open

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Add asyncio and threading performance tests for resampling operations#32

Add asyncio and threading performance tests for resampling operations#32
shauneccles wants to merge 3 commits intotuxu:masterfrom
LedFx:tests/add-threading-asyncio-tests

shauneccles commented Nov 19, 2025

Uh oh!

shauneccles commented Nov 19, 2025 •

edited

Loading

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

Conversation

shauneccles commented Nov 19, 2025

Uh oh!

shauneccles commented Nov 19, 2025 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

Reviewers

Assignees

Labels

Projects

Milestone

Development

Uh oh!

2 participants

shauneccles commented Nov 19, 2025 •

edited

Loading