vs

 1 : Are the Numba programs faster? At a glance.

Each chart bar shows, for one unidentified benchmark, how much the fastest Numba program used compared to the fastest Cython program.

(Memory use is only compared for tasks that require memory to be allocated.)


These are not the only compilers and interpreters. These are not the only programs that could be written. These are not the only tasks that could be solved. These are just 10 tiny examples.

 2 : Are the Numba programs faster? Approximately.

Each table row shows, for one named benchmark, how much the fastest Numba program used compared to the fastest Cython program.

(Memory use is only compared for tasks that require memory to be allocated.)

 Numba used what fraction? used how many times more? 
Benchmark Time Memory Code
 binary-trees±±
 fibonacci † ±
 spectral-norm ±
 fib50 † ±
 templates †20× ±
 iobench1,261×?±
 Numba used what fraction? used how many times more? 
Time-used  |-  |---  25% median  75%  ---|  -|
(Elapsed secs)20×41×1,261×

† possible mismatch - one-core program compared to multi-core program.

± read the measurements and then read the program source code.

 3 : Are the Numba programs faster? Measurements.

These are not the only tasks that could be solved. These are just 10 tiny examples. These are not the only compilers and interpreters. These are not the only programs that could be written.

For each named benchmark, measurements of the fastest Numba program are shown for comparison against measurements of the fastest Cython program.

Program Source Code CPU secs Elapsed secs Memory KB Code B ≈ CPU Load
 binary-trees 
Numba7.112.5866,496702  96% 95% 87% 93%
Cython3.050.8767,172706  85% 86% 87% 94%
 fibonacci 
Numba2.714.5075,816160  100% 100% 100% 100% †
Cython0.790.799,400184  1% 99% 0% 1% †
 spectral-norm 
Numba2.4815.4594,924663  100% 100% 100% 100%
Cython7.792.1755,920575  90% 89% 90% 92%
 fib50 
Numba4.377.7975,964152  100% 100% 100% 100% †
Cython1.091.098,816156  100% 2% 4% 0% †
 templates 
Numba4.5528.0180,772335  100% 100% 100% 100% †
Cython1.401.4016,644328  100% 0% 4% 1% †
 iobench 
Numba14.8441.6076,508384  100% 100% 100% 100%
Cython0.030.03?366  0% 0% 100% 25%
 n-body 
No program
Cython25.5425.568,7121315  100% 1% 0% 0%
 fasta-redux 
No program
Cython0.330.331,8761115  0% 0% 100% 0%
 reverse-complement 
No program
Cython0.030.03?458  100% 25% 0% 0%
 pystone 
No program
Cython0.220.228762288  0% 5% 0% 100%
 fasta 
No program
Cython8.538.548,976904  5% 100% 3% 3%
 meteor-contest 
No program
Cython4.504.5011,3201334  2% 0% 0% 100%
 fannkuch-redux 
No program
Cython21.845.6856,0801271  98% 96% 97% 97%
 pidigits 
No program
Cython0.040.10?379  0% 90% 0% 0%
 chameneos-redux 
No program
Cython160.1988.278,8361203  40% 39% 39% 39%
 regex-dna 
No program
Cython0.060.06?524  0% 100% 0% 0%
 jsonbench 
No program
Cython4.904.9011,996323  2% 0% 100% 1%
 richards 
No program
Cython1.001.008,8282467  1% 1% 100% 0%
 k-nucleotide 
No program
Cython0.120.12?594  8% 0% 100% 0%
 mandelbrot 
No program
Cython101.2126.53189,8043832  99% 100% 96% 97%
 thread-ring 
No program
Cython0.360.361,816407  0% 100% 0% 3%
 binary-trees-redux
   No programs

† possible mismatch - one-core program compared to multi-core program.

 4 : Are there other Numba programs for these benchmarks?

Remember - those are just the fastest Numba and Cython programs measured on this OS/machine. Check if there are other implementations of these benchmark programs for Numba.

Maybe one of those other Numba programs is fastest on a different OS/machine.

 5 : Are there other faster programs for these benchmarks?

Remember - those are just the fastest Numba and Cython programs measured on this OS/machine. Check if there are faster implementations of these benchmark programs for other programming languages.

Maybe one of those other programs is fastest on a different OS/machine.

 Numba : Numba JIT compiler 

 

Revised BSD license

  Home   Conclusions   License   Play