Performance test MVar vs TMVar vs IORef. Just read.

MVar vs TMVar vs IORef

 ~/develop/bench -$ cat Main.hs                                                   iMac-Alexej@chemist :)
import Control.Concurrent
import Control.Concurrent.STM
import Criterion.Main
import Data.IORef


main = do
    mv <- newMVar ()
    tv <- newTMVarIO ()
    ir <- newIORef ()
    defaultMain [ bench "MVar" (readMVar mv)
                , bench "TMVar" (atomically $ readTMVar tv)
                , bench "IORef" (readIORef ir)
                ]


 ~/develop/bench -$ ghc -O2 -threaded Main.hs -o mv-tm-ir                         iMac-Alexej@chemist :)


 ~/develop/bench -$ ./mv-tm-ir                                                    iMac-Alexej@chemist :)
warming up
estimating clock resolution...
mean is 1.314451 us (640001 iterations)
found 3421 outliers among 639999 samples (0.5%)
  2748 (0.4%) high severe
estimating cost of a clock call...
mean is 49.09204 ns (9 iterations)
found 2 outliers among 9 samples (22.2%)
  1 (11.1%) high mild
  1 (11.1%) high severe

benchmarking MVar
mean: 40.45142 ns, lb 40.01889 ns, ub 40.83322 ns, ci 0.950
std dev: 2.081636 ns, lb 1.655347 ns, ub 3.136299 ns, ci 0.950
found 11 outliers among 100 samples (11.0%)
  1 (1.0%) low severe
  10 (10.0%) high mild
variance introduced by outliers: 49.465%
variance is moderately inflated by outliers

benchmarking TMVar
mean: 59.49208 ns, lb 59.13058 ns, ub 59.94610 ns, ci 0.950
std dev: 2.065702 ns, lb 1.718641 ns, ub 2.541805 ns, ci 0.950
found 5 outliers among 100 samples (5.0%)
  5 (5.0%) high mild
variance introduced by outliers: 30.673%
variance is moderately inflated by outliers

benchmarking IORef
mean: 4.987461 ns, lb 4.976263 ns, ub 5.002600 ns, ci 0.950
std dev: 66.13710 ps, lb 52.75414 ps, ub 84.57218 ps, ci 0.950
found 20 outliers among 100 samples (20.0%)
  13 (13.0%) high mild
  7 (7.0%) high severe
variance introduced by outliers: 6.565%
variance is slightly inflated by outliers
 ~/develop/bench -$