Native RNGs do not pass SmallCrush #803
Description
Both GPUArrays' and CUDA's native RNGs do not support RNGTest's SmallCrush:
julia> rng = RNGTest.wrap(CUDA.gpuarrays_rng(), UInt32)'
julia> RNGTest.smallcrushTestU01(rng)
========= Summary results of SmallCrush =========
Version: TestU01 1.2.3
Generator:
Number of statistics: 15
Total CPU time: 00:00:05.25
The following tests gave p-values outside [0.001, 0.9990]:
(eps means a value < 1.0e-300):
(eps1 means a value < 1.0e-15):
Test p-value
----------------------------------------------
1 BirthdaySpacings eps
2 Collision eps
3 Gap eps
4 SimpPoker 7.4e-4
5 CouponCollector 4.9e-4
6 MaxOft eps
7 WeightDistrib 1.4e-5
9 HammingIndep 1.3e-4
----------------------------------------------
All other tests were passed
julia> rng = RNGTest.wrap(CUDA.RNG(), UInt32);
julia> RNGTest.smallcrushTestU01(rng)
========= Summary results of SmallCrush =========
Version: TestU01 1.2.3
Generator:
Number of statistics: 15
Total CPU time: 00:00:05.21
The following tests gave p-values outside [0.001, 0.9990]:
(eps means a value < 1.0e-300):
(eps1 means a value < 1.0e-15):
Test p-value
----------------------------------------------
1 BirthdaySpacings eps
2 Collision eps
3 Gap eps
4 SimpPoker eps
5 CouponCollector eps
6 MaxOft AD 1 - eps1
7 WeightDistrib eps
8 MatrixRank eps
9 HammingIndep eps
10 RandomWalk1 H 5.0e-5
----------------------------------------------
All other tests were passed
CURAND does, of course:
julia> rng = RNGTest.wrap(CUDA.curand_rng(), UInt32);
julia> RNGTest.smallcrushTestU01(rng)
========= Summary results of SmallCrush =========
Version: TestU01 1.2.3
Generator:
Number of statistics: 15
Total CPU time: 00:00:05.53
All tests were passed
I'm out of my depth here, so help is appreciated.