With regard different versions of Windows: Windows 95 and above: Can use CryptGenRandom Windows Vista and above: Can use BCryptGenRandom ( as posted ) As we progress from Windows 95, Microsoft stepped up and used the latest recommendation from the NIST with regard what the random number generator is...

Thanks, St_W. I am not up to speed with all the commands yet. :( CryptoRnd.bas is only a slight change to CryptoRndBufferCNG.inc but here it is. CryptoRnd.bas #include once "windows.bi" #include once "win/bcrypt.bi" #inclib "bcrypt" Dim Shared As Byte Ptr hRand Dim Shar...

Finally hit 1TB with a clean sweep except for the two unusual values during the 256MB run. I am glad about them - I think that I would find it a little disconcerting to get a clean sweep. The test went from 256MB, 512MB, ..., 1TB giving a total crunch of 2.38TB. I started another test and it wasn't ...

My_RNG was written in 'PowerBASC mode'. PowerBASIC has a command where an array and a string can share the same memory; without which I resorted to copying. The FreeBASIC docs make it clear that we should use StrPtr to point to a string's character data with VarPtr returning the address of the inter...

I have been using PractRand for a few years now to examine the output of pseudo-random number generators. Unlike most compilers FreeBASIC offers a selection of generators so I subjected them to PractRand. PractRand is a very unforgiving suite of tests. Way back in 1995 George Marsaglia published Die...

I am having problems piping to PractRand ... I was using MKI without realising that what I actually needed was a MKUI, unsigned integer, and that doesn't exist. <laugh> So, what I am now doing is to populate a Ulong array with CryptoDW, copy the contents to a string and then stream the strings via ...

Tested to 64GB - only one anomaly at 512MB. I haven't seen that many tests do that well. If any random number generator never finds itself in the tail of a bell curve then we would have a contradiction. 128GB would take a further 140 minutes and doubling each time. A TB would take a further 35 hours...

I am having problems piping to PractRand so I dumped 256MB of CryptoDW ( 1GB ) to disk. 32MB, 64MB, 128MB, 256MB, 512MB and 1GB were tested. Four of them had no anomalies and two had one unusual value. Nothing wrong here - I have seen worse with Intel RdRand and quantum random number dumps. While I ...

Here is a typical run of the above:

I now have a library: http://deltarho.org.uk/Downloads/CryptoRnd.zip The download is a zipped folder which unzips to CryptoRnd where you will find CryptoRnd.bi and libCryptoRnd.a. The library is only 4612 bytes. I thought that using a library would slow things down a bit, but it hasn't. For the time...

multiply by a floating point constant Did some tests and averaged. 125MB of CryptoS took 493.309ms with the constant multiplier approach and 467.097ms with the code as published. I thought your constant multiplier may have an edge but I was wrong, although we only have a difference of 5%. Credit to...

or 2.328306436538696289063e-10 ( from 1/2^32 ) for 32 bit [0,1), but where is the fun in that? Anyway the main use of pseudo-random number generators in most applications is correlation breaking, which fortunately isn't too difficult. I was wondering what that has to do with the price of bacon but t...

We are spoilt for choice with regard random number generators. Algorithm 5 provides cryptographic random numbers. For Windows, the CryptoAPI is used via CryptGenRandom. A better quality generator is with 'Cryptography API: Next Generation' introduced in Windows Vista via BCryptGenRandom. Both genera...

A request for n-1 milliseconds would see a delay of, roughly, n-1 to n milliseconds. Occasionally, we overshoot n milliseconds so there is no need to poll QueryPerformanceCounter. We can test if an overshoot has occurred but this would incur a greater overshoot. A solution is to use Sleep( n-2,1 )....

Here is another macro to add to the above although it doesn't time an interval but creates one. A simple analogy of the method employed is to consider amplification in a Hi-Fi system. We could have a powerful Class B amplifier which doesn't sound too good but didn't cost much. On the other hand, we ...