That may mean that you could use this for normal simulations, but secure applications is a different story though: the issue of entropy remains. It is akin to the issue of having state that can be determined without exhausting the pseudorandom sequence for a PRNG, which is why even PRNGs that pass statistical tests with flying colours do not necessarily qualify as cryptographically secure PRNGs, although they certainly could see other uses.
Look up a C++ Reference and learn How To Ask Questions The Smart WayOriginally Posted by Bjarne Stroustrup (2000-10-14)
It can still be used in security software but for the simple reason of security it shouldn't ne used on its own so for example one could include multple RNGs and add their results together making it much harder to predict the final number used, to be honest I don't think even the online ones should be used on their own, its better to mix everything and increase the security of the final result
Is that command doing what you think it does? That command redirects the executable file itself to the standard input of the "dieharder" program. It doesn't execute ./mcc_rnd_die.elf.
If you know better then please, I'm all eyes
christop is right though. My guess is that's supposed to be:
But I haven't read the relevant documentation so I could be wrong.Code:./mcc_rnd_die.elf | dieharder -a
Look up a C++ Reference and learn How To Ask Questions The Smart Way
From dieharder(1) - Linux man page
Tim S.Code:cat /dev/urandom | ./dieharder -a -g 200
"...a computer is a stupid machine with the ability to do incredibly smart things, while computer programmers are smart people with the ability to do incredibly stupid things. They are,in short, a perfect match.." Bill Bryson
A simple test for the code at post #7... Take a look at:
This will create a graphic file where each pixel is scaled to 24 bits and used as RGB value in a 1024x1024 surface. It will result in a "noisy" surface like this:Code:#include <stdio.h> #include <stdlib.h> #include <time.h> // Scale 'n' to 24 bits (0xBBGGRR). #define rand2rgb(n) \ ( ( int ) ( ( (n) * 16777215.0 ) / RAND_MAX ) ) int main ( void ) { int x, y, r; srand( time(NULL) ); fputs( "P6\n1024 1024\n255\n", stdout ); for ( y = 0; y < 1024; y++ ) for ( x = 0; x < 1024; x++ ) { r = rand2rgb(rand()); fwrite( &r, 3, 1, stdout ); // assume little endian. } }
Now, take a look at your routine in #7, mildly modified (my comments):
This will create a graphic like this:Code:#include <stdio.h> #include <stdlib.h> #include <time.h> #include <limits.h> // Scale n to 24 bit value (0xBBGGRR). // OBS: Uses long double because 'long' is 64 bits long in x86-64 systems. #define rand2rgb(n) \ ( ( int ) ( ( (n) * 16777215.0L ) / LONG_MAX ) ) // using size_t instead of ptrdiff_t (same thing) typedef size_t mcc_rnd_t; static long mcc_rnd ( mcc_rnd_t *seed, long min, long max ); int main ( void ) { int x, y; int r; long seed = 1; fputs ( "P3 1024 1024 255\n", stdout ); for ( y = 0; y < 1024; y++ ) for ( x = 0; x < 1024; x++ ) { r = rand2rgb ( mcc_rnd ( &seed, 0, LONG_MAX ) ); fwrite ( &r, 3, 1, stdout ); } } long mcc_rnd ( mcc_rnd_t *seed, long min, long max ) { // Most of the time tmp will point to the SAME memory block // for every process. Notice that, probably, tmp is aligned, // meaning the lower bits aren't 'random' at all... // and the upper bits isn't random as well... void *tmp = malloc ( 1 ); long val = time ( NULL ) + ( size_t ) tmp; free ( tmp ); // This isn't 'random' seed. Taking the argument address means // it will be copied to the stack and the stack pointer isn't // that 'random'... The value on stack isn't that unpredictable // as you think! if ( ! seed ) seed = ( mcc_rnd_t * ) ( &seed ); // Well... this isn't random, again, isn't it? if ( ! *seed ) *seed = 1; val %= *seed; val *= clock(); // The seed, again, isn't random! *seed <<= 1; // Saturation get even less random values! return ( val > max ) ? max : ( val < min ? min : val ); }
See the patterns? Not random at all!
Have you tried the updated code instead of old code?A simple test for the code at post #7... Take a look at:
This will create a graphic file where each pixel is scaled to 24 bits and used as RGB value in a 1024x1024 surface. It will result in a "noisy" surface like this:Code:#include <stdio.h> #include <stdlib.h> #include <time.h> // Scale 'n' to 24 bits (0xBBGGRR). #define rand2rgb(n) \ ( ( int ) ( ( (n) * 16777215.0 ) / RAND_MAX ) ) int main ( void ) { int x, y, r; srand( time(NULL) ); fputs( "P6\n1024 1024\n255\n", stdout ); for ( y = 0; y < 1024; y++ ) for ( x = 0; x < 1024; x++ ) { r = rand2rgb(rand()); fwrite( &r, 3, 1, stdout ); // assume little endian. } }
Now, take a look at your routine in #7, mildly modified (my comments):
This will create a graphic like this:Code:#include <stdio.h> #include <stdlib.h> #include <time.h> #include <limits.h> // Scale n to 24 bit value (0xBBGGRR). // OBS: Uses long double because 'long' is 64 bits long in x86-64 systems. #define rand2rgb(n) \ ( ( int ) ( ( (n) * 16777215.0L ) / LONG_MAX ) ) // using size_t instead of ptrdiff_t (same thing) typedef size_t mcc_rnd_t; static long mcc_rnd ( mcc_rnd_t *seed, long min, long max ); int main ( void ) { int x, y; int r; long seed = 1; fputs ( "P3 1024 1024 255\n", stdout ); for ( y = 0; y < 1024; y++ ) for ( x = 0; x < 1024; x++ ) { r = rand2rgb ( mcc_rnd ( &seed, 0, LONG_MAX ) ); fwrite ( &r, 3, 1, stdout ); } } long mcc_rnd ( mcc_rnd_t *seed, long min, long max ) { // Most of the time tmp will point to the SAME memory block // for every process. Notice that, probably, tmp is aligned, // meaning the lower bits aren't 'random' at all... // and the upper bits isn't random as well... void *tmp = malloc ( 1 ); long val = time ( NULL ) + ( size_t ) tmp; free ( tmp ); // This isn't 'random' seed. Taking the argument address means // it will be copied to the stack and the stack pointer isn't // that 'random'... The value on stack isn't that unpredictable // as you think! if ( ! seed ) seed = ( mcc_rnd_t * ) ( &seed ); // Well... this isn't random, again, isn't it? if ( ! *seed ) *seed = 1; val %= *seed; val *= clock(); // The seed, again, isn't random! *seed <<= 1; // Saturation get even less random values! return ( val > max ) ? max : ( val < min ? min : val ); }
See the patterns? Not random at all!
Been reading the docs and seems to be fine with stdin but to satisfy your curiosity I'll try it when I get home
This one?
it is worse:Code:#include <stdio.h> #include <stdlib.h> #include <time.h> #include <limits.h> // Scale n to 24 bit value (0xBBGGRR). // OBS: Uses long double because 'long' is 64 bits long in x86-64 systems. #define rand2rgb(n) \ ( ( int ) ( ( (n) * 16777215.0L ) / ULONG_MAX ) ) // using size_t instead of ptrdiff_t (same thing) typedef size_t mcc_rnd_t; static unsigned long mcc__rnd( mcc_rnd_t *seed ); int main( void ) { int x, y; int r; long seed = 1; fputs( "P6\n1024 1024\n255\n", stdout ); for ( y = 0; y < 1024; y++ ) for ( x = 0; x < 1024; x++ ) { r = rand2rgb( mcc__rnd( &seed ) ); fwrite( &r, 3, 1, stdout ); } } unsigned long mcc__rnd( mcc_rnd_t *seed ) { /* Initial random value */ unsigned long val = ( unsigned long )( &seed ) * ( unsigned long )clock() * ( unsigned long )clock(); if ( seed ) { if ( *seed == 0 ) *seed = 1; val %= *seed; *seed <<= 1; } return val; }
Try without the seed set, ie NULL, also how did you get your terminal to do that? I've tried but it doesn't do thatThis one?
it is worse:Code:#include <stdio.h> #include <stdlib.h> #include <time.h> #include <limits.h> // Scale n to 24 bit value (0xBBGGRR). // OBS: Uses long double because 'long' is 64 bits long in x86-64 systems. #define rand2rgb(n) \ ( ( int ) ( ( (n) * 16777215.0L ) / ULONG_MAX ) ) // using size_t instead of ptrdiff_t (same thing) typedef size_t mcc_rnd_t; static unsigned long mcc__rnd( mcc_rnd_t *seed ); int main( void ) { int x, y; int r; long seed = 1; fputs( "P6\n1024 1024\n255\n", stdout ); for ( y = 0; y < 1024; y++ ) for ( x = 0; x < 1024; x++ ) { r = rand2rgb( mcc__rnd( &seed ) ); fwrite( &r, 3, 1, stdout ); } } unsigned long mcc__rnd( mcc_rnd_t *seed ) { /* Initial random value */ unsigned long val = ( unsigned long )( &seed ) * ( unsigned long )clock() * ( unsigned long )clock(); if ( seed ) { if ( *seed == 0 ) *seed = 1; val %= *seed; *seed <<= 1; } return val; }
Well currently re-running dieharderwith this so far:
I'm cancelling at this point and trying something in additionCode:make --no-print-directory dieharder_rnd gcc -ggdb -Wall -lpthread -DBUILD_FOR_DIEHARDER -o ./mcc_rnd_die.elf mcc_rnd.c ./mcc_rnd_die.elf > diehard.txt dieharder -a -f diehard.txt -t 100 -Y 1 #=============================================================================# # dieharder version 3.31.1 Copyright 2003 Robert G. Brown # #=============================================================================# rng_name | filename |rands/second| mt19937| diehard.txt| 8.32e+07 | #=============================================================================# test_name |ntup| tsamples |psamples| p-value |Assessment #=============================================================================# diehard_birthdays| 0| 100| 100|0.85295815| PASSED diehard_operm5| 0| 1000000| 100|0.95085512| PASSED diehard_rank_32x32| 0| 40000| 100|0.84994506| PASSED diehard_rank_6x8| 0| 100000| 100|0.88766341| PASSED diehard_bitstream| 0| 2097152| 100|0.56787971| PASSED diehard_opso| 0| 2097152| 100|0.30634647| PASSED diehard_oqso| 0| 2097152| 100|0.97180228| PASSED diehard_dna| 0| 2097152| 100|0.24078918| PASSED diehard_count_1s_str| 0| 256000| 100|0.30519436| PASSED diehard_count_1s_byt| 0| 256000| 100|0.51351868| PASSED diehard_parking_lot| 0| 12000| 100|0.89432206| PASSED diehard_2dsphere| 2| 8000| 100|0.99812802| WEAK diehard_2dsphere| 2| 8000| 200|0.96469714| PASSED diehard_3dsphere| 3| 4000| 100|0.96650039| PASSED diehard_squeeze| 0| 100000| 100|0.77568347| PASSED diehard_sums| 0| 100| 100|0.00051839| WEAK diehard_sums| 0| 100| 200|0.00400734| WEAK diehard_sums| 0| 100| 300|0.00090761| WEAK diehard_sums| 0| 100| 400|0.00006661| WEAK diehard_sums| 0| 100| 500|0.00002646| WEAK diehard_sums| 0| 100| 600|0.00000203| WEAK diehard_sums| 0| 100| 700|0.00000045| FAILED diehard_runs| 0| 100000| 100|0.99080465| PASSED diehard_runs| 0| 100000| 100|0.86282952| PASSED diehard_craps| 0| 200000| 100|0.80909198| PASSED diehard_craps| 0| 200000| 100|0.25693590| PASSED marsaglia_tsang_gcd| 0| 10000000| 100|0.18319920| PASSED marsaglia_tsang_gcd| 0| 10000000| 100|0.77544038| PASSED sts_monobit| 1| 100000| 100|0.27733181| PASSED sts_runs| 2| 100000| 100|0.81845021| PASSED sts_serial| 1| 100000| 100|0.46403118| PASSED sts_serial| 2| 100000| 100|0.11862501| PASSED sts_serial| 3| 100000| 100|0.42804125| PASSED sts_serial| 3| 100000| 100|0.96526517| PASSED sts_serial| 4| 100000| 100|0.32388866| PASSED sts_serial| 4| 100000| 100|0.14881751| PASSED sts_serial| 5| 100000| 100|0.69732895| PASSED sts_serial| 5| 100000| 100|0.74439824| PASSED sts_serial| 6| 100000| 100|0.44819093| PASSED sts_serial| 6| 100000| 100|0.27052991| PASSED sts_serial| 7| 100000| 100|0.68303103| PASSED sts_serial| 7| 100000| 100|0.98073079| PASSED sts_serial| 8| 100000| 100|0.05490760| PASSED sts_serial| 8| 100000| 100|0.46844079| PASSED sts_serial| 9| 100000| 100|0.94170852| PASSED sts_serial| 9| 100000| 100|0.37246207| PASSED sts_serial| 10| 100000| 100|0.78708430| PASSED sts_serial| 10| 100000| 100|0.43940563| PASSED sts_serial| 11| 100000| 100|0.95682412| PASSED sts_serial| 11| 100000| 100|0.91341183| PASSED sts_serial| 12| 100000| 100|0.94059147| PASSED sts_serial| 12| 100000| 100|0.91720462| PASSED sts_serial| 13| 100000| 100|0.29977341| PASSED sts_serial| 13| 100000| 100|0.80663660| PASSED sts_serial| 14| 100000| 100|0.65174466| PASSED sts_serial| 14| 100000| 100|0.10931696| PASSED sts_serial| 15| 100000| 100|0.31703586| PASSED sts_serial| 15| 100000| 100|0.03218640| PASSED sts_serial| 16| 100000| 100|0.01650420| PASSED sts_serial| 16| 100000| 100|0.21720490| PASSED rgb_bitdist| 1| 100000| 100|0.96762642| PASSED rgb_bitdist| 2| 100000| 100|0.52038768| PASSED rgb_bitdist| 3| 100000| 100|0.75233658| PASSED rgb_bitdist| 4| 100000| 100|0.37138772| PASSED rgb_bitdist| 5| 100000| 100|0.88690941| PASSED rgb_bitdist| 6| 100000| 100|0.64131006| PASSED
Last edited by awsdert; 01-10-2020 at 04:59 PM.
I just noticed this part of your dieharder output (from 10 hours ago):
It looks like dieharder is using the "mt19937" RNG by default (the man page for dieharder confirms this). That is the "Mersenne Twister" using 19937 bits of state. This is a pretty good RNG, as you can see from dieharder's analysis.Code:rng_name |rands/second| Seed | mt19937| 7.47e+07 |2183355501|
Edit to add: you need to call dieharder with the "-g 200" option to make it read your "random" data on standard input.
Last edited by christop; 01-10-2020 at 04:58 PM.
stahta01's manual extract confirms my example, except that you might want to look at whether the other options are relevant. The pipe does use stdin: recall that it pipes the output of your program (to stdout) as input to the other program (as stdin). If you really did run those tests with input redirection instead of piping though, then it would seem that this utility ran statistical tests on an executable program rather than the output of said program... I'd have thought that the content of an executable wouldn't be statistically random enough, but oh well.
EDIT:
So the utility just discarded what was fed to it via standard input?
Last edited by laserlight; 01-10-2020 at 05:02 PM.
Look up a C++ Reference and learn How To Ask Questions The Smart Way
Well the change at least caused a pass on some previously failed tests
Still running, for reference the change I made is this:Code:make --no-print-directory dieharder_rnd gcc -ggdb -Wall -lpthread -DBUILD_FOR_DIEHARDER -o ./mcc_rnd_die.elf mcc_rnd.c ./mcc_rnd_die.elf > diehard.txt dieharder -a -f diehard.txt -t 100 -Y 1 #=============================================================================# # dieharder version 3.31.1 Copyright 2003 Robert G. Brown # #=============================================================================# rng_name | filename |rands/second| mt19937| diehard.txt| 7.96e+07 | #=============================================================================# test_name |ntup| tsamples |psamples| p-value |Assessment #=============================================================================# diehard_birthdays| 0| 100| 100|0.92161723| PASSED diehard_operm5| 0| 1000000| 100|0.95034250| PASSED diehard_rank_32x32| 0| 40000| 100|0.04762000| PASSED diehard_rank_6x8| 0| 100000| 100|0.98629424| PASSED diehard_bitstream| 0| 2097152| 100|0.68113481| PASSED diehard_opso| 0| 2097152| 100|0.82427499| PASSED diehard_oqso| 0| 2097152| 100|0.73382595| PASSED diehard_dna| 0| 2097152| 100|0.80381661| PASSED diehard_count_1s_str| 0| 256000| 100|0.88826073| PASSED diehard_count_1s_byt| 0| 256000| 100|0.69131699| PASSED diehard_parking_lot| 0| 12000| 100|0.42137346| PASSED diehard_2dsphere| 2| 8000| 100|0.78353142| PASSED diehard_3dsphere| 3| 4000| 100|0.83751604| PASSED diehard_squeeze| 0| 100000| 100|0.97324599| PASSED diehard_sums| 0| 100| 100|0.03890804| PASSED diehard_runs| 0| 100000| 100|0.37897033| PASSED diehard_runs| 0| 100000| 100|0.28134077| PASSED diehard_craps| 0| 200000| 100|0.98255307| PASSED diehard_craps| 0| 200000| 100|0.19981783| PASSED marsaglia_tsang_gcd| 0| 10000000| 100|0.52946603| PASSED marsaglia_tsang_gcd| 0| 10000000| 100|0.11235286| PASSED sts_monobit| 1| 100000| 100|0.94853669| PASSED sts_runs| 2| 100000| 100|0.98915854| PASSED sts_serial| 1| 100000| 100|0.53037634| PASSED sts_serial| 2| 100000| 100|0.99259101| PASSED sts_serial| 3| 100000| 100|0.69499416| PASSED sts_serial| 3| 100000| 100|0.25469722| PASSED sts_serial| 4| 100000| 100|0.31794760| PASSED sts_serial| 4| 100000| 100|0.65630293| PASSED sts_serial| 5| 100000| 100|0.10030703| PASSED sts_serial| 5| 100000| 100|0.33876643| PASSED sts_serial| 6| 100000| 100|0.11512964| PASSED sts_serial| 6| 100000| 100|0.92113071| PASSED sts_serial| 7| 100000| 100|0.35443944| PASSED sts_serial| 7| 100000| 100|0.73691352| PASSED sts_serial| 8| 100000| 100|0.80353276| PASSED sts_serial| 8| 100000| 100|0.86551832| PASSED sts_serial| 9| 100000| 100|0.93505112| PASSED sts_serial| 9| 100000| 100|0.98897060| PASSED sts_serial| 10| 100000| 100|0.84157456| PASSED sts_serial| 10| 100000| 100|0.89642801| PASSED sts_serial| 11| 100000| 100|0.94733618| PASSED sts_serial| 11| 100000| 100|0.88809914| PASSED sts_serial| 12| 100000| 100|0.62251654| PASSED sts_serial| 12| 100000| 100|0.74017302| PASSED sts_serial| 13| 100000| 100|0.61114068| PASSED sts_serial| 13| 100000| 100|0.70827034| PASSED sts_serial| 14| 100000| 100|0.99409109| PASSED sts_serial| 14| 100000| 100|0.86867344| PASSED sts_serial| 15| 100000| 100|0.87824157| PASSED sts_serial| 15| 100000| 100|0.60437611| PASSED sts_serial| 16| 100000| 100|0.65594647| PASSED sts_serial| 16| 100000| 100|0.91562830| PASSED rgb_bitdist| 1| 100000| 100|0.30958867| PASSED rgb_bitdist| 2| 100000| 100|0.63426304| PASSED rgb_bitdist| 3| 100000| 100|0.14087628| PASSED rgb_bitdist| 4| 100000| 100|0.07156920| PASSED rgb_bitdist| 5| 100000| 100|0.13166009| PASSED rgb_bitdist| 6| 100000| 100|0.10396914| PASSED rgb_bitdist| 7| 100000| 100|0.85533507| PASSED rgb_bitdist| 8| 100000| 100|0.76165246| PASSED rgb_bitdist| 9| 100000| 100|0.90668495| PASSED rgb_bitdist| 10| 100000| 100|0.81370263| PASSED rgb_bitdist| 11| 100000| 100|0.56225235| PASSED rgb_bitdist| 12| 100000| 100|0.62578504| PASSED rgb_minimum_distance| 2| 10000| 1000|0.14524290| PASSED rgb_minimum_distance| 3| 10000| 1000|0.71798653| PASSED rgb_minimum_distance| 4| 10000| 1000|0.62044967| PASSED rgb_minimum_distance| 5| 10000| 1000|0.92519925| PASSED rgb_permutations| 2| 100000| 100|0.80781758| PASSED rgb_permutations| 3| 100000| 100|0.22465441| PASSED rgb_permutations| 4| 100000| 100|0.06486625| PASSED rgb_permutations| 5| 100000| 100|0.88229895| PASSED rgb_lagged_sum| 0| 1000000| 100|0.29016285| PASSED rgb_lagged_sum| 1| 1000000| 100|0.44218168| PASSED rgb_lagged_sum| 2| 1000000| 100|0.87158911| PASSED rgb_lagged_sum| 3| 1000000| 100|0.40383383| PASSED rgb_lagged_sum| 4| 1000000| 100|0.20346561| PASSED rgb_lagged_sum| 5| 1000000| 100|0.06987294| PASSED rgb_lagged_sum| 6| 1000000| 100|0.96547707| PASSED rgb_lagged_sum| 7| 1000000| 100|0.65228771| PASSED rgb_lagged_sum| 8| 1000000| 100|0.49786721| PASSED rgb_lagged_sum| 9| 1000000| 100|0.03422759| PASSED rgb_lagged_sum| 10| 1000000| 100|0.34623429| PASSED rgb_lagged_sum| 11| 1000000| 100|0.34719122| PASSED rgb_lagged_sum| 12| 1000000| 100|0.37062504| PASSED rgb_lagged_sum| 13| 1000000| 100|0.74491138| PASSED rgb_lagged_sum| 14| 1000000| 100|0.88931778| PASSED rgb_lagged_sum| 15| 1000000| 100|0.13501736| PASSED rgb_lagged_sum| 16| 1000000| 100|0.06369016| PASSED rgb_lagged_sum| 17| 1000000| 100|0.88156651| PASSED rgb_lagged_sum| 18| 1000000| 100|0.51469867| PASSED rgb_lagged_sum| 19| 1000000| 100|0.16741199| PASSED rgb_lagged_sum| 20| 1000000| 100|0.42849592| PASSED rgb_lagged_sum| 21| 1000000| 100|0.40387072| PASSED rgb_lagged_sum| 22| 1000000| 100|0.25274004| PASSED rgb_lagged_sum| 23| 1000000| 100|0.28240284| PASSED rgb_lagged_sum| 24| 1000000| 100|0.66642375| PASSED rgb_lagged_sum| 25| 1000000| 100|0.83348324| PASSED rgb_lagged_sum| 26| 1000000| 100|0.39567244| PASSED rgb_lagged_sum| 27| 1000000| 100|0.98195499| PASSED rgb_lagged_sum| 28| 1000000| 100|0.04585665| PASSED rgb_lagged_sum| 29| 1000000| 100|0.98016458| PASSED rgb_lagged_sum| 30| 1000000| 100|0.99756248| WEAK rgb_lagged_sum| 30| 1000000| 200|0.97082170| PASSED rgb_lagged_sum| 31| 1000000| 100|0.61949755| PASSED rgb_lagged_sum| 32| 1000000| 100|0.72894547| PASSED rgb_kstest_test| 0| 10000| 1000|0.09472064| PASSED dab_bytedistrib| 0| 51200000| 1|0.27907397| PASSED dab_dct| 256| 50000| 1|0.64134898| PASSED Preparing to run test 207. ntuple = 0 dab_filltree| 32| 15000000| 1|0.97399778| PASSED dab_filltree| 32| 15000000| 1|0.95593012| PASSED Preparing to run test 208. ntuple = 0 dab_filltree2| 0| 5000000| 1|0.01867152| PASSED dab_filltree2| 1| 5000000| 1|0.23876266| PASSED Preparing to run test 209. ntuple = 0 dab_monobit2| 12| 65000000| 1|0.35946069| PASSED Compilation finished successfully.
to this:Code:return val;
in the ulong versionCode:return val >> (val % (sizeof(long) * CHAR_BIT));
Last edited by awsdert; 01-10-2020 at 05:30 PM. Reason: Updated output again
