stan

The document at nist draft publication of entropy assessment recommendations gives a series of tests that can be performed on random number generators to validate them. There is a description page with a link to the program here users guide to running draft test suite When I run those tests on the output of rtl2832-entropyd, it passes with entropy estimates a few tenths below the size of the bits tested on the data for bit sizes of 1, 2, 3, 4, 5, 6, 7. For instance, the estimate for 1 bit sample...

The document at nist draft publication of entropy assessment recommendations gives a series of tests that can be performed on random number generators to validate them. There is a description page with a link to the program here users guide to running draft test suite When I run those tests on the output of rtl2832-entropyd, it passes with entropy estimates a few tenths below the size of the bits tested on the data for bit sizes of 1, 2, 3, 4, 5, 6, 7. These sizes also validate with their sanity...

The document at gives a series of tests that can be performed on random number generators to validate them. There is a description page with a link to the program here When I run those tests on the output of rtl2832-entropyd, it passes with entropy estimates a few tenths below the size of the bits tested on the data for bit sizes of 1, 2, 3, 4, 5, 6, 7. These sizes also validate with their sanity checker. It passes for bit sizes of 8 bits with an estimate of approximately 7.9 bits of entropy per...

Starting in 4.13 kernel, the cryptographic pseudo random number generator used by the kernel changed from a modified mersenne twister (I think) to chacha20. The way that entropy is used was also changed, so that it isn't utilized as much as it used to be. rtl2832-entropyd still works because the api it calls is still there, but the entropy it provides is not fed continuously into the kernel entropy pool, because entropy only leaves the pool now when the kernel get_random function is called. This...

Moved to 3.0 because the internal code was pretty extensively modified, though functionality remains the same. The display of sprites is tuned to the size of the sprite when use background and keep background are set. So the appearance of the screen changes from an always large black background to a black background the size of the sprite. There were flaws in the background save and restoration code, fixed now. The size and name of the center and periphery sprite used for a run are now saved to files...

https://www.sciencedaily.com/releases/2017/06/170619092518.htm If it can do this for the elderly, it probably can do this for people of all ages. It's never too early to incorporate good habits.

This change fixes a bug that was in 2.3, and adds some features. The program functionality remains the same. There were variables that could be evaluated out of order in 2.3, causing a crash. This fixes that. Reworked the scoring algorithm, so that it yields lower values. Lower is still better, but scores in the 100 to 200 range are now excellent. Added the parameters for the scoring algorithm to the config file so that scoring can be adjusted by the user. Added all the new outputs to the final statistics...

This change fixes a bug that was in 2.2, and adds a few new features. Functionality remains the same. When I was packaging up 2.2, I was editing the config file and realized that file_append more accurately described the new file save function than file_write, so I changed it, meaning to change it in the program. But I got sidetracked and forgot. So this fixes that functionality, files will now be saved if the option is set. Added new options to save the raw data from a run. This saves the times...