What's new - change history

Back to main page

Here is a sketched history of changes on this site.
Since unfortunately I currently do not know how to setup an RSS feed or similar stuff I can only offer to inform you on updates via e-mail.
Comments or questions? E-mail: info@sirrida.de

2023-01-23

• Update / speed up of the routines for greatest common divisor (gcd) by using a variant without modulo operations. Thanks to Stefan Kanthak!

2021-08-30

• After a successful and exciting transition journey: Since today my new official name is Sigrid Johanna Neumann.

2020-08-30

• The program calcperm.cpp / calcperm.pas crashed with an access violation and also gave wrong results for the new merge sublist method for some specially optimized cases. Fixed. Many thanks to Robert Clausecker who found these bugs!

2020-03-17

• After a way too long time finally some progress. I implemented and commented the "new" Lee's algorithm (merge lists) and enhanced calcperm.cpp and its Pascal equivalent accordingly. The online PHP version however still misses this improvement.
• Having slowly started my transition about a year ago I no longer want to keep my future name Sigrid a secret (even if it sounds very similar). Incidentally a reference to this name has been on my site from the very beginning but this usage really had never been anticipated. Looking forward in curiosity…

2018-06-01

• Moved hashing stuff to a dedicated page.

2018-05-28

• Added some notes on privacy policy.

2018-05-07

• An implementation in x86 assembler of bit_permute_step, bit_permute_step_simple, bit_permute_step2.

2018-01-25

• More on and fixup of C++ style method references.

2018-01-15

• More on C++ style method references.

2018-01-11

• More on calculations with offsets and C++ style method references.

2017-12-30

• More on calculations with offsets.

2017-03-02

• Filling holes in dispatching on strings in theory.

2017-02-19

• Just more about dispatching on strings in theory.
• I at least temporarily disabled my guestbook due to way too much spam. I am very sorry about that.

2017-02-14

• More about dispatching on strings in theory.

2016-07-02

• Scalar multiplication (even fields) in SIMD/SWAR operations.

2016-06-30

• Scalar multiplication (upper half) in SIMD/SWAR operations.
• Boolean SWAR functions yielding MSB, LSB, or mask values.

2016-06-27

• Scalar multiplication in SIMD/SWAR operations.

2016-06-23

• Add/Subtract with carry in SIMD/SWAR operations.

2016-06-17

• A new chapter about SIMD/SWAR operations.

2015-10-06

• Update of the online permutation code generator to optionally disallow Beneš / Butterfly routing.

2015-10-06

• Update of haroldbot: Online bitwise calculator and theorem solver by Harold Aptroot

2014-12-25

• Added the proposal to emit a Pascal hint when using implicit dereferencing.
• Added the proposal to emit a Pascal hint when comparing boolean values.
• Added a proposal on how to deal with relations.

2014-12-13

• Added a section about iterators.

2014-09-25

• Added the function odd in perm_bas.*. This function is intrinsic in Pascal.
• Added a link to the influence on a future C++ standard in a section about the intended source usage.

2014-09-20

• Added a function to calculate the multiplicative inverse mul_inv in perm_bas.*.

2014-09-16

• Added a missing typecast to the routine testing tbm in testperm.* where the result of tbm was inverted which failed for types smaller than int due to the peculiar integer casting of C; the Pascal version did not fail but nevertheless I added the typecast. Unfortunately even now GCC incorrectly bemoans the comparison when specifying -Wextra as "warning: comparison of promoted ~unsigned with unsigned".
• Renamed
  • nr_leading_zeros to nr_leading_0bits,
  • nr_trailing_zeros to nr_trailing_0bits,
  • is_contiguous_bits to is_contiguous_1bits.

2014-09-15

• Renamed nr_bits to nr_1bits.

2014-09-10

• Changed the name bit_index_rot to bit_index_ror since we are rotating to the right.

2014-09-01

• Enhanced some comments in xshift.asm featuring macros for bitwise shifts on xmm registers.

2014-08-31

• The sources in sources.zip now are stored in a subdirectory. This avoids accidentally polluting your working directory. Other archives such as the one below might follow.
• Submitted some source files of a prototype of a perfect hashing library in order to give some facts and some code to play with to the section on perfect hashing.

2014-08-04

• Renamed the rotate functions vror / vrol / vrot to vror_bfly / vrol_bfly / vrot_bfly.
• Introduced vror / vrol / vrot, blend and simd_odd.

2014-07-23

• Corrected types in bit_permute_step, bit_permute_step_simple, bit_permute_step2, bit_index_complement, bit_index_swap. bit_index_swap_complement, permute_bpc, general_reverse_bits in perm_bas.*.
  This was not critical but nevertheless wrong.

2014-07-07

• Precautions about unsuitable systems.

2014-07-07

• Renamed the rotate functions fror / frol / frot to fror_bfly / frol_bfly / frot_bfly and fror_alt / frol_alt / frot_alt to fror / frol / frot.
• Introduced frorc / frolc / frotc.
• Corrected calcperm.cpp: Inverse Butterfly was signaled as always true. The Pascal variant was correct.

2014-07-06

• Cleaned up the sources where on few places shifts ≥ the used word size were used. Added some comments on this.
• Some more comments on word sizes and representation of numbers.
• Some tidying up on several places.

2014-06-27

• Renamed count_bits to nr_bits since a proper function should describe the result and not the action to achieve the result.

2014-06-13

• Changed the copyright year from 2013 to 2014.

2014-06-11

• Fixed the changed link to the Free Pascal development forum below the contents of my Pascal proposals page.

2014-06-09

• Thanked some more people and institutes.
• Offered to inform on updates via e-mail.
• Started a section on perfect hashing using more than one auxiliary hash function and lookup into additional arrays, exemplified by the method of Bob Jenkins. Probably this section will be grow largely.

2014-02-14

• Corrected once more links to AMD documents at several places since the place appears to have changed.

2014-02-11

• More about manipulating rightmost bits and an implementation in perm_bas.* of this function tbm.

2014-02-10

• A new chapter about manipulating rightmost bits.

2014-01-13

• Provided a link to our document about integer hashing in Hashing integers.

2013-10-01

• A new and faster solution to signed average.
• Some changes on my proposal of the extension to setcc. Swapped the actions of setcc xxx,zero. Added keywords to the table. Added the signed average example.

2013-09-06

• Changed the method for compacted perfect hashing.

2013-08-22

• Added some external links and enhanced their hints.

2013-08-21

• Changed the link to haroldbot - Bitwise calculator and theorem solver. The old server did no longer work correctly.

2013-08-15

• Added two links to the main page.
• Added mask_bit_i. See perm_32.asm and perm_64.asm.

2013-06-07

• A nice user (Harold Aptroot) gave me a link to first timing tests of PEXT and PDEP.
  Both commands have a latency of 3 cycles and are fully pipelined, i.e. they a reciprocal throughput of 1 cycle. I have adapted the relevant programs (calcperm.*) and the online permutation code generator to these findings.
• He also gave me the hint that all AMD and Intel processors seem to leave the target of BSR as is if fed with 0. AMD documents this but Intel leaves it as unspecified. I verified this with Intel P2, Atom, i7 (32 bit OS).

2013-05-08

• Added a chapter about hashing integers.
• Replaced all post-increment/decrement operations in the C/C++ sources by postfix operations.

2013-05-03

• Added a general chapter about hashing.

2013-04-27

• Added a formula to calculate the expected number of comparisons from the load factor for hash maps.

2013-04-18

• There is an active programming language supporting dimensions: F#! I was not aware of it before.

2013-04-16

• Added chapters about Dispatching on strings in theory and reality. Also moved some stuff about in related chapters.

2013-04-12

• I made the string switch solutions thread-safe, see case_*.* in Switch on strings (C/C++). However in order to safely use the solutions in a multi-threaded environment, you should activate USE_MUTEX in general.h and make sure that mutexes work by supplying a suitable compiler option such as -pthread for GCC. It might even be sensible to utilize the memory barriers.
• Added a section about Singletons / Do once only.

2013-04-05

• I measured some times of string compares and hash functions in order to get a better estimate of what the switches really cost. So the string switch solutions using self-made hash containers have new cost factors: case_m*.* in Switch on strings (C/C++).

2013-04-02

• All solutions of Switch on strings (C/C++) now accept a (superfluous) break statement.
• Provided a variant of multiple case targets (CASE_MULTI0) for all solutions of Switch on strings (C/C++).

2013-04-01

• Changed the following case_m*.*:
  • Optional dump of statistics (DEBUG_CASE_STATISTICS).
  • Optional dupe recognition (DEBUG_CASE_DUPE) at runtime.
  • Fixed a minor memory issue.
• Provided multiple case targets without using goto for all solutions of Switch on strings (C/C++).
• Added a description of the formal grammar for the Modula-style program flow statements for C / C++ in modula.txt.
• I developed a better version of my FOR macro in modula.h, see Counting loops in C/C++.

2013-03-30

• Added a hash container for STL strings: case_ms.cpp.
• Repaired ceil_power_2 used in case_map.*. This caused bad performance of the hash map (only 2 slots). The error was introduced on 2013-03-27.
• Calculated the lengths of the string constant beforehand and changed the string comparison in case_map.* to memcmp which ought to be faster than strcmp; also string identity is checked now which might speed up things.

2013-03-29

• Extracted the headers from some C sources.

2013-03-28

• Renamed some test programs: loop.c => testloop.c and case.* => testcase.*.

2013-03-26

• Added yet another alternative heuristic for Shift by multiplication in calcperm.*

2013-03-25

• Parameter include files for calcperm.*.

2013-03-24

• Changed the internal cost approximation and added an alternative heuristic for Shift by multiplication in calcperm.*

2013-03-23

• Reworked indenting according to Ratliff style.

2013-03-22

• After a self-test the programs calcperm.* end now.

2013-03-20

• Renamed some parameters for calcperm.*.
• Reversed the odometer loops in testperm.*.

2013-03-19

• Described the new permutation method Shift by multiplication and provided an implementation in calcperm.*.

2013-03-17

• Changed the default of using BMI operations to false in calcperm.*.
• Added a switch to control the direction of the permutation in calcperm.*. The new switch is named /indexes.

2013-03-14

• Shifted the internal case number range's lower bound to 0 in the hash map based implementations of Switch on strings (C/C++).

2013-03-05

• More on Counting loops in C/C++, also and other macros.

2013-03-03

• Added a discussion of Counting loops in C/C++.

2013-02-22

• More about dispatching on strings: Switch on strings (C/C++).

2013-02-15

• Added a program to test endian access records: endian.*:
• Restructured the include files in order to avoid duplication.

2013-02-14

• Added the following functions to testperm.*:
  • identity_perm
  • bfly_parity
  • benes_parity
• Created a new page programming.html.
• Moved the topic speeding up case of string to this new page.

2013-02-09

• Changed the copright year.
• Finally got a coarse translation of calcperm.pas to C++ running: calcperm.cpp. I used C++ instead of C here in order to ease string handling.

2013-01-31

• Fixed some signed/unsigned hints in perm_bas.c.

2013-01-19

• Removed transparency from the diagrams in order to improve anti-aliasing, i.e. make the embedded texts much better readable and the slanted lines less ragged.

2013-01-18

• Discussed some method of speeding up case of string in Pascal.

2012-12-30

• Refreshed some images.

2012-12-21

• Stated that demultiplexers can be used instead of multiplexers for switching networks.

2012-12-16

• Changed an outdated link.

2012-11-15

• Added small spaces between digits of long numbers.

2012-11-12

• Corrected some minor errors in calcperm.pas and perm_bas.c.

2012-11-04

• Reordered the main page.

2012-10-23

• The permutation code generator in Pascal (calcperm.pas) now optimizes by optionally pre- and post-fixing the permutation with ROL and/or BSWAP.

2012-10-20

• Removed empty parentheses for functions without parameters from the Pascal sources to avoid compiler problems.
• Somewhat cleaned up the permutation code generator in Pascal (calcperm.pas).

2012-10-19

• Restructured the comments on the permutation code generators.

2012-10-10

• Added the function bswap.
• Changes to the permutation code generator in Pascal (calcperm.pas):
  • Command line option parameters now working correctly.
  • Now there is an optional optimization by prefixing the permutation with ROL and/or BSWAP.
    This massively slows down the computation and the slow down becomes relevant for bits > 32 since the whole calculation for the remaining permutation is done for every combination with ROL and/or BSWAP (up to about 3*bits times).
    Do not expect the online permutation code generator to include this feature, since PHP is too slow for such escapades, even for 32 bits.

2012-10-06

• Added a primitive check for the assumptions of bit_permute_step_simple in testperm.*.

2012-10-05

• Added a mask count in the permutation code generator in Pascal (calcperm.pas).
• Removed set operations from perm_bas.pas.
• Gave a hint how to utilize the old 16 bit Pascal compilers Borland Pascal and Delphi-1.

2012-10-04

• Adjusted the cost boni in calcperm.ini for the permutation code generator in Pascal (calcperm.pas).

2012-10-03

• Cleaned up the permutation code generator in Pascal (calcperm.pas) and added a configuration file.
  Zeros were not correctly dumped.
• Changed the indentation of "end" in all Pascal sources so that "end" appears as the last statement of a statement list. This is also called Ratliff style.

2012-10-02

• Renamed the page of the online permutation code generator from calc.php to calcperm.php.
• Added a first source version of the permutation code generator in Pascal: calcperm.pas. Much stuff such as a configuration file and a version in C is still missing.

2012-10-01

• All those web search machines are too stupid to distinguish a domain name from the address where the site is hosted. Also a quasi-static ip address can change.
  So, finally I ordered "real" web space. That means that the domain name now appears on every page and that also redirection via index.php works again.

2012-09-27

• The server's ip address has changed again due to a problem on the server's provider Unitymedia.
• There are the following new functions:
  • used_source_bits
  • used_target_bits

2012-09-23

• Added sources for a word size of 8 bit.
• Corrected the types in random_int (auxiliary function).
• Added C sources for a word size of 128 bit. This appears to work with GCC under a 64 bit OS, albeit veeery slooow, also the constant definitions are clumsy.
  I could not find a Pascal compiler which can work with 128 bit integers.

2012-09-21

• The permutation sources are now ready to act on the word sizes of 16, 32 and 64 bit at your choice. Support for other sizes can be easily added as needed.
• I came to the conclusion that in order to support multiple word sizes without duplicating most of the source code it is necessary to get rid of most of the suffixes such as "_32" and "_64". In the assembler files these prefixes remain necessary.
  The sources are now broken up into multiple files: The include files are perm_bas.* and perm_b#.* where # is 16, 32 or 64. The main files have not changed their name.
  You can now choose which functions to use by including one of two small bit-specific files. The remaining stuff will need another include files.
  A command line version of the online permutation code generator will obviously also include these files.
• The following changes have been made on the whole site:
  • "t_32u" => "t_bits"
  • "randseed_32" => "my_randseed"
  • "random_32" => "random_bits"
  • "one_32" => "lo_bit"
  • "hi_32" => "hi_bit"
  • "all_32" => "all_bits"
  • "_32" => ""
  I hope that I did not break too much…
• The page containing the function descriptions (testperm.html) has been renamed to perm_fn.html.
• There are the following new functions:
  • nr_1bits
  • nr_leading_0bits
  • nr_trailing_0bits
  • is_contiguous_1bits
• All source files now reside in the archive sources.zip since we have now many quite small files which really need not be downloaded one by one.

2012-09-18

• Combined the starting pages index.php and index.html in order to ease automated indexing and to avoid unnecessary reloading by the browser.
  The appearance has not changed but redirection still works.

2012-09-17

• Described the new permutation method gather/scatter utilizing PEXT and PDEP.
• Enhanced the online permutation code generator to optionally challenge the methods gather/scatter and sheep and goats" utilizing PEXT and PDEP.

2012-09-12

• Added a summary.
• Added the functions rol to testperm.*.

2012-09-11

• Changed the cost of bit_permute_step_simple from 3 to 4 cycles in the online permutation code generator and in the comments in the test programs. 3 cycles are possible with the new RORX statement (BMI2) which shaves one MOV.
• Implemented bit group moving in the online permutation code generator.
• Comments to bit group moving.

2012-09-09

• Added the "don't care" (wildcard) entries of the actual permutation as dotted lines to the dynamically created image of the online permutation code generator.
• Added the following functions to testperm.*:
  • invert_perm
  • random_perm
  • gen_benes_ex
  • benes_fwd_ex
  • benes_bwd_ex
• The bit indexes (t_bit_index) now include the value -1 with the meaning "don't care" (wildcard); these added functions cope therewith:
  • invert_perm
  • gen_benes
  • gen_benes_ex

2012-09-07

• Added a dynamically created image to the online permutation code generator.

2012-09-06

• Corrected an error in the online permutation code generator while detecting BPC permutations.

2012-09-05

• Some options to the online permutation code generator.

2012-09-04

• Now there is an online permutation code generator for creating code for an arbitrary fixed 32 bit permutation!
  Probably C and Pascal source for 32 and/or 64 bit will follow soon.
• I also gave some hint on the inner workings of this generator in the chapter about speeding up permutations.

2012-07-23

• Now there is a backlink from Hacker's Delight.

2012-07-22

• Commented several links to external sources.

2012-07-18

• Corrected all_bits in testperm.pas.
• Replaced all inline copies of the bit_permute_step functions by a call thereof in testperm.*.
• Proposed another variant of the bit permutation primitives using rotate operations instead of shifts.
• The special case Y=X+1 in hypercube.

2012-07-17

• A new chapter about permutations on a hypercube.
• Corrected the wrong date and added "_32" to the bit_permute_step function names in testperm.*.
  ("_32" removed later on 2012-09-21)
• Added some documentation to the bit_permute_step functions.
• Added a link to Hacker's Delight, for the bit permutation primitives.

2012-07-13

• Corrected the text to Knuth's algorithm for compress-flip in arbitrary permutations.
• Provided an example for Knuth's algorithm for SAG and compress-flip.
• Mentioned the "shuffled row major" permutation.

2012-07-12

• I have sketched Knuth's algorithm to use SAG or compress-flip to perform arbitrary permutations in at most sw steps.
• Added gray_code and inverse_gray_code to testperm.*.

2012-07-05

• More about speeding up permutations by means of PSHUFB.
• Proposal of a new variant of the bit permutation primitive
• A new assembler file xshift.asm featuring macros for bitwise shifts on xmm registers.

2012-06-19

• Some work on the chapter about alternatives.
• New chapters about speeding up permutations and citations.

2012-06-18

• Corrected the stripped down variant of my proposal of a bit permutations instruction once more because it appears that amazingly only one primitive suffices for most applications.
• The bit permutation primitive is now a separate chapter.

2012-06-17

• Corrected the stripped down variant of my proposal of a bit permutations instruction because we need two routines.

2012-06-11

• I corrected the bit index rotation direction for shuffle and unshuffle.
• bit_index_ror actually rotates right instead of left.
  These documentation bugs have been there since the first version.

2012-06-06

• Added a comment to a stripped down variant of my proposal of a bit permutations instruction.
• Added a chapter about alternatives for calculating bit permutations.

2012-06-01

• One year sirrida.de!

2012-05-07

• Referenced bookmark on every page.

2012-03-29

• The redirection feature ceased to work since the address parameter no longer comes through. This is a pity!
• Replaced a dead link regarding dimensions.

2012-02-13

• Added some comments on the used programming languages.

2012-02-12

• Added routines to invert bit permutation vectors and BPC permutation parameters.
• Added a routine to calculate BPC permutations.

2012-02-04

• Enhanced the text of Bit index manipulations.

2012-01-29

• Enhanced the table of Bit index manipulations.
• Added an index for my ideas for Pascal to the contents.

2012-01-28

• Optimized index_of_bit_i by assuming that i is in range. See perm_32.asm and perm_64.asm.
• In all the images the steering lines are blue now. I also changed the multiplexers and steering bits which are inactive for a Waksman network into very bright instead of dark.

2012-01-18

• Changed the parameter order of index_of_bit_i to save 2 MOV instructions in 64 bit mode and also optimized these routines. See perm_32.asm and perm_64.asm.

2012-01-15

• Added an extra example for the usage of PDEP which makes a loopless evaluation of the index of the i-th set bit possible (index_of_bit_i) in perm_32.asm and perm_64.asm.

2012-01-11

• Changed the copyright's year.
• Added a small site icon next to the address line for all pages.
• Enhanced the index.

2012-01-09

• Added an index to some important words for my discussion of bit permutations.

2012-01-08

• Comment about the parity of permutations in the text for the Beneš network.

2012-01-01

• Enhanced the text for dimensions.

2011-12-31

• Added some ideas to enhance Pascal. Some of these ideas apply to other languages as well.
• Corrected some minor HTML errors.
• Changed some lists to tables.

2011-12-28

• Added some hints to the origin of some procedures in the source files.
• Several other minor changes.

2011-12-23

• Added some text and diagrams to clarify the correspondence of transpositions and shuffle/unshuffle.

2011-12-21 … 2011-12-23

• The server unfortunately was unavailable for about 2½ days because it had hung up and the supervisor was on holidays.

2011-12-18

• Added an image of a (hyper)cube as a visualization of the arrangement of butterfly stages.

2011-12-17

• Enhanced some text in Backstage.

2011-12-16

• Noted that some steering bits of a Beneš network can be forced to be always 0, i.e. the hardware implementation can save some multiplexers, creating a Waksman network. Added a test therefor in testperm.*.

2011-12-12

• Noted that butterfly stages can be visualized as actions on a hypercube.

2011-12-11

• Noted that the omega-flip network is also called a shuffle exchange network and added a diagram to justify this.

2011-12-08

• Changed some text and added images to describe the workings of multiplexers for the butterfly network

2011-12-07

• Added an image of the Beneš network and changed the related text a bit.

2011-12-06

• Added a test for the Beneš network in testperm.*.

2011-12-03

• Corrected a link to an AMD document at several places since the place appears to have changed.

2011-12-02

• Implemented the generation of the configuration of a Beneš network for arbitrary bit permutations in gen_benes

2011-11-18

• Added a comment hint about optimization by loop-induction.

2011-10-26, 2011-10-25, 2011-10-24

• The server's ip address has changed again three times due to server maintenance.

2011-10-18

• Added a comment about equation syntax in intro.

2011-10-07

• Several changes in Bit index manipulations for BPC permutation.
• Several changes in Generalized bit reversal, e.g. comment that they can mirror hyperrectangles.

2011-10-04

• More about Bit index manipulations.
• Added some new functions to the test programs together with test procedures and brief description:
  • bit_index_ror
  • transpose
  • shuffle_power
  • unshuffle_power

2011-09-28

• Corrected the comment about Gray permutations with Swap by primitives, which can do not only this but also other permutations. I really should read more carefully…
• Added some comments about bit index functions Bit index manipulations.

2011-09-27

• Added a chapter about Bit index manipulations and some related functions. Also added some references therefor on several places.
• Added some words to the Word definition list and created a word index.

2011-09-25

• Added a comment that the two innermost stages of a Beneš network can easily be replaced by one, see butterfly network.

2011-09-21

• Yet another comment on "rep movsb" in Gather: rep xlat
• Globally renamed rolc to rolc_lo and analog similarly named ones.
• New auxiliary function rol_lo (currently not used).
• I wrote an intro to bit permutations.

2011-09-15

• Globally renamed rotc to frolc and analog similarly named ones.

2011-09-14

• Globally renamed the fixed count SWAR rotate functions: rot to frot, ror to fror, rol to frol, and analog similarly named ones.
• New functions frot and vrot
• I decorated shift, rotate, frot, and vrot with images.
• Description what a word is.

2011-09-13

• Introduced a direction flag in order to be able to combine the functionality of the left and right variant of routines to one.
  Introduced these combined routines.
• Renamed shuffle_bits to shuffle and unshuffle_bits to unshuffle globally.
• The new page perm_fn.html adds a detailed function description of testperm.*. I provided links to this list in Comments on the test programs. Also added several links to specific functions.

2011-09-10

• Reordered the test programs testperm.*.
  Added all the missing compound routines.
  Cleaned up the comments in the butterfly operation "class".
  Added tests for the compress_mask routines.
• In testperm.pas I had wrongly specified $o- instead of $q- to disable overflow check.
• In Compress and expand asserted that compress_mask(m,sw) = compress(m,m,sw).
  Made the assumptions more general and also described the assumptions more clearly.

2011-09-09

• Added functions to testperm.* which calculate the extra masks needed for the extended versions of Compress and expand when using PEXT and PDEP.
• Added the verification of the claim mentioned above to the test programs.
• Added a global error flag to testperm.*.

2011-09-07

• Added a comment on capitalization in Comments on the test programs.

2011-09-06

• Added the verification of [un]shuffle with Compress and expand to testperm.*.

2011-09-05

• Tidied up documentation in perm_32.asm and perm_64.asm.
• Modified the asm function names to conform to testperm.*, i.e. added the parameter values to their name which are missing to meet the full blown cousins.
• Added a comment in Shuffle and unshuffle on how to emulate [un]shuffle with Compress and expand or PEXT and PDEP.
• Added a comment in Compress and expand of the possible implementation via PEXT and PDEP.
• Added some ideas about the site's future in Motivation.
• Added comments about patents and affiliation in Intro.

2011-08-31

• I found the second public reference to my pages at Chess programming – Bit-Twiddling introduced 2011-08-27.

2011-08-27

• New x86 proposal: Gather and scatter with configurable factor
• The second public reference to my pages at Chess programming – Bit-Twiddling.

2011-08-23

• The images in Bit permutations got titles beneath.

2011-08-22

• Provided a link to Intel's documentation in the assembler sources perm_32.asm and perm_64.asm.
• I refined the documentation of time stamps.
• Added a comment that I want to be informed of errors.

2011-08-19

• On 2011-08-18 we had a very rainy evening (a new record for Aachen) causing several power outages. The server was off-line for some hours and unfortunately its ip address got changed thereafter. On the next day about 12:00 all was up and running again.
• Notice about possible ip address changes in Bookmarking.

2011-07-21

• Comments on the time format.
• Comments on the used language in the Intro.
• I have placed the references to the extra source files in an own chapter.
• Put a link to the assembler routines of PEXT and PDEP just under the contents listing.
• Added a comment that this is my private hobby.

2011-07-20

• The assembler routines [un]shuffle of PEXT and PDEP have only one parameter. I removed the wrong parameter.
• In the assembler routines I had written xbx instead of ebx.
  I normally use pseudo registers such as xbx to mean ebx or rbx dependent on the used environment.
• I made different versions for 32 and 64 bit assembly as extra files and removed the lengthy assembler code stuff from the HTML page.

2011-07-19

• Added some test routines to testperm.*.
• Sketched the conversion of the configuration between butterfly and omega-flip network, like done in test routines.
• Minor changes.

2011-07-18

• Comment on [un]shuffle applied sw times.
• Added butterfly, omega and flip to testperm.*.
• In the diagrams with multiplexers (butterfly and omega-flip) I marked the lines which connect to an "1"-input with the color magenta. In the other diagrams I marked some connectors as well.
• I changed some colors in several diagrams.
• Some text changes and additions on Omega-flip network.

2011-07-16

• I have drawn some more diagrams and added the texts Source and Target to all diagrams.

2011-07-15

• Introduction of this list as a public means to see what has changed. The earlier entries are partly guesswork and all but complete.
• I found an error!
  In testperm.*: shuffle and unshuffle repaired: The array index was off by one. The once working routines broke by a refactoring operation. Unfortunately a test routine for these procedures is still missing. Who wants to help me writing one?

2011-07-14

• First version of the chapter about the Omega-flip network.

2011-07-01

• Translation of the test program in C testperm.c to the Pascal version testperm.pas.

2011-06-13

• First public reference at The Aggregate Magic Algorithms by Hank Dietz.

2011-06-01

• Registration of the top level domain sirrida.de.
• Creation of programming.sirrida.de.

2011-05-28

• Creation of the logos.
• First diagrams.

2011-05-18

• First pages in HTML.

About 2011-02

• Development of some of the routines you find in testperm.c, e.g. the gen_cef_* routines.