Thank you very much, everyone! I was aware that my solution was far from perfect, as the average probability to move each number to another place is the same, but it is just a probability and therefore some numbers just stay at their place. What was really surprising was that the amount of numbers t...

Hi everyone! While my first post to this forum was in 2005, during the last 5-6 years I did not interact with the community. It's my fault and I hope to be regarded as a prodigal son rather than as a black sheep, especially because (as usual!) I am back here for a reason. I need to generate a list o...

The 'Doctor Web' antivirus does the same false report, based on heuristics checks. I think this is the only explanation, and I am glad to know I'm not the only one who got a virus alert. In fact, I checked the "infected" FBIDETEMP.exe sending it to virustotal.com and it turned out that th...

About one month ago my AVIRA found a virus in the FBIDETEMP.exe that was compiled while I was executing a small piece of code. When I tried to run AVIRA for a full scan, right after that detection, it turned out that several api-ms-win-downlevel-*-l1-1-0.dll (the * replaces several names) were missi...

Thanks, dkl! I was pretty sure that FB was not involved in what happened and, if a fake virus alert was the only problem, I would not have been much concerned. The missing api-ms-win-downlevel-*-l1-1-0.dll files (I realized that *-user32-* is not the only missing dll) however, are a real problem bec...

Hi everyone! While I'm no more an active user of this forum since a few years (not enough time, unfortunately!), I still know it's the only place where one can find advice about every FB-related issue. Yesterday I noticed that a new version was available (1.01) and I installed it on my WIn7 32 bit P...

As always I learned quite a bit from the forum. Thanks a lot everyone!

Thanks, Richard and MichaelW! I was (partly) aware of the rounding problems due to the different basis, as well as of those related to the byte order, but what really puzzled me was the fact that I was reading a value from a binary file that was apparently interpreted (i.e. shown in console output) ...

I wrote a very simple program to read a single precision array from a binary file generated by Surfer 9 (a popular geostatical tool). Missing data are coded in the array as the minimum 32 bit floating point value (-3.402823e+38) and I want to recognize them as easily as possible. Although there are ...

For the purposes here the fix is to simply remove this instruction, with the probable side effect of making the code slightly faster. I have updated my posts. Thanks a lot, MichaelW. Now it works alike a charm and it's actually the fastest solution. I'm used to write 99% of my code for computationa...

I just ran a few more tests on both types of solutions, i.e. (1) processing only one value at a time and (2) processing the whole array. As for single value processing, I completely removed data I/O from my test program, because it took so much more time than the actual conversion that its variabili...

Thanks to MichaelW and I3I2UI/I0 for their asm-based solutions, I'm going to test them and post the results ASAP!

Can you post at least one of the two solutions you have validated? Yes, of course. Basically it's just a test with s sample HDF file containing chlorophyll concentration data from the MODISA remote sensing mission. The data file mentioned in the code is available at http://oceandata.sci.gsfc.nasa.g...

I ran a few test with Richard's reverse() function as well as with SARG's union-based solution. Reading a very large data file (37324800 4-bytes floating point values) the reverse() function was 15% to 20% faster than the union-based method. Unfortunately I wasn't able (my fault) to adapt fxm's &quo...

Thanks a lot everybody for your quick replies. I had problems with a couple of potential solutions (my fault, i guess, I'm not used to handle pointers, but I'll try again later), while Richard's reverse() function (as well as its sub counterpart) works very nicely. So, special thanks to Richard, I'l...