## Search found 2927 matches

### Re: Squares

It gives me a little insight into how the compressor is working..Maybe i can figure something out I believe you are searching for something that does not exist. When you claim you can compress whole files down to a residue or seed of only one byte, you are effectively saying that you can only regen...

- May 25, 2019 20:31
- Forum: General
- Topic: Need faster Exponentiate
- Replies:
**32** - Views:
**2174**

### Re: Need faster Exponentiate

The problem with approximating diadic functions like a^b is that they cover an area rather than a line, so usually require two, or possibly three, approximations to get one answer. There are a few possible ways to speed it up further. 1. Implement it as macro “inline code” to remove the function cal...

- May 25, 2019 10:12
- Forum: General
- Topic: Need faster Exponentiate
- Replies:
**32** - Views:
**2174**

### Re: Need faster Exponentiate

Latest version is slower by two multiplies and two adds. I have reduced error dependency on the power b, by increasing the order and accuracy of the Log2 approximation. It is also now less kinky and better behaved. Use this version when higher power errors are more of a problem than speed. Maximum e...

- May 21, 2019 16:34
- Forum: General
- Topic: Need faster Exponentiate
- Replies:
**32** - Views:
**2174**

### Re: Need faster Exponentiate

This more mature version is faster now that several instructions have been removed, merged or modified. The error function has unfortunately increased slightly, but is much better behaved since I have polished the polynomial coefficients. The step changes and reflections in the error function when c...

- May 19, 2019 20:30
- Forum: General
- Topic: Need faster Exponentiate
- Replies:
**32** - Views:
**2174**

### Re: Need faster Exponentiate

I'm happy that you find them useful. You can credit me as “Richard Freebasic”. I am interested in the particular n-gram techniques you are using for the solution of substitution ciphers. Do you have any references to papers or books that cover n-gram based solution. These routines were written as “p...

- May 19, 2019 5:51
- Forum: General
- Topic: Need faster Exponentiate
- Replies:
**32** - Views:
**2174**

### Re: Need faster Exponentiate

@Provoni The 12% errors in the above crude approximation can be reduced by taking the log2 of the mantissa and the antilog2 of the result. That can be done with simple polynomials faster than the intrinsic function calls. It is important to use only primitive +, –, * and bitwise operators. Avoid div...

### Re: Squares

Albert wrote:I got lossless compression working...

Then it is simply another a "hash function". It is not "lossless compression".Albert wrote:The compressor works. I just can't figure out how to decompress it.

### Re: Squares

@albert.

I believe in the science of information theory which says that you cannot compress truly random data.

I think you have tricked yourself, not Zlib.

I believe in the science of information theory which says that you cannot compress truly random data.

I think you have tricked yourself, not Zlib.

- May 16, 2019 22:16
- Forum: General
- Topic: Need faster Exponentiate
- Replies:
**32** - Views:
**2174**

### Re: Need faster Exponentiate

'======================================================================= ' approximate a ^ b Function approx_power( Byval a As Double, Byval b As Double ) As Double Dim As Long Ptr fp = 1 + Cptr( Long Ptr, @a ) Dim As Long tmp = ( *fp - 1072632447 ) * b *fp = tmp + 1072632447 Return a End Function ...

### Re: Squares

Now you need to find out how CUshort is supposed to work. Screen 19 Do cls ' make a string of bytes Dim As String v, s = "0123456789" Dim as integer i For i = 0 To len(s) - 1 s[ i ] = Int( Rnd * 256 ) Next i Dim As Short ts Dim As Ushort tus ' extract pairs of bytes Print "..." F...

### Re: Squares

@Albert.

You are using Short integers, so half the time you should expect a minus sign as the first digit.

Use Ushort integers instead.

You are using Short integers, so half the time you should expect a minus sign as the first digit.

Use Ushort integers instead.

### Re: Squares

@Albert MKI will output an integer for your platform, 32 or 64 bits. Use MKL to force 4 bytes. “ MKI supports an optional <bits> parameter before the argument. If bits is 16, MKShort will be called instead; if bits is 32, MKL will be called; if bits is 64, MKLongInt will be called. The length of the...

### Re: Squares

Richard has been a downer on the compression idea , but I've used so much of his code in my programs, i thought I'd include him anyways. @Albert. Please don't include or associate me with your data compression ideas. I would really like to maintain some distance between myself and the liability. I ...

### Re: Squares

I think i got the compressor working.. There's no way to decompress yet.. @Albert. There is no way your grid approach is going to be reversible. You need to stop posting hash functions and get back to us when you have a package that will compress to 50% or more, and will decompress to the original ...

### Re: Squares

Albert wrote:I found an algorithm that works!!

How do you know it works if you have not thoroughly tested it.Albert wrote:The compression and decompression are extremely fast... I still need to run some tests , to see if it decompresses properly..

Your compression factor is subject to the Dunning-Kruger effect.