digitalmars.D - Is there a good lib out there to handle large integer of know size ?

deadalnix (3/3) Jan 06 2016 In my case, I'm interested in 160 and 256 bits integers. Using

H. S. Teoh via Digitalmars-d (6/9) Jan 06 2016 If not, it may be something nice to add to Phobos. Integer width can be
Guillaume Piolat (3/6) Jan 06 2016 https://github.com/d-gamedev-team/gfm/blob/master/math/gfm/math/wideint....

deadalnix (5/11) Jan 06 2016 That is awesome ! Won't do the 160 bits case, but I can deal with

Andrei Alexandrescu (2/13) Jan 06 2016 Yes, we need to add that to phobos. Who could be the champion? -- Andrei

H. S. Teoh via Digitalmars-d (15/31) Jan 06 2016 Here are a few things that should probably be addressed before merging

Andrei Alexandrescu (2/30) Jan 06 2016 This is a good list, and not difficult to implement. Thanks! -- Andrei

H. S. Teoh via Digitalmars-d (7/10) Jan 08 2016 [...]
H. S. Teoh via Digitalmars-d (7/8) Jan 08 2016 [...]
H. S. Teoh via Digitalmars-d (9/15) Jan 08 2016 [...]

rumbu (16/16) Jan 09 2016 Here is what I've done as a side project for a decimal data type:

Andrei Alexandrescu (2/4) Jan 09 2016 Who can champion ONE fixed large integer library for Phobos? -- Andrei

Era Scarecrow (9/11) Jun 10 2017 Got a possible one. My implementation is heavy on assembly

Andrei Alexandrescu (3/13) Jun 10 2017 That's cool as long as the assembler is guarded by version(X_86) and has...

Era Scarecrow (6/12) Jun 10 2017 My computer, and no portable alternative yet. I've got a ways

deadalnix (4/17) Jun 11 2017 I ended up doing my own. There are just no way to do it well

Era Scarecrow (13/16) Jun 11 2017 And what timing, I just finished getting it working with

Era Scarecrow (3/6) Jun 11 2017 I should probably mention that's only for UCent, Cent isn't
Era Scarecrow (10/14) Jun 11 2017 Doing a 10240bit int I ran though 1,000 factorials in 6 seconds

Era Scarecrow (11/12) Jun 11 2017 It's complete, more or less now. I just a matter of filling in
Era Scarecrow (5/8) Jun 11 2017 Got the compatible/generic code to run far faster, about 7 times

deadalnix (7/24) Jun 12 2017 You misunderstood. We need cent/ucent supported by the compiler

Era Scarecrow (4/10) Jun 12 2017 Agreed. But until larger types are natively avaliable (either by

deadalnix (3/13) Jun 12 2017 I work with SDC on that one. That's the only reasonable path

Era Scarecrow (14/16) Jun 12 2017 Should i do a pull request and push it under std.experimental?

Nicholas Wilson (3/12) Jun 10 2017 You should be able to use the overflow intrinsics to similar

Guillaume Piolat (3/16) Jan 06 2016 Well I've never ever used them after writing. Perhaps should be
Guillaume Piolat (7/8) Jan 06 2016 Now that I think of it, wideint could be used as a base for fixed

H. S. Teoh via Digitalmars-d (9/17) Jan 06 2016 Looks nice. With a bit of work, it could be a potential Phobos addition

deadalnix <deadalnix gmail.com> writes:

In my case, I'm interested in 160 and 256 bits integers. Using 
BigInt seems wasteful as I know the integer size I want. Do we 
have something like this around ?

Jan 06 2016

"H. S. Teoh via Digitalmars-d" <digitalmars-d puremagic.com> writes:

On Wed, Jan 06, 2016 at 09:25:45PM +0000, deadalnix via Digitalmars-d wrote:
 In my case, I'm interested in 160 and 256 bits integers. Using BigInt
 seems wasteful as I know the integer size I want. Do we have something
 like this around ?

If not, it may be something nice to add to Phobos.  Integer width can be
passed as a compile-time parameter, making it completely generic.


T

-- 
Right now I'm having amnesia and deja vu at the same time. I think I've
forgotten this before.

Jan 06 2016

Guillaume Piolat <first.last gmail.com> writes:

On Wednesday, 6 January 2016 at 21:25:45 UTC, deadalnix wrote:
 In my case, I'm interested in 160 and 256 bits integers. Using 
 BigInt seems wasteful as I know the integer size I want. Do we 
 have something like this around ?

https://github.com/d-gamedev-team/gfm/blob/master/math/gfm/math/wideint.d

public domain. Slow division. Only power-of-2 bits supported.

Jan 06 2016

deadalnix <deadalnix gmail.com> writes:

On Wednesday, 6 January 2016 at 21:57:13 UTC, Guillaume Piolat 
wrote:
 On Wednesday, 6 January 2016 at 21:25:45 UTC, deadalnix wrote:
 In my case, I'm interested in 160 and 256 bits integers. Using 
 BigInt seems wasteful as I know the integer size I want. Do we 
 have something like this around ?

 https://github.com/d-gamedev-team/gfm/blob/master/math/gfm/math/wideint.d

 public domain. Slow division. Only power-of-2 bits supported.

That is awesome ! Won't do the 160 bits case, but I can deal with 
that. Did you consider this as a worthy addition to phobos ? I 
think it is.

Jan 06 2016

Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:

On 1/6/16 5:18 PM, deadalnix wrote:
 On Wednesday, 6 January 2016 at 21:57:13 UTC, Guillaume Piolat wrote:
 On Wednesday, 6 January 2016 at 21:25:45 UTC, deadalnix wrote:
 In my case, I'm interested in 160 and 256 bits integers. Using BigInt
 seems wasteful as I know the integer size I want. Do we have
 something like this around ?

 https://github.com/d-gamedev-team/gfm/blob/master/math/gfm/math/wideint.d

 public domain. Slow division. Only power-of-2 bits supported.

 That is awesome ! Won't do the 160 bits case, but I can deal with that.
 Did you consider this as a worthy addition to phobos ? I think it is.

Yes, we need to add that to phobos. Who could be the champion? -- Andrei

Jan 06 2016

"H. S. Teoh via Digitalmars-d" <digitalmars-d puremagic.com> writes:

On Wed, Jan 06, 2016 at 05:23:12PM -0500, Andrei Alexandrescu via Digitalmars-d
wrote:
 On 1/6/16 5:18 PM, deadalnix wrote:
On Wednesday, 6 January 2016 at 21:57:13 UTC, Guillaume Piolat wrote:
On Wednesday, 6 January 2016 at 21:25:45 UTC, deadalnix wrote:
In my case, I'm interested in 160 and 256 bits integers. Using
BigInt seems wasteful as I know the integer size I want. Do we have
something like this around ?

https://github.com/d-gamedev-team/gfm/blob/master/math/gfm/math/wideint.d

public domain. Slow division. Only power-of-2 bits supported.

That is awesome ! Won't do the 160 bits case, but I can deal with
that.  Did you consider this as a worthy addition to phobos ? I think
it is.

 
 Yes, we need to add that to phobos. Who could be the champion? --

Here are a few things that should probably be addressed before merging
to Phobos:

- Better error message when user attempts to instantiate a wide int with
  non-power-of-2 bits. (Currently spews a whole bunch of internal
  compile errors.)

- toString() needs to:
  1) Be the non-allocating overload:

  	void toString(scope void delegate(const(char)[]) dg) {...}

  2) Output in decimal by default instead of hex.

- Need a nice way of initializing wide ints from literals.

- [optional] Division performance can probably be improved.


T

-- 
I don't trust computers, I've spent too long programming to think that they can
get anything right. -- James Miller

Jan 06 2016

Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:

On 01/06/2016 05:47 PM, H. S. Teoh via Digitalmars-d wrote:
 On Wed, Jan 06, 2016 at 05:23:12PM -0500, Andrei Alexandrescu via
Digitalmars-d wrote:
 On 1/6/16 5:18 PM, deadalnix wrote:
 On Wednesday, 6 January 2016 at 21:57:13 UTC, Guillaume Piolat wrote:
 On Wednesday, 6 January 2016 at 21:25:45 UTC, deadalnix wrote:
 In my case, I'm interested in 160 and 256 bits integers. Using
 BigInt seems wasteful as I know the integer size I want. Do we have
 something like this around ?

 https://github.com/d-gamedev-team/gfm/blob/master/math/gfm/math/wideint.d

 public domain. Slow division. Only power-of-2 bits supported.

 That is awesome ! Won't do the 160 bits case, but I can deal with
 that.  Did you consider this as a worthy addition to phobos ? I think
 it is.

 Yes, we need to add that to phobos. Who could be the champion? --

 Here are a few things that should probably be addressed before merging
 to Phobos:

 - Better error message when user attempts to instantiate a wide int with
    non-power-of-2 bits. (Currently spews a whole bunch of internal
    compile errors.)

 - toString() needs to:
    1) Be the non-allocating overload:

    	void toString(scope void delegate(const(char)[]) dg) {...}

    2) Output in decimal by default instead of hex.

 - Need a nice way of initializing wide ints from literals.

 - [optional] Division performance can probably be improved.

This is a good list, and not difficult to implement. Thanks! -- Andrei

Jan 06 2016

"H. S. Teoh via Digitalmars-d" <digitalmars-d puremagic.com> writes:

On Wed, Jan 06, 2016 at 02:47:24PM -0800, H. S. Teoh via Digitalmars-d wrote:
[...]
 - Better error message when user attempts to instantiate a wide int with
   non-power-of-2 bits. (Currently spews a whole bunch of internal
   compile errors.)

[...]

https://github.com/d-gamedev-team/gfm/pull/143


T

-- 
Ph.D. = Permanent head Damage

Jan 08 2016

"H. S. Teoh via Digitalmars-d" <digitalmars-d puremagic.com> writes:

On Wed, Jan 06, 2016 at 02:47:24PM -0800, H. S. Teoh via Digitalmars-d wrote:
[...]
 - Need a nice way of initializing wide ints from literals.

[...]

https://github.com/d-gamedev-team/gfm/pull/144


T

-- 
Try to keep an open mind, but not so open your brain falls out. -- theboz

Jan 08 2016

"H. S. Teoh via Digitalmars-d" <digitalmars-d puremagic.com> writes:

On Wed, Jan 06, 2016 at 02:47:24PM -0800, H. S. Teoh via Digitalmars-d wrote:
[...]
 - toString() needs to:
   1) Be the non-allocating overload:
 
   	void toString(scope void delegate(const(char)[]) dg) {...}
 
   2) Output in decimal by default instead of hex.

[...]

https://github.com/d-gamedev-team/gfm/pull/145


T

-- 
Debugging is twice as hard as writing the code in the first place.
Therefore, if you write the code as cleverly as possible, you are, by
definition, not smart enough to debug it. -- Brian W. Kernighan

Jan 08 2016

rumbu <rumbu rumbu.ro> writes:

Here is what I've done as a side project for a decimal data type:

https://github.com/rumbu13/fixed/blob/master/src/fixedtest/fixed.d

- signed and unsigned fixed width types;
- all current operators supported (even signed shift or power);
- optimized division and multiplication;
- formatting with specifiers;
- conversion (currently 16 and 10-based numbers)
- casting;
- endianess aware;
- minimal dependendencies; (std.traits, std.format - needed for 
FormatSpec).

Working on:
- floating point conversion optimization;
- random base conversion;
- rangify current strings conversions;
- some bugs for the signed data type;

Jan 09 2016

Andrei Alexandrescu <SeeWebsiteForEmail erdani.org> writes:

On 1/9/16 6:31 AM, rumbu wrote:
 Here is what I've done as a side project for a decimal data type:

 https://github.com/rumbu13/fixed/blob/master/src/fixedtest/fixed.d

Who can champion ONE fixed large integer library for Phobos? -- Andrei

Jan 09 2016