On Wed, Aug 17, 2011 at 7:44 PM, Maciej Marcin Piechotka wrote:

I posted bug for various similar instructions some time ago:

http://hackage.haskell.org/trac/ghc/ticket/4102

Probably the update should be done in batch (i.e. popCount, trailing/leading zeros etc.)

I'd be happy to see these as well. If someone would like to do the GHC work needed to have them translate into single machine instructions where available, have a look at the implementation of the popCnt# primops here: https://github.com/ghc/ghc/commit/2d0438f329ac153f9e59155f405d27fac0c43d65 https://github.com/ghc/packages-ghc-prim/commit/cefc19afafe5107ff98d5205c204... https://github.com/ghc/testsuite/commit/2f4d13348e2140f7fc0d4b8b995a2be0fa62... I don't have time to do so at the moment. I'd prefer to not wait with adding popCount until this is done, unless someone takes it on them to add the others before GHC 7.4. -- Johan

First, +1 for a popCount primitive On Wednesday 17 August 2011, 22:33:29, Johan Tibell wrote:

On Wed, Aug 17, 2011 at 7:44 PM, Maciej Marcin Piechotka

wrote:

...

I posted bug for various similar instructions some time ago:

http://hackage.haskell.org/trac/ghc/ticket/4102

Probably the update should be done in batch (i.e. popCount, trailing/leading zeros etc.)

Very desirable to have these too.

I'd be happy to see these as well. If someone would like to do the GHC work needed to have them translate into single machine instructions where available, have a look at the implementation of the popCnt# primops here:

I intend to see whether I can learn from those how to do it, but

https://github.com/ghc/ghc/commit/2d0438f329ac153f9e59155f405d27fac0c43 d65 https://github.com/ghc/packages-ghc-prim/commit/cefc19afafe5107ff98d520 5c204b190da1d497b https://github.com/ghc/testsuite/commit/2f4d13348e2140f7fc0d4b8b995a2be 0fa6291f0

I don't have time to do so at the moment.

nor have I, no time for hacking before September :(

I'd prefer to not wait with adding popCount until this is done, unless someone takes it on them to add the others before GHC 7.4.

No, don't wait. While it would be nice to have the others equally soon, it's better to have one goody than none. Cheers, Daniel

On Thu, Aug 18, 2011 at 3:13 PM, Ryan Newton wrote:

+1

...

...

I'd prefer to not wait with adding popCount until this is done, unless someone takes it on them to add the others before GHC 7.4.

No, don't wait. While it would be nice to have the others equally soon, it's better to have one goody than none.

How about making the API change in one go -- i.e. add the default implementations for all of them, which is easy.  Then people can come along and do the GHC work necessary to make the others efficient later. In fact, changing the API could be a "forcing function" to make the others happen ;-).

If someone wants to do the work I'm happy to review the patches.