• trevorparscal (10/10) Mar 04 2008 I've read everything I can about proposals and ideas for adding some sor...
• Walter Bright (4/16) Mar 04 2008 There always seems to be something more important. It's also a difficult...
• trevorparscal (4/4) Mar 04 2008 Thank you for the clarity.
• Robert Fraser (2/8) Mar 04 2008 What's GDC's status on SIMD?
• downs (6/16) Mar 05 2008 Autovectorization, which GCC supports in principle, is broken on GDC.

trevorparscal <trevorparscal hotmail.com> writes:

I've read everything I can about proposals and ideas for adding some sort of
SIMD capability to D, and I have not seen any conclusion. I am NOT trying to
revamp the discussion of how or why - only query the status of 2 things...

1. Currently there's no way to align data on the stack, making movaps
difficult/impossible to use properly. Is there where D still stands?

2. Has there been any consensus on if or how Walter may or may not integrate
SIMD capability into the language?

Obviously we have all seen the many (some very convincing) proposals,
but whats the status?

- Trevor

Walter Bright <newshound1 digitalmars.com> writes:

trevorparscal wrote:
 I've read everything I can about proposals and ideas for adding some sort of
 SIMD capability to D, and I have not seen any conclusion. I am NOT trying to
 revamp the discussion of how or why - only query the status of 2 things...
 
 1. Currently there's no way to align data on the stack, making movaps
 difficult/impossible to use properly. Is there where D still stands?

Yes.

 2. Has there been any consensus on if or how Walter may or may not integrate
 SIMD capability into the language?

There always seems to be something more important. It's also a difficult 
thing to implement.

 Obviously we have all seen the many (some very convincing) proposals,
 but whats the status?
 
 - Trevor

trevorparscal <trevorparscal hotmail.com> writes:

Thank you for the clarity.

D is wonderful - I can live with the current lack of SIMD, but look forward to
future development in the area.

- Trevor

Robert Fraser <fraserofthenight gmail.com> writes:

trevorparscal wrote:
 Thank you for the clarity.
 
 D is wonderful - I can live with the current lack of SIMD, but look forward to
 future development in the area.
 
 - Trevor

What's GDC's status on SIMD?

downs <default_357-line yahoo.de> writes:

Robert Fraser wrote:
 trevorparscal wrote:
 Thank you for the clarity.

 D is wonderful - I can live with the current lack of SIMD, but look
 forward to
 future development in the area.

 - Trevor

 
 What's GDC's status on SIMD?

Autovectorization, which GCC supports in principle, is broken on GDC.
David has said, back when I brought it up in d.gnu, that he'd like to fix it
but it's not a priority issue.
So, little hope :)

asm and SSE is of course supported.

 --downs