• BCS (4/4) Sep 06 2010 I assuem that a 64bit DMD building 64bit binaries is one of the objectiv...
• Walter Bright (3/6) Sep 06 2010 There's not much reason to provide a 64 bit binary of dmd. I doubt it wo...
• dsimcha (6/8) Sep 06 2010 How about using a computer that you don't have admin access to that does...
• bearophile (5/7) Sep 06 2010 I don't know. But currently it's not hard for me to reach about 1 GB of ...
• BCS (4/14) Sep 06 2010 I've hit 1GB compileing a single file. It wouldn't be hard to hit 3GB.
• Don (7/21) Sep 07 2010 Yes, but I'm almost certain that's just DMD's poor memory management,

BCS <none anon.com> writes:

I assuem that a 64bit DMD building 64bit binaries is one of the objective 
right now. How about a 64bit->32bit? (Or for that matter, 32bit->64bit.)

-- 
... <IXOYE><

Walter Bright <newshound2 digitalmars.com> writes:

BCS wrote:
 I assuem that a 64bit DMD building 64bit binaries is one of the 
 objective right now. How about a 64bit->32bit? (Or for that matter, 
 32bit->64bit.)

There's not much reason to provide a 64 bit binary of dmd. I doubt it would run 
any faster, and there's no need for a compiler to consume 5 GB of memory.

dsimcha <dsimcha yahoo.com> writes:

== Quote from Walter Bright (newshound2 digitalmars.com)'s article
 There's not much reason to provide a 64 bit binary of dmd.

How about using a computer that you don't have admin access to that doesn't
offer
good support for 32-bit binaries?  Then again, for these cases you could just
compile a 64 binary from source.

 there's no need for a compiler to consume 5 GB of memory.

CTFE/fancy metaprogramming?  Lately DMD seems to use ~500MB of memory to compile
Phobos.

bearophile <bearophileHUGS lycos.com> writes:

Walter Bright:

I doubt it would run any faster,<

I can't know.


and there's no need for a compiler to consume 5 GB of memory.

I don't know. But currently it's not hard for me to reach about 1 GB of used
RAM with template/CTFE-heavy code (once a certain CTFE bug is fixed, the needed
memory probably gets smaller).

Bye,
bearophile

BCS <none anon.com> writes:

Hello Walter,

 BCS wrote:
 
 I assuem that a 64bit DMD building 64bit binaries is one of the
 objective right now. How about a 64bit->32bit? (Or for that matter,
 32bit->64bit.)
 

 There's not much reason to provide a 64 bit binary of dmd. I doubt it
 would run any faster, and there's no need for a compiler to consume 5
 GB of memory.
 

I've hit 1GB compileing a single file. It wouldn't be hard to hit 3GB. 

-- 
... <IXOYE><

Don <nospam nospam.com> writes:

BCS wrote:
 Hello Walter,
 
 BCS wrote:

 I assuem that a 64bit DMD building 64bit binaries is one of the
 objective right now. How about a 64bit->32bit? (Or for that matter,
 32bit->64bit.)

 There's not much reason to provide a 64 bit binary of dmd. I doubt it
 would run any faster, and there's no need for a compiler to consume 5
 GB of memory.

 
 I've hit 1GB compileing a single file. It wouldn't be hard to hit 3GB.

Yes, but I'm almost certain that's just DMD's poor memory management, 
rather than an intrinsic memory-hungry problem. Array literals consume a 
*ridiculous* amount of memory, especially when you start modifying them 
in CTFE.
The solution is to improve the memory management, rather than give the 
compiler another 200GB to waste <g>.