www.digitalmars.com         C & C++   DMDScript  

digitalmars.D - dmd omit framepointer

reply Benjamin Thaut <code benjamin-thaut.de> writes:
Does dmd omit framepointers in release buidls? (on windows)
If yes is there a way to turn it off?

-- 
Kind Regards
Benjamin Thaut
Sep 30 2011
parent reply "Vladimir Panteleev" <vladimir thecybershadow.net> writes:
On Fri, 30 Sep 2011 19:51:46 +0300, Benjamin Thaut  
<code benjamin-thaut.de> wrote:

 Does dmd omit framepointers in release buidls? (on windows)
Yes.
 If yes is there a way to turn it off?
I think the only option is not to use -release. (For the record, I also don't think that putting stack frame generation and disabling of asserts/etc. behind one option was a good idea.) -- Best regards, Vladimir mailto:vladimir thecybershadow.net
Sep 30 2011
next sibling parent reply Benjamin Thaut <code benjamin-thaut.de> writes:
Am 30.09.2011 19:22, schrieb Vladimir Panteleev:
 On Fri, 30 Sep 2011 19:51:46 +0300, Benjamin Thaut
 <code benjamin-thaut.de> wrote:

 Does dmd omit framepointers in release buidls? (on windows)
Yes.
 If yes is there a way to turn it off?
I think the only option is not to use -release. (For the record, I also don't think that putting stack frame generation and disabling of asserts/etc. behind one option was a good idea.)
That is really unfortunate. I would like to profile my application obviously in release build. Because of the omitted framepointer the profiler can often not walk the callstack correctly and this results in wrong profiling output. Not building with -release would not really result in representable results when profiling. -- Kind Regards Benjamin Thaut
Sep 30 2011
parent reply "Martin Nowak" <dawg dawgfoto.de> writes:
On Fri, 30 Sep 2011 19:48:18 +0200, Benjamin Thaut  
<code benjamin-thaut.de> wrote:

 Am 30.09.2011 19:22, schrieb Vladimir Panteleev:
 On Fri, 30 Sep 2011 19:51:46 +0300, Benjamin Thaut
 <code benjamin-thaut.de> wrote:

 Does dmd omit framepointers in release buidls? (on windows)
Yes.
 If yes is there a way to turn it off?
I think the only option is not to use -release. (For the record, I also don't think that putting stack frame generation and disabling of asserts/etc. behind one option was a good idea.)
That is really unfortunate. I would like to profile my application obviously in release build. Because of the omitted framepointer the profiler can often not walk the callstack correctly and this results in wrong profiling output. Not building with -release would not really result in representable results when profiling.
You should try the good built-in profiler. It's an instrumenting one and besides a severe slowdown in certain situations the results are way preciser than those from a sampling profiler. martin
Sep 30 2011
parent Benjamin Thaut <code benjamin-thaut.de> writes:
Am 30.09.2011 20:23, schrieb Martin Nowak:
 You should try the good built-in profiler.
 It's an instrumenting one and besides a severe slowdown in certain
 situations the results are way preciser than those from a sampling
 profiler.

 martin
As I want to profile a realtime 3d game the builtin profiler is not usable for me, as it slows my game down to unplayable framerates. Using a sampling profiler however is possible without a gamebreaking performance impact. -- Kind Regards Benjamin Thaut
Sep 30 2011
prev sibling parent reply Walter Bright <newshound2 digitalmars.com> writes:
On 9/30/2011 10:22 AM, Vladimir Panteleev wrote:
 (For the record, I also don't think that putting stack frame generation and
 disabling of asserts/etc. behind one option was a good idea.)
Please add an enhancement request to Bugzilla.
Sep 30 2011
parent "Vladimir Panteleev" <vladimir thecybershadow.net> writes:
On Fri, 30 Sep 2011 21:52:02 +0300, Walter Bright  
<newshound2 digitalmars.com> wrote:

 On 9/30/2011 10:22 AM, Vladimir Panteleev wrote:
 (For the record, I also don't think that putting stack frame generation  
 and
 disabling of asserts/etc. behind one option was a good idea.)
Please add an enhancement request to Bugzilla.
Done: http://d.puremagic.com/issues/show_bug.cgi?id=6752 -- Best regards, Vladimir mailto:vladimir thecybershadow.net
Oct 01 2011