• Dejan Lekic (15/15) Sep 15 2014 Yesterday I stumbled upon this excellent research paper:
• eles (5/8) Sep 15 2014 Well, the same was believed about the ASM vs C or C++ compilers.
• Paulo Pinto (8/16) Sep 15 2014 Only when exploring vector instructions or when targeting simple

"Dejan Lekic" <dejan.lekic gmail.com> writes:

Yesterday I stumbled upon this excellent research paper:

Reconsidering Custom Memory Allocation - (PDF: 
http://people.cs.umass.edu/~emery/pubs/berger-oopsla2002.pdf )

It got the "Most Influential OOPSLA Paper Award" in 2012.

Abstract (for those lazy to read it):

"Custom memory management is often used in systems software for 
the purpose of decreasing the cost of allocation and tightly 
controlling memory footprint of the software. Until 2002, it was 
taken for granted that application-specific memory allocators 
were superior to general purpose libraries. Berger, Zorn and 
McKinley’s paper demonstrated through a rigorous empirical study 
that this assumption is not well-founded, and gave insights into 
the reasons why general purpose allocators can outperform 
handcrafted ones. The paper also stands out for the quality of 
its empirical methodology."

"eles" <eles eles.com> writes:

On Monday, 15 September 2014 at 09:26:37 UTC, Dejan Lekic wrote:
 Yesterday I stumbled upon this excellent research paper:

 gave insights into the reasons why general purpose allocators 
 can outperform handcrafted ones.

Well, the same was believed about the ASM vs C or C++ compilers. 
This too was infirmed, at least at global scale, with time. Very 
specialized applications would still perform better when in ASM 
(I believe).

"Paulo Pinto" <pjmlp progtools.org> writes:

On Monday, 15 September 2014 at 09:43:41 UTC, eles wrote:
 On Monday, 15 September 2014 at 09:26:37 UTC, Dejan Lekic wrote:
 Yesterday I stumbled upon this excellent research paper:

 gave insights into the reasons why general purpose allocators 
 can outperform handcrafted ones.

 Well, the same was believed about the ASM vs C or C++ 
 compilers. This too was infirmed, at least at global scale, 
 with time. Very specialized applications would still perform 
 better when in ASM (I believe).

Only when exploring vector instructions or when targeting simple 
processors.

I seriously doubt someone can hold the Haswell ISA in their head 
as we did back when targeting the Z80, 6800, 80x86 (up to 
Pentium) and similar.

--
Paulo