[m-rev.] for review: conditional structure reuse for LLDS backends

[Peter Wang]

On 2008-02-11, Peter Wang <novalazy at gmail.com> wrote:
> On 2008-02-11, Julien Fischer <juliensf at csse.unimelb.edu.au> wrote:
> >
> > On Wed, 30 Jan 2008, Peter Wang wrote:
> >
> >> [Pending bootchecks in more grades.]
> >
> > Could you please post some benchmarks between conditional structure
> > reuse / structure reuse without static data / no structure reuse,
> > at some point.
> 
> Will do.

Here are some speed_test results from neptune.  The compiler built with
ctgc and intermodule optimisation doesn't work, so I doubt much reuse is
actually happening.

% mmc --grade asm_fast.gc ...

EXTRA_MCFLAGS = 
EXTRA_CFLAGS = 
mercury_compile.01 average of 6 with ignore=1     23.91 (100.00%)

EXTRA_MCFLAGS = --structure-reuse --no-use-atomic-cells
EXTRA_CFLAGS = 
mercury_compile.02 average of 6 with ignore=1     23.83 ( 99.66%)

EXTRA_MCFLAGS = --structure-reuse --no-static-ground-terms --no-use-atomic-cells
EXTRA_CFLAGS = -DMR_UNCONDITIONAL_STRUCTURE_REUSE
mercury_compile.03 average of 6 with ignore=1     26.08 (109.08%)

% mmc --grade hlc.gc ...

EXTRA_MCFLAGS = 
EXTRA_CFLAGS = 
mercury_compile.01 average of 6 with ignore=1     22.35 (100.01%)

EXTRA_MCFLAGS = --structure-reuse --no-use-atomic-cells
EXTRA_CFLAGS = 
mercury_compile.02 average of 6 with ignore=1     22.37 (100.10%)

EXTRA_MCFLAGS = --structure-reuse --no-static-ground-terms --no-use-atomic-cells
EXTRA_CFLAGS = -DMR_UNCONDITIONAL_STRUCTURE_REUSE
mercury_compile.03 average of 6 with ignore=1     24.04 (107.54%)

--------------------------------------------------------------------------
mercury-reviews mailing list
Post messages to:       mercury-reviews at csse.unimelb.edu.au
Administrative Queries: owner-mercury-reviews at csse.unimelb.edu.au
Subscriptions:          mercury-reviews-request at csse.unimelb.edu.au
--------------------------------------------------------------------------