Re: At the edge of LNUM patch

["Bogdan Marinescu" <bogdan.marinescu@...>]

Hi Mike,

Can you please let me know where did you get the files for the benchmarks you mentioned below?

Thanks,
Bogdan

On Thu, Mar 27, 2008 at 3:32 AM, Mike Pall <mikelu-0803@mike.de> wrote:

Miles Bader wrote:
> Asko Kauppi <askok@dnainternet.net> writes:
> > The easiest way to achieve this is to only use the int32 mode. It would
> > boost performance on non-FPU platforms, and cause absolutely no
> > behaviour or accuracy changes on double+int32 configured desktop
> > machines.
>
> Is there any real _advantage_ on typical desktop machines (there's an
> obvious disadvange, which is code bloat)?

Well ... draw your own conclusions:

Here are some benchmarks for Lua 5.1.3 vs. Lua patched with LNUM
260308 (double/int32) on an Intel Core 2 @ 2.13 GHz. Both are
compiled with GCC 4.1.2 -O3 -fomit-frame-pointer.

Benchmark    | LNUM better (+) or worse (-) than Lua
-------------+--------------------------------------
binarytrees  |  -2%
chameneos    |  -3%
cheapconcr   |       LNUM: Lua error
cheapconcw   |       LNUM: Lua stack overflow
fannkuch     |  -4%
fasta        |       LNUM: Segmentation fault
knucleotide  |  -4%
mandelbrot   | -20%
nbody        |       LNUM: Segmentation fault
nsieve       |  -3%
nsievebits   |  -9%
partialsums  | -12%
pidigits     |       LNUM: Segmentation fault
recursive    |  -3%
regexdna     | +-0%
revcomp      |  -8%
spectralnorm |  +1%
sumfile      | +14%  (this mainly measures strtol vs. strtod)
SciMark      |       LNUM: Segmentation fault

--Mike