DragonFly kernel List (threaded) for 2007-06
DragonFly BSD
DragonFly kernel List (threaded) for 2007-06
[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]

Re: implemented features (Re: Decision time....)

From: Erik Wikström <erik-wikstrom@xxxxxxxxx>
Date: Wed, 06 Jun 2007 15:08:20 +0200

On 2007-06-06 14:30, km b wrote:
On 04 Jun 2007 11:58:18 GMT, Rahul Siddharthan <rsidd120@gmail.com> wrote:

I made no claim about "all" binaries, and I claimed the exact opposite
of what you say I claimed for my own binaries: I said the 32-bit
binaries are 50% slower than the 64-bit binaries.  (Eg, running time
25 secs for 64-bit, 38 secs for 32-bit.)

i did some homework for you and conducted the FFT benchmark (double precision only, single precision disabled) of the specific library you mention (GSL) using fftbench http://www.fftw.org/benchfft/ in two different execution environments on Intel Core 2 Duo 1.6G machine

1. pure 64-bit - FreeBSD , CFLAGS="-m64 -march=nocono -O3 -fomit-frame-pointer"

2. 32-bit - DragonFlyBSD, CFLAGS="-march=prescott -O3 -fomit-frame-pointer"

And here I am attaching the results for you to study. Let me know
where do you see the 50 % performance increase?

The place where he found the 50 % performance increase is indicated right in the text you quoted, in *his* binaries compiled from *his* code. He also did mention that this seemed to be true for *some* parts of the GSL, not all of it. So unless you can get hold of the same program as he is talking about and showing that there's no difference between 64-bit and 32-bit machines then there's nothing you can do to prove him wrong.

Regarding your tests I find it hard to put any faith into number derived from running different tests (32 vs. 64 bits) on different machines (is it even the same hardware?). By the way, the files you sent were not PDFs, but PS-files (took me a while to figure that one out).

Erik Wikström

[Date Prev][Date Next]  [Thread Prev][Thread Next]  [Date Index][Thread Index]