Date: Fri, 5 Apr 1996 14:13:11 -0500 (EST) From: Marc Ramirez <mrami@mramirez.sy.yale.edu> To: Satoshi Asami <asami@cs.berkeley.edu> Cc: current@freebsd.org, nisha@cs.berkeley.edu, tege@matematik.su.se, hasty@rah.star-gate.com Subject: Re: fast memory copy for large data sizes Message-ID: <Pine.BSF.3.91.960405141056.2754A-100000@mramirez.sy.yale.edu> In-Reply-To: <199604050935.BAA24263@silvia.HIP.Berkeley.EDU>
next in thread | previous in thread | raw e-mail | index | archive | help
On Fri, 5 Apr 1996, Satoshi Asami wrote:
> We've put together a fast memory copy that uses floating point
> registers to speed up large transfers. The original idea was taken
> from Amancio Hasty's old post to use floating point registers to move
> 8 bytes at a time. (We tried using integer registers too but with our
> wits we could only get 10MB/s less than the FP case.)
P5's and up only, please. :) AMD 486/66:
mrami[~/bcopy]$ make
>> running tests
sh runtests
size libc ours
32 7.629395 MB/s 7.629395 MB/s
64 12.207031 MB/s 4.695012 MB/s
128 12.207031 MB/s 3.487723 MB/s
256 12.207031 MB/s 6.424753 MB/s
512 12.520032 MB/s 7.076540 MB/s
1024 12.682630 MB/s 7.454676 MB/s
2048 12.682630 MB/s 7.570252 MB/s
4096 10.146104 MB/s 7.629395 MB/s
8192 12.682630 MB/s 7.666830 MB/s
16384 12.703252 MB/s 7.689469 MB/s
32768 12.682630 MB/s 7.517440 MB/s
65536 12.183236 MB/s 7.472501 MB/s
131072 12.090144 MB/s 7.526040 MB/s
262144 12.569762 MB/s 7.717717 MB/s
524288 12.419583 MB/s 7.534773 MB/s
1048576 11.838803 MB/s 7.718551 MB/s
2097152 12.164192 MB/s 7.725020 MB/s
4194304 12.290410 MB/s 7.719504 MB/s
mrami[~/bcopy]$
Marc.
--
Every four seconds a woman has a baby. Our problem is to find this
woman and stop her.
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?Pine.BSF.3.91.960405141056.2754A-100000>