Date: Thu, 28 Jul 2005 17:05:19 +0800 From: Erich Dollansky <oceanare@pacific.net.sg> To: ray@redshift.com Cc: freebsd-amd64@freebsd.org Subject: Re: Benchmarks: AMD64 vs i386 on Dual 246 Opteron Message-ID: <42E89FCF.5070307@pacific.net.sg> In-Reply-To: <3.0.1.32.20050728013152.00a4d188@pop.redshift.com> References: <3.0.1.32.20050728013152.00a4d188@pop.redshift.com>
next in thread | previous in thread | raw e-mail | index | archive | help
Hi, ray@redshift.com wrote: > MySQL, re-ran the benchmarks. Much to my surprise, just changing the OS from 64 > bit to 32 bit caused the machine to double in speed. The results are attached I do not wonder at all. I did some software development for PA-RISC and SPARC machines. Naiv as I am, I started with the perfect 64 bit program which resulted in bad performance. I then collected hints of how to mix 8, 16, 32 and 64 bit data to make the program much faster. The TCP/IP based application run then five times faster on the same machine compared to the plain 64 bit program. 64 bit programs copy to many unused information around. It will be very difficult to tune a huge source base to maximum performance and keep it compatible between 32 and 64 bit. The pointers mentioned by you are only a small part of the problem. The main problem is - at least in my program, and as far as what I saw in the sources here - the data handling. Erich
Want to link to this message? Use this URL: <https://mail-archive.FreeBSD.org/cgi/mid.cgi?42E89FCF.5070307>