Re: v6 speed compared to previous versions
On Tue, Nov 14, 2006 at 10:57:01AM +0300, John Smith wrote: On 11/14/06, Kris Kennaway [EMAIL PROTECTED] wrote: When was this corrected? Is 6.2 going to be faster than 6.1? As I said earlier, some of the most important changes that went into 6.0 were fixing performance problems in 5.x. Since then it's been a process of polishing and improving, instead of architectural changes. So 6.2 is expected to be better than 6.1 but probably not measurably faster for general workloads. Did the guys port the libthr changes into 6.2 or not? Dunno what you mean here. Kris I mean the libthr threading library, MySQL works better with it in 6.1 so I hope the last changes done by David Xu is ported to 6.2 Dunno, you can check CVS or ask David. Kris pgpv6QjvlnY6h.pgp Description: PGP signature
Re: v6 speed compared to previous versions
On Mon, Nov 13, 2006 at 11:39:16AM -0800, Jim Pazarena wrote: When I switched to 6.0, then 6.1, it was noticed by most of my clients that my php/mysql/apache system slowed down a fair bit compared to previous version (5.XX). I always like to be on the bleeding edge of FreeBSD, but the performance hit is being commented about by my (few) mysql/php clients. I've seen the trolls of past about speed and previous versions, and I would really be interested to hear the actual truth about 6.XX speed. It is my understanding that 6.XX is more optimized for multi-processors, and that for a single processor, 5.XX (or even 4.XX) outperforms 6.XX. Would someone please outline the choices/drawbacks/concerns of even considering going back a series? 6.x is 5.x with performance bottlenecks fixed. This applies both to UP and SMP systems. Therefore it's pretty surprising that you're seeing a slowdown between 5.x and 6.x, so you should try to look into exactly why your system seems to be running slower. Perhaps it's just a simple misconfiguration, or related to some other change you made at the same time as you updated. Kris pgpMPF3bdJMoS.pgp Description: PGP signature
Re: v6 speed compared to previous versions
Fair to say that those tools should be recompiled on a 6 system to ensure full update-ness? On 11/13/06, Kris Kennaway [EMAIL PROTECTED] wrote: On Mon, Nov 13, 2006 at 11:39:16AM -0800, Jim Pazarena wrote: When I switched to 6.0, then 6.1, it was noticed by most of my clients that my php/mysql/apache system slowed down a fair bit compared to previous version (5.XX). I always like to be on the bleeding edge of FreeBSD, but the performance hit is being commented about by my (few) mysql/php clients. I've seen the trolls of past about speed and previous versions, and I would really be interested to hear the actual truth about 6.XX speed. It is my understanding that 6.XX is more optimized for multi-processors, and that for a single processor, 5.XX (or even 4.XX) outperforms 6.XX. Would someone please outline the choices/drawbacks/concerns of even considering going back a series? 6.x is 5.x with performance bottlenecks fixed. This applies both to UP and SMP systems. Therefore it's pretty surprising that you're seeing a slowdown between 5.x and 6.x, so you should try to look into exactly why your system seems to be running slower. Perhaps it's just a simple misconfiguration, or related to some other change you made at the same time as you updated. Kris ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: v6 speed compared to previous versions
On Mon, Nov 13, 2006 at 01:51:49PM -0800, Jeff Mohler wrote: Fair to say that those tools should be recompiled on a 6 system to ensure full update-ness? Shouldn't really matter, the relevant changes were in the kernel. Kris pgptUG0zqiPwb.pgp Description: PGP signature
Re: v6 speed compared to previous versions
On 11/14/06, Kris Kennaway [EMAIL PROTECTED] wrote: On Mon, Nov 13, 2006 at 01:51:49PM -0800, Jeff Mohler wrote: Fair to say that those tools should be recompiled on a 6 system to ensure full update-ness? Shouldn't really matter, the relevant changes were in the kernel. Kris When was this corrected? Is 6.2 going to be faster than 6.1? Did the guys port the libthr changes into 6.2 or not? Thank you, -J ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
Re: v6 speed compared to previous versions
On Tue, Nov 14, 2006 at 09:27:47AM +0300, John Smith wrote: On 11/14/06, Kris Kennaway [EMAIL PROTECTED] wrote: On Mon, Nov 13, 2006 at 01:51:49PM -0800, Jeff Mohler wrote: Fair to say that those tools should be recompiled on a 6 system to ensure full update-ness? Shouldn't really matter, the relevant changes were in the kernel. Kris When was this corrected? Is 6.2 going to be faster than 6.1? As I said earlier, some of the most important changes that went into 6.0 were fixing performance problems in 5.x. Since then it's been a process of polishing and improving, instead of architectural changes. So 6.2 is expected to be better than 6.1 but probably not measurably faster for general workloads. Did the guys port the libthr changes into 6.2 or not? Dunno what you mean here. Kris pgpep9g5bMpMU.pgp Description: PGP signature
Re: v6 speed compared to previous versions
On 11/14/06, Kris Kennaway [EMAIL PROTECTED] wrote: When was this corrected? Is 6.2 going to be faster than 6.1? As I said earlier, some of the most important changes that went into 6.0 were fixing performance problems in 5.x. Since then it's been a process of polishing and improving, instead of architectural changes. So 6.2 is expected to be better than 6.1 but probably not measurably faster for general workloads. Did the guys port the libthr changes into 6.2 or not? Dunno what you mean here. Kris I mean the libthr threading library, MySQL works better with it in 6.1 so I hope the last changes done by David Xu is ported to 6.2 Thank you, -J ___ freebsd-questions@freebsd.org mailing list http://lists.freebsd.org/mailman/listinfo/freebsd-questions To unsubscribe, send any mail to [EMAIL PROTECTED]
