On Mon, 30 Dec 2002, Marvin P. Dickens wrote: > > > THis may all be well & true, however it still doesnt' address the fact > > that disabling swap is not a performance enhancement, but rather a > > performance degradation. > > > > -- > > ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ > > Lonni J Friedman [EMAIL PROTECTED] > > Not so. Here is a quote directly from the "Swap-Mini-Howto": > > "It is now feasible (in the Linux 2.5.40 timeframe) to eliminate
I'm not running 2.5.40. I'd wager that the majority of folks running linux aren't either. If this is still the recommendation when 2.6.0 comes out, then you might have an argument. > swap devices (partitions) completely and not be penalized in > performance." > > Performance gain proof in networking swap vs no swap: > > Results of netperf TCP/IP Stream performance tests on Linux/MVME-162 > and Linux/MVME-167 are found here: > > http://www.esrf.fr/computing/cs/sysadmin/rtk/emlinux/doc/netperf/vmetcpstream.htm " Last update 12 June 1998." I'd hardly call that a reliable reference on modern computing when their reference kernel is 2.0.29. > Performance gain proof as related to I/O. Go to the linux-audio-dev mailing list > and query for "swap" and you'll get tons of info supporting my position. Go to the > kernal (Developer list) mailing list and you'll find tons of info supporting my > position. That's nice. The average user isn't running into memory limitations with tcp/ip streaming or audio. They're running into limitations with bloated PoS software like KDE. I'm not going to even waste time finding the multitude of performance nightmares that are documented on the various KDE mailing lists from folks who keep hitting a wall with insufficient memory, which would be an order of magnitude worse without swap. If youv'e got a reference to documentation that indicates that swap is a performance inhibiter under a _recent_ 2.4.x kernel, then i'll entertain your claim. > Lonnie, This fact is nothing new... It's been a fact since the invention of the > hard drive. How can you make the claim that a swap partition on a hard drive is > faster than 10ns RAM? Swap pages vs RAM is Snail vs Jackrabbit. For the record, > I'll say it one more time: " I know alot of people do not want to run a system > without a swap space." Lonnie, you are one of them which is fine with me. But, > don't mix fact with fiction by claiming that using a swap space is a performance > enhancement. It' an insurance policy for underprovisioned machines. No where did i claim that physical memory wasn't faster than a HD. _YOU_ originally claimed that removing swap was a performance enhancement which is utterly false. Now you're back peddling by providing a single reference to research that is years old (and might as well be centuries when we're talking about the linux kernel), or you provided no references, and just did a lot of hand waving about what is happening on other mailing lists with respect to a bleeding edge _unstable_ kernel. Let's talk about the current stable kernel release, ok? -- ~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ Lonni J Friedman [EMAIL PROTECTED] Linux Step-by-step & TyGeMo http://netllama.ipfox.com _______________________________________________ Linux-users mailing list [EMAIL PROTECTED] Unsubscribe/Suspend/Etc -> http://www.linux-sxs.org/mailman/listinfo/linux-users
