<?xml version="1.0" encoding="UTF-8"?><rss version="2.0"
	xmlns:content="http://purl.org/rss/1.0/modules/content/"
	xmlns:dc="http://purl.org/dc/elements/1.1/"
	xmlns:atom="http://www.w3.org/2005/Atom"
	xmlns:sy="http://purl.org/rss/1.0/modules/syndication/"
	xmlns:georss="http://www.georss.org/georss" xmlns:geo="http://www.w3.org/2003/01/geo/wgs84_pos#" xmlns:media="http://search.yahoo.com/mrss/"
		>
<channel>
	<title>Comments on: Poor man&#8217;s SSD: Confusion and disappointment</title>
	<atom:link href="http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/feed/" rel="self" type="application/rss+xml" />
	<link>http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/</link>
	<description>K.Mandla's blog of Linux experiences</description>
	<lastBuildDate>Mon, 11 Feb 2013 18:19:59 +0000</lastBuildDate>
	<sy:updatePeriod>hourly</sy:updatePeriod>
	<sy:updateFrequency>1</sy:updateFrequency>
	<generator>http://wordpress.com/</generator>
	<item>
		<title>By: Jose Catre-Vandis</title>
		<link>http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49293</link>
		<dc:creator><![CDATA[Jose Catre-Vandis]]></dc:creator>
		<pubDate>Tue, 10 May 2011 21:33:10 +0000</pubDate>
		<guid isPermaLink="false">http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49293</guid>
		<description><![CDATA[Got fed up with the noise from my 4GB Fujitsu HDD in my ageing Dell Cpt 333GT Latitude I bit the bullet and bought a cf adapter and an 8 gb cf card. The card hasn&#039;t arrived yet, so I threw a 256mb cf card in that I have in my ancient Kodak camera

Booted up Slitaz cooking 3, formatted with gparted and installed slitaz

Rebooted, and oh the silence. Its quick of course as Slitaz is running from ram, when the new card comes I&#039;ll try out my xubuntu respin (1.5gb installed)

Thanks for the tips]]></description>
		<content:encoded><![CDATA[<p>Got fed up with the noise from my 4GB Fujitsu HDD in my ageing Dell Cpt 333GT Latitude I bit the bullet and bought a cf adapter and an 8 gb cf card. The card hasn&#8217;t arrived yet, so I threw a 256mb cf card in that I have in my ancient Kodak camera</p>
<p>Booted up Slitaz cooking 3, formatted with gparted and installed slitaz</p>
<p>Rebooted, and oh the silence. Its quick of course as Slitaz is running from ram, when the new card comes I&#8217;ll try out my xubuntu respin (1.5gb installed)</p>
<p>Thanks for the tips</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: tidux</title>
		<link>http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49238</link>
		<dc:creator><![CDATA[tidux]]></dc:creator>
		<pubDate>Thu, 05 May 2011 14:33:19 +0000</pubDate>
		<guid isPermaLink="false">http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49238</guid>
		<description><![CDATA[I&#039;d recommend SDHC cards with a decent filesystem on them (ext3 or jfs - ext2 if you have a crappy card and want to preserve write cycles) for more modern systems.  They come in sizes up to 32GB, and you can always swap them out like system floppies back in the dark ages if you want to try a different OS.  Linux on one card, Haiku on a second, FreeDOS on a 1B FAT32 card if you need to do BIOS hackery, etc.  Sort of GRUB version -1.]]></description>
		<content:encoded><![CDATA[<p>I&#8217;d recommend SDHC cards with a decent filesystem on them (ext3 or jfs &#8211; ext2 if you have a crappy card and want to preserve write cycles) for more modern systems.  They come in sizes up to 32GB, and you can always swap them out like system floppies back in the dark ages if you want to try a different OS.  Linux on one card, Haiku on a second, FreeDOS on a 1B FAT32 card if you need to do BIOS hackery, etc.  Sort of GRUB version -1.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: ancientforest</title>
		<link>http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49200</link>
		<dc:creator><![CDATA[ancientforest]]></dc:creator>
		<pubDate>Mon, 02 May 2011 19:50:34 +0000</pubDate>
		<guid isPermaLink="false">http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49200</guid>
		<description><![CDATA[I have a 400x speed Transcend card in my 1 GHz ThinkPad. It easily saturates the buss. Cheers.]]></description>
		<content:encoded><![CDATA[<p>I have a 400x speed Transcend card in my 1 GHz ThinkPad. It easily saturates the buss. Cheers.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: George Thomas</title>
		<link>http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49198</link>
		<dc:creator><![CDATA[George Thomas]]></dc:creator>
		<pubDate>Mon, 02 May 2011 16:35:52 +0000</pubDate>
		<guid isPermaLink="false">http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49198</guid>
		<description><![CDATA[I have dual 8GB CF connected through the IDE bus in a Dell Precision M60 notebook.  The results are less than spectacular.

The write cache on these things are in the 1-4 kB range, so you see a high percentage of &quot;Wa&quot; in &#039;top&#039;.  If there were a bigger write cache on the devices, I am sure they would work better as a general purpose hard drive.

The kernel is booted with &quot;elevator=noop&quot;, disks mounted with &quot;noatime&quot;, but I was unable to do much with &#039;hdparm&#039; to attempt to control cache.  Moving and disabling cache for browsers helped some, as did putting tmp and var/log onto tmpfs mount points (2 GB RAM has lots to spare).  

I&#039;d like to move to a compressed file system that reads from CF on boot, caches all changes in RAM, then commits changes back to the CF during shutdown.  I just haven&#039;t put in the effort to follow up on the few articles on the web to set this up.]]></description>
		<content:encoded><![CDATA[<p>I have dual 8GB CF connected through the IDE bus in a Dell Precision M60 notebook.  The results are less than spectacular.</p>
<p>The write cache on these things are in the 1-4 kB range, so you see a high percentage of &#8220;Wa&#8221; in &#8216;top&#8217;.  If there were a bigger write cache on the devices, I am sure they would work better as a general purpose hard drive.</p>
<p>The kernel is booted with &#8220;elevator=noop&#8221;, disks mounted with &#8220;noatime&#8221;, but I was unable to do much with &#8216;hdparm&#8217; to attempt to control cache.  Moving and disabling cache for browsers helped some, as did putting tmp and var/log onto tmpfs mount points (2 GB RAM has lots to spare).  </p>
<p>I&#8217;d like to move to a compressed file system that reads from CF on boot, caches all changes in RAM, then commits changes back to the CF during shutdown.  I just haven&#8217;t put in the effort to follow up on the few articles on the web to set this up.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: mulenmar</title>
		<link>http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49189</link>
		<dc:creator><![CDATA[mulenmar]]></dc:creator>
		<pubDate>Sun, 01 May 2011 22:55:19 +0000</pubDate>
		<guid isPermaLink="false">http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49189</guid>
		<description><![CDATA[The browser has an in-memory cache, and this laptop has only 512 MiB of memory, so that&#039;s as good as I know how to get it.

As for disabling the logs, that requires hunting down everything that generates a log, and I haven&#039;t been successful in that yet. Until I get around to finishing that, forcing them all to write to memory instead of the CF card seems to be good enough.]]></description>
		<content:encoded><![CDATA[<p>The browser has an in-memory cache, and this laptop has only 512 MiB of memory, so that&#8217;s as good as I know how to get it.</p>
<p>As for disabling the logs, that requires hunting down everything that generates a log, and I haven&#8217;t been successful in that yet. Until I get around to finishing that, forcing them all to write to memory instead of the CF card seems to be good enough.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: AgentOss</title>
		<link>http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49185</link>
		<dc:creator><![CDATA[AgentOss]]></dc:creator>
		<pubDate>Sun, 01 May 2011 14:14:39 +0000</pubDate>
		<guid isPermaLink="false">http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49185</guid>
		<description><![CDATA[This website has a lot of info concerning card benchmarks :

http://www.hjreggel.net/cardspeed/index.html


also, the link below 
http://www.thinkwiki.org/wiki/Compact_Flash_boot_drive

suggests that a kernel patch might be necessary to force faster UDMA modes]]></description>
		<content:encoded><![CDATA[<p>This website has a lot of info concerning card benchmarks :</p>
<p><a href="http://www.hjreggel.net/cardspeed/index.html" rel="nofollow">http://www.hjreggel.net/cardspeed/index.html</a></p>
<p>also, the link below<br />
<a href="http://www.thinkwiki.org/wiki/Compact_Flash_boot_drive" rel="nofollow">http://www.thinkwiki.org/wiki/Compact_Flash_boot_drive</a></p>
<p>suggests that a kernel patch might be necessary to force faster UDMA modes</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Bill Frankenhopper</title>
		<link>http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49184</link>
		<dc:creator><![CDATA[Bill Frankenhopper]]></dc:creator>
		<pubDate>Sun, 01 May 2011 12:52:12 +0000</pubDate>
		<guid isPermaLink="false">http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49184</guid>
		<description><![CDATA[two thoughts on this:

- it could be worthwile to disable logging altogether. I mean, how often goes something wrong and how often do you peruse log files anyway? Furthermore, if something really goes wrong and isn&#039;t fixed by rebooting ;) you could then always re-enable logging and go bug hunting.

- you could also put the browser cache into memory. Saves some bandwidth and improves your &#039;browsing experience&#039;... ;)]]></description>
		<content:encoded><![CDATA[<p>two thoughts on this:</p>
<p>- it could be worthwile to disable logging altogether. I mean, how often goes something wrong and how often do you peruse log files anyway? Furthermore, if something really goes wrong and isn&#8217;t fixed by rebooting <img src='http://s1.wp.com/wp-includes/images/smilies/icon_wink.gif' alt=';)' class='wp-smiley' />  you could then always re-enable logging and go bug hunting.</p>
<p>- you could also put the browser cache into memory. Saves some bandwidth and improves your &#8216;browsing experience&#8217;&#8230; <img src='http://s1.wp.com/wp-includes/images/smilies/icon_wink.gif' alt=';)' class='wp-smiley' /> </p>
]]></content:encoded>
	</item>
	<item>
		<title>By: mulenmar</title>
		<link>http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49178</link>
		<dc:creator><![CDATA[mulenmar]]></dc:creator>
		<pubDate>Sun, 01 May 2011 01:08:43 +0000</pubDate>
		<guid isPermaLink="false">http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49178</guid>
		<description><![CDATA[I have these entries in my fstab specifically to avoid as much writing to the CF card as possible -- if the code tags don&#039;t work, please feel free to remove/fix them! :)

[code]
# Mount often-written directories in memory to avoid CF card wear
tmpfs		/tmp		tmpfs	noatime,nosuid		0	0
tmpfs		/var/log	tmpfs	noatime			0	0
tmpfs		/var/run	tmpfs	noatime			0	0
tmpfs		/var/lock	tmpfs	noatime			0	0
tmpfs		/var/tmp	tmpfs	noatime			0	0[/code]

I also disable the disk cache for Firefox AND set it to zero -- I perceived a difference even with the disk cache already disabled.

Finally (of what I can recall off-hand), I use hdparm to force-enable write caching, which doesn&#039;t seem to be enabled by default on CF cards -- they&#039;re removable when not used in IDE adapters, so this is a wise default.

The command to enable it is:

sudo hdparm -W 1 /dev/sdX

Where, of course, X is replaced by the assigned label of the CF card drive.


These tweaks all Work happily with a medium-weight Debian setup on a 600MHz-1.6GHz Pentium M laptop . . . still have occasional writes out to disk that spike CPU usage, but it works much better than without these tweaks.]]></description>
		<content:encoded><![CDATA[<p>I have these entries in my fstab specifically to avoid as much writing to the CF card as possible &#8212; if the code tags don&#8217;t work, please feel free to remove/fix them! <img src='http://s0.wp.com/wp-includes/images/smilies/icon_smile.gif' alt=':)' class='wp-smiley' /> </p>
<pre class="brush: plain; title: ; notranslate">
# Mount often-written directories in memory to avoid CF card wear
tmpfs		/tmp		tmpfs	noatime,nosuid		0	0
tmpfs		/var/log	tmpfs	noatime			0	0
tmpfs		/var/run	tmpfs	noatime			0	0
tmpfs		/var/lock	tmpfs	noatime			0	0
tmpfs		/var/tmp	tmpfs	noatime			0	0</pre>
<p>I also disable the disk cache for Firefox AND set it to zero &#8212; I perceived a difference even with the disk cache already disabled.</p>
<p>Finally (of what I can recall off-hand), I use hdparm to force-enable write caching, which doesn&#8217;t seem to be enabled by default on CF cards &#8212; they&#8217;re removable when not used in IDE adapters, so this is a wise default.</p>
<p>The command to enable it is:</p>
<p>sudo hdparm -W 1 /dev/sdX</p>
<p>Where, of course, X is replaced by the assigned label of the CF card drive.</p>
<p>These tweaks all Work happily with a medium-weight Debian setup on a 600MHz-1.6GHz Pentium M laptop . . . still have occasional writes out to disk that spike CPU usage, but it works much better than without these tweaks.</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: AgentOss</title>
		<link>http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49175</link>
		<dc:creator><![CDATA[AgentOss]]></dc:creator>
		<pubDate>Sat, 30 Apr 2011 21:20:10 +0000</pubDate>
		<guid isPermaLink="false">http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49175</guid>
		<description><![CDATA[Seems that better writing speeds can be attained when partitions area aligned!

https://lwn.net/Articles/428584/

I need to have a closer look on that too, my 8Gb Sandisk can read at 30MB/s but writes at only 3.0MB/S !!! (on a 1.6GHz system)]]></description>
		<content:encoded><![CDATA[<p>Seems that better writing speeds can be attained when partitions area aligned!</p>
<p><a href="https://lwn.net/Articles/428584/" rel="nofollow">https://lwn.net/Articles/428584/</a></p>
<p>I need to have a closer look on that too, my 8Gb Sandisk can read at 30MB/s but writes at only 3.0MB/S !!! (on a 1.6GHz system)</p>
]]></content:encoded>
	</item>
	<item>
		<title>By: Links 30/4/2011: Systemd and a Lot of Ubuntu Coverage &#124; Techrights</title>
		<link>http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49170</link>
		<dc:creator><![CDATA[Links 30/4/2011: Systemd and a Lot of Ubuntu Coverage &#124; Techrights]]></dc:creator>
		<pubDate>Sat, 30 Apr 2011 14:48:12 +0000</pubDate>
		<guid isPermaLink="false">http://kmandla.wordpress.com/2011/04/29/poor-mans-ssd-confusion-and-disappointment/#comment-49170</guid>
		<description><![CDATA[[...] Poor man’s SSD: Confusion and disappointment [...]]]></description>
		<content:encoded><![CDATA[<p>[...] Poor man’s SSD: Confusion and disappointment [...]</p>
]]></content:encoded>
	</item>
</channel>
</rss>
