<?xml version="1.0" encoding="UTF-8"?>
<rss xmlns:content="http://purl.org/rss/1.0/modules/content/" xmlns:dc="http://purl.org/dc/elements/1.1/" xmlns:rdf="http://www.w3.org/1999/02/22-rdf-syntax-ns#" xmlns:taxo="http://purl.org/rss/1.0/modules/taxonomy/" version="2.0">
  <channel>
    <title>topic [iMX6Q] Poor SATA Write Performance in i.MX Processors</title>
    <link>https://community.nxp.com/t5/i-MX-Processors/iMX6Q-Poor-SATA-Write-Performance/m-p/451368#M70025</link>
    <description>&lt;HTML&gt;&lt;HEAD&gt;&lt;/HEAD&gt;&lt;BODY&gt;&lt;P&gt;Hello&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;I have an i.MX6Q based SabreSDB running Freescale kernel 3.14.28, alternatively a home-brew vanilla 4.1.0 kernel.&lt;/P&gt;&lt;P&gt;All four cores running at 996MHz, "cpufreq" set to "performance".&lt;/P&gt;&lt;P&gt;Using a SATA-to-CFast adapter and a SATA-II capable CFast card, I tested the SATA write perfomance of the i.MX6Q.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;To simulate a "real-life" situation (with all the system overhead included), I created a shell script which copies the same file (10MiB in size) 20 times (using different target names) from a ramdisk to a 1GiB VFAT partition on the CFast card.&lt;/P&gt;&lt;P&gt;The time is stopped once the final "sync" returns ensuring all the internal buffers are flushed.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;Above test sequence results in a SATA write speed of at most ~50MiB/s.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;I also tried an Intel 330 Series SSD attached directly to the SATA port of the SabreSB port, but the result is more or less the same.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;The same test done on Intel based systems (using vanilla 3.14.28 and 4.1.0 kernels) results in SATA write speeds of up to ~130MiB/s therefore it seems that the i.MX6 itself (hardware and/or software) is the limiting factor.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;What "real-life" SATA write speed could I expect from a i.MX6Q based system?&lt;/P&gt;&lt;P&gt;Are there any bandwidth limitiation in the SATA area of the i.MX6, similiar to the limitation of the FEC?&lt;/P&gt;&lt;P&gt;Does anybody have some comparative values from similiar "real-life" applications?&lt;/P&gt;&lt;P&gt;Or is the ~50 MiB/s already the upper limit for an i.MX6Q based system?&lt;/P&gt;&lt;/BODY&gt;&lt;/HTML&gt;</description>
    <pubDate>Sat, 10 Oct 2015 05:48:50 GMT</pubDate>
    <dc:creator>davidmüller</dc:creator>
    <dc:date>2015-10-10T05:48:50Z</dc:date>
    <item>
      <title>[iMX6Q] Poor SATA Write Performance</title>
      <link>https://community.nxp.com/t5/i-MX-Processors/iMX6Q-Poor-SATA-Write-Performance/m-p/451368#M70025</link>
      <description>&lt;HTML&gt;&lt;HEAD&gt;&lt;/HEAD&gt;&lt;BODY&gt;&lt;P&gt;Hello&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;I have an i.MX6Q based SabreSDB running Freescale kernel 3.14.28, alternatively a home-brew vanilla 4.1.0 kernel.&lt;/P&gt;&lt;P&gt;All four cores running at 996MHz, "cpufreq" set to "performance".&lt;/P&gt;&lt;P&gt;Using a SATA-to-CFast adapter and a SATA-II capable CFast card, I tested the SATA write perfomance of the i.MX6Q.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;To simulate a "real-life" situation (with all the system overhead included), I created a shell script which copies the same file (10MiB in size) 20 times (using different target names) from a ramdisk to a 1GiB VFAT partition on the CFast card.&lt;/P&gt;&lt;P&gt;The time is stopped once the final "sync" returns ensuring all the internal buffers are flushed.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;Above test sequence results in a SATA write speed of at most ~50MiB/s.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;I also tried an Intel 330 Series SSD attached directly to the SATA port of the SabreSB port, but the result is more or less the same.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;The same test done on Intel based systems (using vanilla 3.14.28 and 4.1.0 kernels) results in SATA write speeds of up to ~130MiB/s therefore it seems that the i.MX6 itself (hardware and/or software) is the limiting factor.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;What "real-life" SATA write speed could I expect from a i.MX6Q based system?&lt;/P&gt;&lt;P&gt;Are there any bandwidth limitiation in the SATA area of the i.MX6, similiar to the limitation of the FEC?&lt;/P&gt;&lt;P&gt;Does anybody have some comparative values from similiar "real-life" applications?&lt;/P&gt;&lt;P&gt;Or is the ~50 MiB/s already the upper limit for an i.MX6Q based system?&lt;/P&gt;&lt;/BODY&gt;&lt;/HTML&gt;</description>
      <pubDate>Sat, 10 Oct 2015 05:48:50 GMT</pubDate>
      <guid>https://community.nxp.com/t5/i-MX-Processors/iMX6Q-Poor-SATA-Write-Performance/m-p/451368#M70025</guid>
      <dc:creator>davidmüller</dc:creator>
      <dc:date>2015-10-10T05:48:50Z</dc:date>
    </item>
    <item>
      <title>Re: [iMX6Q] Poor SATA Write Performance</title>
      <link>https://community.nxp.com/t5/i-MX-Processors/iMX6Q-Poor-SATA-Write-Performance/m-p/451369#M70026</link>
      <description>&lt;HTML&gt;&lt;HEAD&gt;&lt;/HEAD&gt;&lt;BODY&gt;&lt;P&gt;Hi David&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;75 MB/s was obtained on ARM2 validation board (ER3 release, DVFS off),&lt;/P&gt;&lt;P&gt;which close to your results.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;Best regards&lt;/P&gt;&lt;P&gt;igor&lt;/P&gt;&lt;P&gt;-----------------------------------------------------------------------------------------------------------------------&lt;/P&gt;&lt;P&gt;Note: If this post answers your question, please click the Correct Answer button. Thank you!&lt;/P&gt;&lt;P&gt;-----------------------------------------------------------------------------------------------------------------------&lt;/P&gt;&lt;/BODY&gt;&lt;/HTML&gt;</description>
      <pubDate>Mon, 12 Oct 2015 03:55:16 GMT</pubDate>
      <guid>https://community.nxp.com/t5/i-MX-Processors/iMX6Q-Poor-SATA-Write-Performance/m-p/451369#M70026</guid>
      <dc:creator>igorpadykov</dc:creator>
      <dc:date>2015-10-12T03:55:16Z</dc:date>
    </item>
    <item>
      <title>Re: [iMX6Q] Poor SATA Write Performance</title>
      <link>https://community.nxp.com/t5/i-MX-Processors/iMX6Q-Poor-SATA-Write-Performance/m-p/451370#M70027</link>
      <description>&lt;HTML&gt;&lt;HEAD&gt;&lt;/HEAD&gt;&lt;BODY&gt;&lt;P&gt;Dear Igor&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;Thank you for your information.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;What puzzles me is the fact the reading from the same CFast card is around twice as fast as writing.&lt;/P&gt;&lt;P&gt;&lt;/P&gt;&lt;P&gt;Do you have an idea what could cause this asymmetry in read/write performance? &lt;/P&gt;&lt;/BODY&gt;&lt;/HTML&gt;</description>
      <pubDate>Wed, 14 Oct 2015 10:49:27 GMT</pubDate>
      <guid>https://community.nxp.com/t5/i-MX-Processors/iMX6Q-Poor-SATA-Write-Performance/m-p/451370#M70027</guid>
      <dc:creator>davidmüller</dc:creator>
      <dc:date>2015-10-14T10:49:27Z</dc:date>
    </item>
  </channel>
</rss>

