<?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 Resource usage by drivers in Operating System - OpenVMS</title>
    <link>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417951#M65541</link>
    <description>I would like to see what driver is eating how much of the cpu and is doing how much IO. E.g. the amds driver and the mscp driver.&lt;BR /&gt;&lt;BR /&gt;How/where can I find that info ? SDA ?&lt;BR /&gt;&lt;BR /&gt;Have one (Duvel) on me&lt;BR /&gt;&lt;BR /&gt;Wim&lt;BR /&gt;</description>
    <pubDate>Tue, 09 Nov 2004 05:27:57 GMT</pubDate>
    <dc:creator>Wim Van den Wyngaert</dc:creator>
    <dc:date>2004-11-09T05:27:57Z</dc:date>
    <item>
      <title>Resource usage by drivers</title>
      <link>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417951#M65541</link>
      <description>I would like to see what driver is eating how much of the cpu and is doing how much IO. E.g. the amds driver and the mscp driver.&lt;BR /&gt;&lt;BR /&gt;How/where can I find that info ? SDA ?&lt;BR /&gt;&lt;BR /&gt;Have one (Duvel) on me&lt;BR /&gt;&lt;BR /&gt;Wim&lt;BR /&gt;</description>
      <pubDate>Tue, 09 Nov 2004 05:27:57 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417951#M65541</guid>
      <dc:creator>Wim Van den Wyngaert</dc:creator>
      <dc:date>2004-11-09T05:27:57Z</dc:date>
    </item>
    <item>
      <title>Re: Resource usage by drivers</title>
      <link>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417952#M65542</link>
      <description>While this does not answer exactly your question, you have a python program that does cpu stats for all processes at &lt;BR /&gt;&lt;BR /&gt;&lt;A href="http://vmspython.dyndns.org/cgiplus-bin/pythoncolor?script=/demosrc/modes.py" target="_blank"&gt;http://vmspython.dyndns.org/cgiplus-bin/pythoncolor?script=/demosrc/modes.py&lt;/A&gt;&lt;BR /&gt;&lt;BR /&gt;sample output&lt;BR /&gt;&lt;A href="http://vmspython.dyndns.org/demosrc/modes_result.txt" target="_blank"&gt;http://vmspython.dyndns.org/demosrc/modes_result.txt&lt;/A&gt;&lt;BR /&gt;&lt;BR /&gt;In fact, it plays with the number of Cpu ticks for a process (in the pcb)&lt;BR /&gt; &lt;BR /&gt;pcb$l_kernel_counter&lt;BR /&gt;pcb$l_exec_counter&lt;BR /&gt;pcb$l_super_counter&lt;BR /&gt;pcb$l_user_counter&lt;BR /&gt;&lt;BR /&gt;&lt;BR /&gt;and for the I/O see&lt;BR /&gt;&lt;BR /&gt;&lt;A href="http://vmspython.dyndns.org/demosrc/procio_result.txt" target="_blank"&gt;http://vmspython.dyndns.org/demosrc/procio_result.txt&lt;/A&gt;&lt;BR /&gt;and&lt;BR /&gt;&lt;A href="http://vmspython.dyndns.org/cgiplus-bin/pythoncolor?script=/demosrc/procio.py" target="_blank"&gt;http://vmspython.dyndns.org/cgiplus-bin/pythoncolor?script=/demosrc/procio.py&lt;/A&gt;&lt;BR /&gt;&lt;BR /&gt;&lt;BR /&gt;&lt;BR /&gt;&lt;BR /&gt;&lt;BR /&gt;&lt;BR /&gt;&lt;BR /&gt;</description>
      <pubDate>Tue, 09 Nov 2004 06:28:00 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417952#M65542</guid>
      <dc:creator>labadie_1</dc:creator>
      <dc:date>2004-11-09T06:28:00Z</dc:date>
    </item>
    <item>
      <title>Re: Resource usage by drivers</title>
      <link>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417953#M65543</link>
      <description>Gerard,&lt;BR /&gt;&lt;BR /&gt;Nice but this is for interrupt stack usage by processes. AMDS and MSCP (and others) don't have a process. How can you find the usage per driver ?&lt;BR /&gt;&lt;BR /&gt;As far as I understand it, it is simply accounted in the interrupt stack without further details. Is it ?&lt;BR /&gt;&lt;BR /&gt;Btw : are drivers without usage by a process such as AMDS also possible on Unix/Windows ?&lt;BR /&gt;&lt;BR /&gt;Wim</description>
      <pubDate>Tue, 09 Nov 2004 06:40:06 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417953#M65543</guid>
      <dc:creator>Wim Van den Wyngaert</dc:creator>
      <dc:date>2004-11-09T06:40:06Z</dc:date>
    </item>
    <item>
      <title>Re: Resource usage by drivers</title>
      <link>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417954#M65544</link>
      <description>Wim,&lt;BR /&gt;&lt;BR /&gt;As you said: "drivers have no processes".&lt;BR /&gt;And since on Alpha there isn't an interrupt stack anymore (AFAIK), whenever a driver routine needs to do something, the code is executed on the kernel stack of the process that is CURrent (or in absence thereof, on a "system" kernel stack).&lt;BR /&gt;So I don't think you can find out how much of your CPU is eaten by driver routines.&lt;BR /&gt;Maybe some of our friends at engineering can shed some light on this.&lt;BR /&gt;&lt;BR /&gt;Greetz,&lt;BR /&gt;Kris&lt;BR /&gt;</description>
      <pubDate>Tue, 09 Nov 2004 06:51:04 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417954#M65544</guid>
      <dc:creator>Kris Clippeleyr</dc:creator>
      <dc:date>2004-11-09T06:51:04Z</dc:date>
    </item>
    <item>
      <title>Re: Resource usage by drivers</title>
      <link>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417955#M65545</link>
      <description>Wim,&lt;BR /&gt;&lt;BR /&gt;there are no statistics kept inside OpenVMS about 'which driver' consumes how much CPU. CPU statistics are only kept regarding processes and overall CPU modes.&lt;BR /&gt;&lt;BR /&gt;IO statistics are typically kept per device (operations count in UCB) and in some other places (e.g. to be displayed by MONI MSCP).&lt;BR /&gt;&lt;BR /&gt;Some of the performacne data collectors have 'system PC sampling' functionality, which would tell you, where the CPU time will be spent (by execlet/driver and IPL).&lt;BR /&gt;&lt;BR /&gt;More recents versions of OpenVMS (V7.3-2) include (unsupported !) SDA extensions (PCS$SDA = PC sampling and IO$SDA = IO statistics), which may provide some additional data.&lt;BR /&gt;&lt;BR /&gt;So what is the problem ?&lt;BR /&gt;&lt;BR /&gt;Volker.</description>
      <pubDate>Tue, 09 Nov 2004 07:00:42 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417955#M65545</guid>
      <dc:creator>Volker Halle</dc:creator>
      <dc:date>2004-11-09T07:00:42Z</dc:date>
    </item>
    <item>
      <title>Re: Resource usage by drivers</title>
      <link>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417956#M65546</link>
      <description>Volker,&lt;BR /&gt;&lt;BR /&gt;No problem, just thinking about monitoring the interruptstack.&lt;BR /&gt;&lt;BR /&gt;So, if I understand correctly, some drivers keep counters (like MSCP) that can be monitored by dedicated software (monitor mscp). Other drivers don't keep the counters and so there are no statistics available.&lt;BR /&gt;&lt;BR /&gt;Last hope : any statistics on the non-process interrupt stack as a whole ?&lt;BR /&gt;&lt;BR /&gt;Kris,&lt;BR /&gt;&lt;BR /&gt;What do you mean with "And since on Alpha there isn't an interrupt stack anymore" ?&lt;BR /&gt;&lt;BR /&gt;Wim</description>
      <pubDate>Tue, 09 Nov 2004 07:09:56 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417956#M65546</guid>
      <dc:creator>Wim Van den Wyngaert</dc:creator>
      <dc:date>2004-11-09T07:09:56Z</dc:date>
    </item>
    <item>
      <title>Re: Resource usage by drivers</title>
      <link>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417957#M65547</link>
      <description>Wim,&lt;BR /&gt;&lt;BR /&gt;I quote from an old ref.guide (Alpha AXP Architecture Concepts, EY-K725E-SG.0003, p. 7-3):&lt;BR /&gt;&lt;BR /&gt;"There are four processor stacks in OpenVMS: kernel, executive, supervisor, user&lt;BR /&gt;...&lt;BR /&gt;There is no separate interrupt stack or system-wide stack as there was in the VAX architecture."&lt;BR /&gt;&lt;BR /&gt;And AFAIK, when interrupts are to be serviced, they are serviced using the kernel stack of the process being CURrent at that moment, but with IPL raised (from memory, from the Internals courses).&lt;BR /&gt;&lt;BR /&gt;Greetz,&lt;BR /&gt;&lt;BR /&gt;Kris</description>
      <pubDate>Tue, 09 Nov 2004 07:54:20 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417957#M65547</guid>
      <dc:creator>Kris Clippeleyr</dc:creator>
      <dc:date>2004-11-09T07:54:20Z</dc:date>
    </item>
    <item>
      <title>Re: Resource usage by drivers</title>
      <link>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417958#M65548</link>
      <description>Wim,&lt;BR /&gt;&lt;BR /&gt;unlike on VAX, there is no 'interrupt stack' on Alpha. Interrupts are serviced on the kernel stack of the current process.&lt;BR /&gt;&lt;BR /&gt;The INT bit (bit 2 in PS) is set, to indicate that the system is executing in interrupt mode, so that when the timer routine runs, it sees that the code is running in interrupt mode and can increment the appropriate 'mode' counter.&lt;BR /&gt;&lt;BR /&gt;Volker.</description>
      <pubDate>Tue, 09 Nov 2004 07:57:23 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417958#M65548</guid>
      <dc:creator>Volker Halle</dc:creator>
      <dc:date>2004-11-09T07:57:23Z</dc:date>
    </item>
    <item>
      <title>Re: Resource usage by drivers</title>
      <link>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417959#M65549</link>
      <description>Volker was right. PC sampling gives the result. But only for cpu usage.&lt;BR /&gt;&lt;BR /&gt;But most important : it's in DECps/performace advisor.&lt;BR /&gt;&lt;BR /&gt;adv coll sys /begin=/end=/out=wim1.lis&lt;BR /&gt;adv coll rep /out=wim2.lis wim1.lis&lt;BR /&gt;&lt;BR /&gt;You can see which process is doing what but also get an overview of which driver was doing how much.&lt;BR /&gt;&lt;BR /&gt;Wim&lt;BR /&gt;</description>
      <pubDate>Tue, 09 Nov 2004 10:32:02 GMT</pubDate>
      <guid>https://community.hpe.com/t5/operating-system-openvms/resource-usage-by-drivers/m-p/3417959#M65549</guid>
      <dc:creator>Wim Van den Wyngaert</dc:creator>
      <dc:date>2004-11-09T10:32:02Z</dc:date>
    </item>
  </channel>
</rss>

