This is the mail archive of the
ecos-discuss@sourceware.cygnus.com
mailing list for the eCos project.
Re: eCos System Response time
- To: lingsu at palmmicro dot com
- Subject: Re: [ECOS] eCos System Response time
- From: Bart Veer <bartv at redhat dot com>
- Date: Tue, 16 May 2000 14:03:12 +0100
- CC: ecos-discuss at sourceware dot cygnus dot com
- References: <00c401bfbbce$cdbbd780$6500a8c0@PALMICRO>
- Reply-to: bartv at redhat dot com
>>>>> "Ling" == Ling Su <lingsu@palmmicro.com> writes:
Ling> I have a realtime application. Since the application needs a
Ling> hard system response time around 50us, I am not sure if eCos
Ling> could support that with a high preformance
Ling> microproessor(like MIPS 133MHz).
eCos comes with a test program tm_basic that performs some performance
measurements. Details of the tm_basic methodology can be found in the
documentation, e.g.:
http://sourceware.cygnus.com/ecos/docs-1.3.1/guides/user-guides.16.html
Results for various of the supported architectures can be found in the
corresponding Getting Started guides, see
http://sourceware.cygnus.com/ecos/docs-1.3.1/
Looking at one specific set of results, for the JMR3904 board with a
TX39 mips processor running (I believe) at 50MHz, the maximum
interrupt latency was just under 15us. Those results may actually have
been for a 1.2.x release, but I have done another test run with the
current sources and the results are comparable.
So currently we achieve 15us @ 50MHz, suggesting that < 50us @ 133MHz
should not be a problem. That assumes the response time figure you are
after is for interrupt latency. If you are worried about dispatch
latency or DSR latency things get more complicated because then it all
depends on what the various interrupt sources are doing. tm_basic
should provide most of the performance information you are after if
you need to do a more detailed analysis.
Bart Veer // eCos net maintainer