It would be also very interesting if you would try to port optimized SETI builds (like AKv8b) to ARM. If you interesting maybe you could sign in Lunatics.kwsn.net community ?
----- Original Message ----- From: Jeff Eastlack To: Bernd Machenschalk ; robert miles ; Raistmer Cc: [email protected] Sent: Tuesday, July 05, 2011 11:19 PM Subject: Re: [boinc_dev] ARM project port Hi All, Thanks for the replys! I am using a ARM ported version of Ubuntu 10.10 running on a Cortex-A9. The community has done a good job with porting many packages to the ARM ISA. FFTW was easily installed on my system using apt-get. I think the end goal to this effort is to understand the limitations of mobile chips (in particular the memory hierarchy), and then work on an Android port with BOINC and the other applications. The later versions of Android support native C/C++ compiled code vs. Java on the older releases. This allows us to exploit HW accelerators such as the Vector floating point unit. So right now I am trying to profile a few other applications to get a better mix. Sensitive economics dictate the features of the embedded processor so things like Cache memory is a very expensive commodity. However the ARM Cortex-A9 has a L2 size range up to 4MB which is very respectable and cache coherency to up to 4 cores, but most licences don't include that much. Right now it appears with SETI that 512k L2 is having issues with locality but I still need to analyze and experiment further by running on a system that has a larger cache and extracting data from performance counters on things like L2 misses, etc. I can send out some slide decks on general performance and comparisons to x86 if anyone is interested? Thanks, Jeff --- On Tue, 7/5/11, Raistmer <[email protected]> wrote: From: Raistmer <[email protected]> Subject: Re: [boinc_dev] ARM project port To: "Bernd Machenschalk" <[email protected]>, "robert miles" <[email protected]> Cc: [email protected] Date: Tuesday, July 5, 2011, 1:23 PM SETI requires FFTW or similar FFT library too. Original poster claimed he looks on SETI@home already. Interesting, what about FFT indeed? ----- Original Message ----- From: Bernd Machenschalk To: robert miles Cc: [email protected] Sent: Monday, July 04, 2011 6:12 PM Subject: Re: [boinc_dev] ARM project port On 04.07.11 07:37, robert miles wrote: > I've found a reference to where to get the source code for Einstein@Home > applications: > > http://einstein.phys.uwm.edu/license.php This requires compiling gsl, fftw and the whole lalsuite for ARM, which I suspect to be a pain to get working. Also too, the resulting application (gravitational wave search) requires a significant amount of disk space to run, and the tasks would (initially) cause quite some download traffic. I don't think that this application is well suited for mobile devices. Is there a reasonably fast FFT(W) implementation for ARM? We might be able to find some different application. Best, Bernd _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address. _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address. _______________________________________________ boinc_dev mailing list [email protected] http://lists.ssl.berkeley.edu/mailman/listinfo/boinc_dev To unsubscribe, visit the above URL and (near bottom of page) enter your email address.
