Re: [casper] Which chip for ROACH?
I would also suggest the SX95. You're likely to run out of BRAM before anything else in most DSP-heavy CASPER applications and the LX is very low on BRAM resources. In three of the KAT-7 instruments, we run out of BRAM first, then logic and usually have spare DSP48s when using the SX95. The designs won't fit on an LX110 'cos it doesn't have enough BRAM. Jason On 01 Feb 2012, at 05:31, Dan Werthimer wrote: > > hi billy, > > the SX95T is the best Roach I chip for correlators, fft's and beamformers, > because it has a lot of DSP48 multiplier/accumulator blocks. > > the LX110T has more slices but doesn't have as many DSP blocks. > > best wishes, > > dan > > On Tue, Jan 31, 2012 at 6:27 PM, Barott, William Chauncey > wrote: > Hi all, > > > > I'm getting my ducks in a row for a ROACH acquisition. The wiki page lists > two different possible FPGAs (SX and LX series), but most recent discussion > in the threads seems to indicate that the SX95T-1FF1136C (speed grade?) is > the one to get. The most recent discussion on this was some time ago as I > can find in the list archive Has there been any change to this opinion? > Any reason to get a different chip? My applications are general casper-y > type apps - correlators, ffts, beamformers, etc. > > > > Thanks, > > > > Billy > > > > > >
Re: [casper] Which chip for ROACH?
hi billy, the SX95T is the best Roach I chip for correlators, fft's and beamformers, because it has a lot of DSP48 multiplier/accumulator blocks. the LX110T has more slices but doesn't have as many DSP blocks. best wishes, dan On Tue, Jan 31, 2012 at 6:27 PM, Barott, William Chauncey wrote: > Hi all, > > ** ** > > I'm getting my ducks in a row for a ROACH acquisition. The wiki page > lists two different possible FPGAs (SX and LX series), but most recent > discussion in the threads seems to indicate that the SX95T-1FF1136C (speed > grade?) is the one to get. The most recent discussion on this was some > time ago as I can find in the list archive Has there been any change to > this opinion? Any reason to get a different chip? My applications are > general casper-y type apps - correlators, ffts, beamformers, etc. > > ** ** > > Thanks, > > ** ** > > Billy > > ** ** > > ** ** >
[casper] Which chip for ROACH?
Hi all, I'm getting my ducks in a row for a ROACH acquisition. The wiki page lists two different possible FPGAs (SX and LX series), but most recent discussion in the threads seems to indicate that the SX95T-1FF1136C (speed grade?) is the one to get. The most recent discussion on this was some time ago as I can find in the list archive Has there been any change to this opinion? Any reason to get a different chip? My applications are general casper-y type apps - correlators, ffts, beamformers, etc. Thanks, Billy
[casper] Postdoc opportunity for NANOGrav instrumentation / algorithms in Charlottesville, VA
Hi All, Sorry if you get this more than once... This is an ad for a 2-3 year postdoc at NRAO working on NANOGrav related instrumentation and algorithms as funded through the NSF PIRE program. Please forward it to whomever you think might be interested. The due date for applications (with the usual letter, CV, and a summary of current/planned research) is March 1, 2012. You must apply through the NRAO careers portal: https://careers.nrao.edu/applicants/Central?quickFind=50763 Thanks, Scott The National Radio Astronomy Observatory (NRAO) plays a key role in the North American Nanohertz Observatory for Gravitational Waves (NANOGrav), a consortium between US and Canadian astronomers whose aim is to detect low-frequency gravitational waves by timing millisecond pulsars. NANOGrav has recently gained significant momentum from a $6.5M NSF Partnerships for International Research and Education (PIRE) grant to strengthen international collaboration with groups in Europe, India, and Australia. NRAO and NANOGrav invite applicants for a postdoctoral fellowship to work with NRAO Astronomer Scott Ransom in the development of future pulsar observing instrumentation and algorithms. This postdoc position, to be located in Charlottesville, VA, is one of five positions in total provided by the PIRE award. The initial appointment, which can begin as early as spring 2012, will be for two years, with the possibility of extending for a third year (subject to project need, satisfactory performance and availability of funds). To be considered, applicants must have a PhD in astronomy, physics or a related field, or expect to have earned a PhD by the starting date of the appointment. Expertise in digital systems development, radio pulsar observations and analysis, and digital signal processing is highly desirable. Job Duties Summary: The principle responsibility of this position is to develop next-generation high-performance pulsar backends for the GBT and Arecibo, particularly for high-precision pulsar timing applications. The successful applicant will be involved in digital system design and implementation (likely involving CASPER and GPU technologies), software and hardware implementations of new interference and interstellar medium mitigation algorithms, and the development of more advanced data processing algorithms to improve the precision of pulsar timing. Up to 50% of the applicant's time will be available for independent research. -- Scott M. RansomAddress: NRAO Phone: (434) 296-0320 520 Edgemont Rd. email: sran...@nrao.edu Charlottesville, VA 22903 USA GPG Fingerprint: 06A9 9553 78BE 16DB 407B FFCA 9BFA B6FF FFD3 2989
Re: [casper] Correct versions to use?
Thanks to all for the information clarifications! Lots of good stuff going on. John > Hello CASPERites > > SKA-SA has setup an organisation-wide homepage on github for our > open-source work: https://github.com/ska-sa > > We are slowly migrating all our existing projects off the CASPER webserver > and from our own local RCS servers. Github is to become the unified home > for all our associated projects. > > This includes: > >* ROACH-1 and ROACH-2: datasheets, schematics, layouts, gerbers, BOMs, > board support packages, manufacturing packs, testing, infrastructure > and software stuff like tcpborphserver, tgtap, uboot and other bits 'n > pieces of the ROACH system. All of them will be migrated to github with > some in new, standalone repositories. > >* mlib_devel: the CASPER Simulink stuff. There are now a number of > these forks across multiple organisations. SKA-SA has a fairly stable > fork because this is the one we use for production work (KAT-7 and > MeerKAT development). While it doesn't have all the newest Matlab stuff > (xblocks, for example), it generally does have the latest ROACH-related > stuff. So if you've got a new ROACH2 board, you probably want to use > this fork rather than the casper-astro one. New users to the CASPER > tools should use the standard github.com/casper-astro/mlib_devel. > >* CORR (the correlator control package, including katcp_wrapper) is now > on github. > >* SPEAD (the Python SPEAD reference implementation) source code. > Migrated from Simon's personal github account. > >* KATCP: Both python and c-based source code should move to github. > >* There are also some new (read: not yet ready for public consumption; > YMMV) projects on the go too, like the new MyHDL flow. > > The pypi packages will continue to be released periodically for katcp, > spead and corr, as before. Consider these stable releases, which is what > you'll get if you do easy_install katcp, for example. The stuff in github > is under active development and YMMV on the bleeding edge. > > Most of the Wiki pages still point to the now redundant SVN and GIT repos > on the Berkeley servers. As we come across these, we'll fix 'em. Feel free > to help with this if you spot one! For now, you'll have to manually look > in these github repos. To be crystal clear: github is now the > authoritative source for all things ROACH related, not the Berkeley > servers. > > Jason > > > > > > On 31 Jan 2012, at 18:47, G Jones wrote: > >> The casper libraries have been migrated to github. I always find this >> most easily by googling casper-astro github which gives this link: >> https://github.com/casper-astro >> >> Glenn >> >> On Tue, Jan 31, 2012 at 7:56 AM, John Ford wrote: >> Hi all. What is the official blessed version of the casper libraries? >> There is a git repository at: >> >> git clone git://casper.berkeley.edu/git/mlib_devel.git >> >> Is that the right one to use, or is there a later version? >> >> John >> >> >> >> > >
Re: [casper] Correct versions to use?
Hello CASPERites SKA-SA has setup an organisation-wide homepage on github for our open-source work: https://github.com/ska-sa We are slowly migrating all our existing projects off the CASPER webserver and from our own local RCS servers. Github is to become the unified home for all our associated projects. This includes: * ROACH-1 and ROACH-2: datasheets, schematics, layouts, gerbers, BOMs, board support packages, manufacturing packs, testing, infrastructure and software stuff like tcpborphserver, tgtap, uboot and other bits 'n pieces of the ROACH system. All of them will be migrated to github with some in new, standalone repositories. * mlib_devel: the CASPER Simulink stuff. There are now a number of these forks across multiple organisations. SKA-SA has a fairly stable fork because this is the one we use for production work (KAT-7 and MeerKAT development). While it doesn't have all the newest Matlab stuff (xblocks, for example), it generally does have the latest ROACH-related stuff. So if you've got a new ROACH2 board, you probably want to use this fork rather than the casper-astro one. New users to the CASPER tools should use the standard github.com/casper-astro/mlib_devel. * CORR (the correlator control package, including katcp_wrapper) is now on github. * SPEAD (the Python SPEAD reference implementation) source code. Migrated from Simon's personal github account. * KATCP: Both python and c-based source code should move to github. * There are also some new (read: not yet ready for public consumption; YMMV) projects on the go too, like the new MyHDL flow. The pypi packages will continue to be released periodically for katcp, spead and corr, as before. Consider these stable releases, which is what you'll get if you do easy_install katcp, for example. The stuff in github is under active development and YMMV on the bleeding edge. Most of the Wiki pages still point to the now redundant SVN and GIT repos on the Berkeley servers. As we come across these, we'll fix 'em. Feel free to help with this if you spot one! For now, you'll have to manually look in these github repos. To be crystal clear: github is now the authoritative source for all things ROACH related, not the Berkeley servers. Jason On 31 Jan 2012, at 18:47, G Jones wrote: > The casper libraries have been migrated to github. I always find this most > easily by googling casper-astro github which gives this link: > https://github.com/casper-astro > > Glenn > > On Tue, Jan 31, 2012 at 7:56 AM, John Ford wrote: > Hi all. What is the official blessed version of the casper libraries? > There is a git repository at: > > git clone git://casper.berkeley.edu/git/mlib_devel.git > > Is that the right one to use, or is there a later version? > > John > > > >
Re: [casper] Correct versions to use?
The casper libraries have been migrated to github. I always find this most easily by googling casper-astro github which gives this link: https://github.com/casper-astro Glenn On Tue, Jan 31, 2012 at 7:56 AM, John Ford wrote: > Hi all. What is the official blessed version of the casper libraries? > There is a git repository at: > > git clone git://casper.berkeley.edu/git/mlib_devel.git > > Is that the right one to use, or is there a later version? > > John > > > >
Re: [casper] GPIO header/connector for Roach
Thanks George! Nimish On Mon, Jan 30, 2012 at 7:08 PM, George Peck wrote: > Hi Nimish, > > We use a 3M IDC (Insulation Displacement) connector. It is necessary to > also buy a ribbon cable, and crimp the cable on to the connector. A part > number is 3M 89120-0001. > > Perhaps a better solution would be to buy a cable. You can buy the cable > with a female connector at only one end, such as 3m M1AXK-2036R, or you can > buy a cable with female connectors at both ends, such as 3M M3AAK-2036R. > Both cables are 36 inches long, but are available in other lengths. > > George > > > > Nimish Sane wrote: > >> Hi all, >> >> We want to use the GPIO pins available on the Roach boards. Is there any >> recommended or widely used part to connect to the on-board GPIO header? Or, >> could anyone share the product specifics if they are using one? >> >> Thanks, >> >> Nimish >> >
[casper] Correct versions to use?
Hi all. What is the official blessed version of the casper libraries? There is a git repository at: git clone git://casper.berkeley.edu/git/mlib_devel.git Is that the right one to use, or is there a later version? John
