June 2017 - Anl-intel-lrz-knr - mailman3.cels.anl.gov

Re: [Anl-intel-csa] 2D FFTs on CSA
by Pope, Adrian C. 29 Jun '17

29 Jun '17

Ok, I see - very cool! Thanks, Adrian On Jun 29, 2017, at 12:31 PM, Hammond, Jeff R <jeff.r.hammond(a)intel.com<mailto:[email protected]>> wrote: Jongsoo is the master of FFTs. I just created a proxy that calls MKL FFTW the same way MiniDFT 1.1 does in order to estimate performance, assuming MKL exists. I mentioned my proxy because it is one way to get a sense of how Quantum Espresso and related codes might perform on a CSA-based system, provided one has an implementation of FFT for CSA already. Jeff From: "Pope, Adrian C." <apope(a)anl.gov<mailto:[email protected]>> Date: Thursday, June 29, 2017 at 10:58 AM To: "Hammond, Jeff R" <jeff.r.hammond(a)intel.com<mailto:[email protected]>> Cc: "Vazquez Mayagoitia, Alvaro" <vama(a)alcf.anl.gov<mailto:[email protected]>>, "anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]>" <anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]>>, Jongsoo Park <jongsoo.park(a)intel.com<mailto:[email protected]>>, christopher j hughes <christopher.j.hughes(a)intel.com<mailto:[email protected]>>, "Petrini, Fabrizio" <fabrizio.petrini(a)intel.com<mailto:[email protected]>> Subject: Re: [Anl-intel-csa] 2D FFTs on CSA Hi Jeff, What’s the relationship between what you’re doing and what Jongsoo Park is working on (eg. $CSASIM_TOP/example/076-fft)? Thanks! Adrian On Jun 29, 2017, at 10:47 AM, Hammond, Jeff R <jeff.r.hammond(a)intel.com<mailto:[email protected]>> wrote: FFTs map very nicely to systolic arrays. There are quite a few papers on this, particularly related to FPGAs. Folks should be using MKL for FFTs, since it should be faster than any other general-purpose implementation out there. If that’s not the case, please let us know. MKL supports the FFTW interface, although this abstraction serves no purpose because MKL already knows the right implementation for each processor, so if you are porting something anyways, you may want to use the MKL FFT interface. Documenting the problem sizes that are important to your applications is the most important thing to do right now. If you are calling FFTs in batches, or the code could be changed to do that, particularly with the same dimensions, that is useful to understand. For example, if you have a plane-wave DFT code that does hundreds of rank-N FFTs (i.e. N is constant), then that means the graph doesn’t need to be reloaded between calls, which will improve the efficiency. For what it’s worth, I created an 3D FFT proxy for MiniDFT 1.1 (does exact-exchange, unlike MiniDFT 1.06) that was small enough to run in Sniper. It’s currently in Fortran but I can convert it for CSA activities. Best, Jeff From: anl-intel-csa <anl-intel-csa-bounces(a)lists.jlse.anl.gov<mailto:[email protected]>> on behalf of "Vazquez Mayagoitia, Alvaro" <vama(a)alcf.anl.gov<mailto:[email protected]>> Date: Wednesday, June 28, 2017 at 3:08 PM To: "Pope, Adrian C." <apope(a)anl.gov<mailto:[email protected]>> Cc: "anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]>" <anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]>> Subject: Re: [Anl-intel-csa] 2D FFTs on CSA Hi, Adrian I added fftpack to the git repo which uses r2r transformations , I haven't gone further since the code is pure fortran 77. It would be good to know if a good replacement could be tested in CSA. We use a lot FFTs in periodic quantum chemistry codes. Thanks - Alvaro Vazquez-Mayagoitia, PhD Theoretical and computational chemistry Argonne National Laboratory https://web.alcf.anl.gov/~vama/ On Jun 28, 2017 1:33 PM, "Pope, Adrian C." <apope(a)anl.gov<mailto:[email protected]>> wrote: Update: Christopher Hughes put me in contact with Jongsoo Park for FFTs on CSA - we’ve been chatting over email, and I’m looping in Venkat Vishwanath since he may have some APS use cases. If anyone else is particularly interested this topic just let me know and I’ll include you in discussions. Cheers, Adrian > On Jun 23, 2017, at 12:16 PM, Pope, Adrian C. <apope(a)anl.gov<mailto:[email protected]>> wrote: > > One of the potential LSST app kernels does many 2D R2R FFTs with sizes ~32x32. They pulled out a simplified version of the code for us and it currently uses FFTW3 interface signatures, so that would be easiest, but we could also adapt to a different interface. I think someone on the call said there was some Intel work on 2D FFT on CSA in the graph(500?) project. If so, could someone send a pointer to that code? > > There may also be another LSST app kernel variant that focuses on 2D FFTs with sizes more like ~4000x4000. I’m still tracking down the details on that, but it would be good to know the size ranges of any 2D FFT work on CSA so far. > > Thanks! > Adrian > > _______________________________________________ > anl-intel-csa mailing list > anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]> > https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa _______________________________________________ anl-intel-csa mailing list anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]> https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa

1 0

Re: [Anl-intel-csa] 2D FFTs on CSA
by Hammond, Jeff R 29 Jun '17

29 Jun '17

Jongsoo is the master of FFTs. I just created a proxy that calls MKL FFTW the same way MiniDFT 1.1 does in order to estimate performance, assuming MKL exists. I mentioned my proxy because it is one way to get a sense of how Quantum Espresso and related codes might perform on a CSA-based system, provided one has an implementation of FFT for CSA already. Jeff From: "Pope, Adrian C." <apope(a)anl.gov> Date: Thursday, June 29, 2017 at 10:58 AM To: "Hammond, Jeff R" <jeff.r.hammond(a)intel.com> Cc: "Vazquez Mayagoitia, Alvaro" <vama(a)alcf.anl.gov>, "anl-intel-csa(a)lists.jlse.anl.gov" <anl-intel-csa(a)lists.jlse.anl.gov>, Jongsoo Park <jongsoo.park(a)intel.com>, christopher j hughes <christopher.j.hughes(a)intel.com>, "Petrini, Fabrizio" <fabrizio.petrini(a)intel.com> Subject: Re: [Anl-intel-csa] 2D FFTs on CSA Hi Jeff, What’s the relationship between what you’re doing and what Jongsoo Park is working on (eg. $CSASIM_TOP/example/076-fft)? Thanks! Adrian On Jun 29, 2017, at 10:47 AM, Hammond, Jeff R <jeff.r.hammond(a)intel.com<mailto:[email protected]>> wrote: FFTs map very nicely to systolic arrays. There are quite a few papers on this, particularly related to FPGAs. Folks should be using MKL for FFTs, since it should be faster than any other general-purpose implementation out there. If that’s not the case, please let us know. MKL supports the FFTW interface, although this abstraction serves no purpose because MKL already knows the right implementation for each processor, so if you are porting something anyways, you may want to use the MKL FFT interface. Documenting the problem sizes that are important to your applications is the most important thing to do right now. If you are calling FFTs in batches, or the code could be changed to do that, particularly with the same dimensions, that is useful to understand. For example, if you have a plane-wave DFT code that does hundreds of rank-N FFTs (i.e. N is constant), then that means the graph doesn’t need to be reloaded between calls, which will improve the efficiency. For what it’s worth, I created an 3D FFT proxy for MiniDFT 1.1 (does exact-exchange, unlike MiniDFT 1.06) that was small enough to run in Sniper. It’s currently in Fortran but I can convert it for CSA activities. Best, Jeff From: anl-intel-csa <anl-intel-csa-bounces(a)lists.jlse.anl.gov<mailto:[email protected]>> on behalf of "Vazquez Mayagoitia, Alvaro" <vama(a)alcf.anl.gov<mailto:[email protected]>> Date: Wednesday, June 28, 2017 at 3:08 PM To: "Pope, Adrian C." <apope(a)anl.gov<mailto:[email protected]>> Cc: "anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]>" <anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]>> Subject: Re: [Anl-intel-csa] 2D FFTs on CSA Hi, Adrian I added fftpack to the git repo which uses r2r transformations , I haven't gone further since the code is pure fortran 77. It would be good to know if a good replacement could be tested in CSA. We use a lot FFTs in periodic quantum chemistry codes. Thanks - Alvaro Vazquez-Mayagoitia, PhD Theoretical and computational chemistry Argonne National Laboratory https://web.alcf.anl.gov/~vama/ On Jun 28, 2017 1:33 PM, "Pope, Adrian C." <apope(a)anl.gov<mailto:[email protected]>> wrote: Update: Christopher Hughes put me in contact with Jongsoo Park for FFTs on CSA - we’ve been chatting over email, and I’m looping in Venkat Vishwanath since he may have some APS use cases. If anyone else is particularly interested this topic just let me know and I’ll include you in discussions. Cheers, Adrian > On Jun 23, 2017, at 12:16 PM, Pope, Adrian C. <apope(a)anl.gov<mailto:[email protected]>> wrote: > > One of the potential LSST app kernels does many 2D R2R FFTs with sizes ~32x32. They pulled out a simplified version of the code for us and it currently uses FFTW3 interface signatures, so that would be easiest, but we could also adapt to a different interface. I think someone on the call said there was some Intel work on 2D FFT on CSA in the graph(500?) project. If so, could someone send a pointer to that code? > > There may also be another LSST app kernel variant that focuses on 2D FFTs with sizes more like ~4000x4000. I’m still tracking down the details on that, but it would be good to know the size ranges of any 2D FFT work on CSA so far. > > Thanks! > Adrian > > _______________________________________________ > anl-intel-csa mailing list > anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]> > https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa _______________________________________________ anl-intel-csa mailing list anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]> https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa

1 0

Re: [Anl-intel-csa] 2D FFTs on CSA
by Pope, Adrian C. 29 Jun '17

29 Jun '17

Hi Jeff, What’s the relationship between what you’re doing and what Jongsoo Park is working on (eg. $CSASIM_TOP/example/076-fft)? Thanks! Adrian On Jun 29, 2017, at 10:47 AM, Hammond, Jeff R <jeff.r.hammond(a)intel.com<mailto:[email protected]>> wrote: FFTs map very nicely to systolic arrays. There are quite a few papers on this, particularly related to FPGAs. Folks should be using MKL for FFTs, since it should be faster than any other general-purpose implementation out there. If that’s not the case, please let us know. MKL supports the FFTW interface, although this abstraction serves no purpose because MKL already knows the right implementation for each processor, so if you are porting something anyways, you may want to use the MKL FFT interface. Documenting the problem sizes that are important to your applications is the most important thing to do right now. If you are calling FFTs in batches, or the code could be changed to do that, particularly with the same dimensions, that is useful to understand. For example, if you have a plane-wave DFT code that does hundreds of rank-N FFTs (i.e. N is constant), then that means the graph doesn’t need to be reloaded between calls, which will improve the efficiency. For what it’s worth, I created an 3D FFT proxy for MiniDFT 1.1 (does exact-exchange, unlike MiniDFT 1.06) that was small enough to run in Sniper. It’s currently in Fortran but I can convert it for CSA activities. Best, Jeff From: anl-intel-csa <anl-intel-csa-bounces(a)lists.jlse.anl.gov<mailto:[email protected]>> on behalf of "Vazquez Mayagoitia, Alvaro" <vama(a)alcf.anl.gov<mailto:[email protected]>> Date: Wednesday, June 28, 2017 at 3:08 PM To: "Pope, Adrian C." <apope(a)anl.gov<mailto:[email protected]>> Cc: "anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]>" <anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]>> Subject: Re: [Anl-intel-csa] 2D FFTs on CSA Hi, Adrian I added fftpack to the git repo which uses r2r transformations , I haven't gone further since the code is pure fortran 77. It would be good to know if a good replacement could be tested in CSA. We use a lot FFTs in periodic quantum chemistry codes. Thanks - Alvaro Vazquez-Mayagoitia, PhD Theoretical and computational chemistry Argonne National Laboratory https://web.alcf.anl.gov/~vama/ On Jun 28, 2017 1:33 PM, "Pope, Adrian C." <apope(a)anl.gov<mailto:[email protected]>> wrote: Update: Christopher Hughes put me in contact with Jongsoo Park for FFTs on CSA - we’ve been chatting over email, and I’m looping in Venkat Vishwanath since he may have some APS use cases. If anyone else is particularly interested this topic just let me know and I’ll include you in discussions. Cheers, Adrian > On Jun 23, 2017, at 12:16 PM, Pope, Adrian C. <apope(a)anl.gov<mailto:[email protected]>> wrote: > > One of the potential LSST app kernels does many 2D R2R FFTs with sizes ~32x32. They pulled out a simplified version of the code for us and it currently uses FFTW3 interface signatures, so that would be easiest, but we could also adapt to a different interface. I think someone on the call said there was some Intel work on 2D FFT on CSA in the graph(500?) project. If so, could someone send a pointer to that code? > > There may also be another LSST app kernel variant that focuses on 2D FFTs with sizes more like ~4000x4000. I’m still tracking down the details on that, but it would be good to know the size ranges of any 2D FFT work on CSA so far. > > Thanks! > Adrian > > _______________________________________________ > anl-intel-csa mailing list > anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]> > https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa _______________________________________________ anl-intel-csa mailing list anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]> https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa

1 0

Re: [Anl-intel-csa] 2D FFTs on CSA
by Glossop, Kent 29 Jun '17

29 Jun '17

> It’s currently in Fortran but I can convert it for CSA activities. Note that you should be able to use the csa-gfortran compiler… From: anl-intel-csa [mailto:[email protected]] On Behalf Of Hammond, Jeff R Sent: Thursday, June 29, 2017 1:47 PM To: Vazquez Mayagoitia, Alvaro <vama(a)alcf.anl.gov>; Pope, Adrian C. <apope(a)anl.gov> Cc: anl-intel-csa(a)lists.jlse.anl.gov Subject: Re: [Anl-intel-csa] 2D FFTs on CSA FFTs map very nicely to systolic arrays. There are quite a few papers on this, particularly related to FPGAs. Folks should be using MKL for FFTs, since it should be faster than any other general-purpose implementation out there. If that’s not the case, please let us know. MKL supports the FFTW interface, although this abstraction serves no purpose because MKL already knows the right implementation for each processor, so if you are porting something anyways, you may want to use the MKL FFT interface. Documenting the problem sizes that are important to your applications is the most important thing to do right now. If you are calling FFTs in batches, or the code could be changed to do that, particularly with the same dimensions, that is useful to understand. For example, if you have a plane-wave DFT code that does hundreds of rank-N FFTs (i.e. N is constant), then that means the graph doesn’t need to be reloaded between calls, which will improve the efficiency. For what it’s worth, I created an 3D FFT proxy for MiniDFT 1.1 (does exact-exchange, unlike MiniDFT 1.06) that was small enough to run in Sniper. It’s currently in Fortran but I can convert it for CSA activities. Best, Jeff From: anl-intel-csa <anl-intel-csa-bounces(a)lists.jlse.anl.gov<mailto:[email protected]>> on behalf of "Vazquez Mayagoitia, Alvaro" <vama(a)alcf.anl.gov<mailto:[email protected]>> Date: Wednesday, June 28, 2017 at 3:08 PM To: "Pope, Adrian C." <apope(a)anl.gov<mailto:[email protected]>> Cc: "anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]>" <anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]>> Subject: Re: [Anl-intel-csa] 2D FFTs on CSA Hi, Adrian I added fftpack to the git repo which uses r2r transformations , I haven't gone further since the code is pure fortran 77. It would be good to know if a good replacement could be tested in CSA. We use a lot FFTs in periodic quantum chemistry codes. Thanks - Alvaro Vazquez-Mayagoitia, PhD Theoretical and computational chemistry Argonne National Laboratory https://web.alcf.anl.gov/~vama/ On Jun 28, 2017 1:33 PM, "Pope, Adrian C." <apope(a)anl.gov<mailto:[email protected]>> wrote: Update: Christopher Hughes put me in contact with Jongsoo Park for FFTs on CSA - we’ve been chatting over email, and I’m looping in Venkat Vishwanath since he may have some APS use cases. If anyone else is particularly interested this topic just let me know and I’ll include you in discussions. Cheers, Adrian > On Jun 23, 2017, at 12:16 PM, Pope, Adrian C. <apope(a)anl.gov<mailto:[email protected]>> wrote: > > One of the potential LSST app kernels does many 2D R2R FFTs with sizes ~32x32. They pulled out a simplified version of the code for us and it currently uses FFTW3 interface signatures, so that would be easiest, but we could also adapt to a different interface. I think someone on the call said there was some Intel work on 2D FFT on CSA in the graph(500?) project. If so, could someone send a pointer to that code? > > There may also be another LSST app kernel variant that focuses on 2D FFTs with sizes more like ~4000x4000. I’m still tracking down the details on that, but it would be good to know the size ranges of any 2D FFT work on CSA so far. > > Thanks! > Adrian > > _______________________________________________ > anl-intel-csa mailing list > anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]> > https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa _______________________________________________ anl-intel-csa mailing list anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]> https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa

1 0

Re: [Anl-intel-csa] 2D FFTs on CSA
by Stevens, Rick L. 29 Jun '17

29 Jun '17

Thanks Jeff It would be great to get a high quality estimate of 2D FFT on csa. Sent from my iPhone On Jun 29, 2017, at 13:47, Hammond, Jeff R <jeff.r.hammond(a)intel.com<mailto:[email protected]>> wrote: FFTs map very nicely to systolic arrays. There are quite a few papers on this, particularly related to FPGAs. Folks should be using MKL for FFTs, since it should be faster than any other general-purpose implementation out there. If that’s not the case, please let us know. MKL supports the FFTW interface, although this abstraction serves no purpose because MKL already knows the right implementation for each processor, so if you are porting something anyways, you may want to use the MKL FFT interface. Documenting the problem sizes that are important to your applications is the most important thing to do right now. If you are calling FFTs in batches, or the code could be changed to do that, particularly with the same dimensions, that is useful to understand. For example, if you have a plane-wave DFT code that does hundreds of rank-N FFTs (i.e. N is constant), then that means the graph doesn’t need to be reloaded between calls, which will improve the efficiency. For what it’s worth, I created an 3D FFT proxy for MiniDFT 1.1 (does exact-exchange, unlike MiniDFT 1.06) that was small enough to run in Sniper. It’s currently in Fortran but I can convert it for CSA activities. Best, Jeff From: anl-intel-csa <anl-intel-csa-bounces(a)lists.jlse.anl.gov<mailto:[email protected]>> on behalf of "Vazquez Mayagoitia, Alvaro" <vama(a)alcf.anl.gov<mailto:[email protected]>> Date: Wednesday, June 28, 2017 at 3:08 PM To: "Pope, Adrian C." <apope(a)anl.gov<mailto:[email protected]>> Cc: "anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]>" <anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]>> Subject: Re: [Anl-intel-csa] 2D FFTs on CSA Hi, Adrian I added fftpack to the git repo which uses r2r transformations , I haven't gone further since the code is pure fortran 77. It would be good to know if a good replacement could be tested in CSA. We use a lot FFTs in periodic quantum chemistry codes. Thanks - Alvaro Vazquez-Mayagoitia, PhD Theoretical and computational chemistry Argonne National Laboratory https://web.alcf.anl.gov/~vama/ On Jun 28, 2017 1:33 PM, "Pope, Adrian C." <apope(a)anl.gov<mailto:[email protected]>> wrote: Update: Christopher Hughes put me in contact with Jongsoo Park for FFTs on CSA - we’ve been chatting over email, and I’m looping in Venkat Vishwanath since he may have some APS use cases. If anyone else is particularly interested this topic just let me know and I’ll include you in discussions. Cheers, Adrian > On Jun 23, 2017, at 12:16 PM, Pope, Adrian C. <apope(a)anl.gov<mailto:[email protected]>> wrote: > > One of the potential LSST app kernels does many 2D R2R FFTs with sizes ~32x32. They pulled out a simplified version of the code for us and it currently uses FFTW3 interface signatures, so that would be easiest, but we could also adapt to a different interface. I think someone on the call said there was some Intel work on 2D FFT on CSA in the graph(500?) project. If so, could someone send a pointer to that code? > > There may also be another LSST app kernel variant that focuses on 2D FFTs with sizes more like ~4000x4000. I’m still tracking down the details on that, but it would be good to know the size ranges of any 2D FFT work on CSA so far. > > Thanks! > Adrian > > _______________________________________________ > anl-intel-csa mailing list > anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]> > https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa _______________________________________________ anl-intel-csa mailing list anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]> https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa _______________________________________________ anl-intel-csa mailing list anl-intel-csa(a)lists.jlse.anl.gov<mailto:[email protected]> https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa

1 0

Re: [Anl-intel-csa] 2D FFTs on CSA
by Hammond, Jeff R 29 Jun '17

29 Jun '17

FFTs map very nicely to systolic arrays. There are quite a few papers on this, particularly related to FPGAs. Folks should be using MKL for FFTs, since it should be faster than any other general-purpose implementation out there. If that’s not the case, please let us know. MKL supports the FFTW interface, although this abstraction serves no purpose because MKL already knows the right implementation for each processor, so if you are porting something anyways, you may want to use the MKL FFT interface. Documenting the problem sizes that are important to your applications is the most important thing to do right now. If you are calling FFTs in batches, or the code could be changed to do that, particularly with the same dimensions, that is useful to understand. For example, if you have a plane-wave DFT code that does hundreds of rank-N FFTs (i.e. N is constant), then that means the graph doesn’t need to be reloaded between calls, which will improve the efficiency. For what it’s worth, I created an 3D FFT proxy for MiniDFT 1.1 (does exact-exchange, unlike MiniDFT 1.06) that was small enough to run in Sniper. It’s currently in Fortran but I can convert it for CSA activities. Best, Jeff From: anl-intel-csa <anl-intel-csa-bounces(a)lists.jlse.anl.gov> on behalf of "Vazquez Mayagoitia, Alvaro" <vama(a)alcf.anl.gov> Date: Wednesday, June 28, 2017 at 3:08 PM To: "Pope, Adrian C." <apope(a)anl.gov> Cc: "anl-intel-csa(a)lists.jlse.anl.gov" <anl-intel-csa(a)lists.jlse.anl.gov> Subject: Re: [Anl-intel-csa] 2D FFTs on CSA Hi, Adrian I added fftpack to the git repo which uses r2r transformations , I haven't gone further since the code is pure fortran 77. It would be good to know if a good replacement could be tested in CSA. We use a lot FFTs in periodic quantum chemistry codes. Thanks - Alvaro Vazquez-Mayagoitia, PhD Theoretical and computational chemistry Argonne National Laboratory https://web.alcf.anl.gov/~vama/ On Jun 28, 2017 1:33 PM, "Pope, Adrian C." <apope(a)anl.gov> wrote: Update: Christopher Hughes put me in contact with Jongsoo Park for FFTs on CSA - we’ve been chatting over email, and I’m looping in Venkat Vishwanath since he may have some APS use cases. If anyone else is particularly interested this topic just let me know and I’ll include you in discussions. Cheers, Adrian > On Jun 23, 2017, at 12:16 PM, Pope, Adrian C. <apope(a)anl.gov> wrote: > > One of the potential LSST app kernels does many 2D R2R FFTs with sizes ~32x32. They pulled out a simplified version of the code for us and it currently uses FFTW3 interface signatures, so that would be easiest, but we could also adapt to a different interface. I think someone on the call said there was some Intel work on 2D FFT on CSA in the graph(500?) project. If so, could someone send a pointer to that code? > > There may also be another LSST app kernel variant that focuses on 2D FFTs with sizes more like ~4000x4000. I’m still tracking down the details on that, but it would be good to know the size ranges of any 2D FFT work on CSA so far. > > Thanks! > Adrian > > _______________________________________________ > anl-intel-csa mailing list > anl-intel-csa(a)lists.jlse.anl.gov > https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa _______________________________________________ anl-intel-csa mailing list anl-intel-csa(a)lists.jlse.anl.gov https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa

1 0

Re: [Anl-intel-csa] BWA extend2 asm version
by Rick Stevens 29 Jun '17

29 Jun '17

Excellent Kaz this is a great example of analysis for example of the Data Pillars.. > On Jun 29, 2017, at 01:23, Kazutomo Yoshii <kazutomo(a)mcs.anl.gov> wrote: > > Hi, > > I've written a CSA asm (hybrid dataflow) version of the innerloop of > BWA's extend2() and its testbench. only tested it with one input data > so far. The asm version is ~6.5 times faster than a compiler-generated > version and uses less resources. > > cycles op_instances > CSA-Compiler 4903 148 > CSA-ASM 759 107 > KNL 7210 3874 -- > E5-2660 1094 -- > > NOTE: > - single thread on CPU > - comparing CSA sim cycles with CPU time-stamp counters > - this is the initial version; there may be some bug or room to improve > - hw_extend2inner.s is a hand-written code > > > To build and run: > > $ git clone /soft/restricted/intel_csa/apps/bwa/bwa.git/ && cd bwa > $ git checkout -t origin/csa > $ source /soft/restricted/intel_csa/csasetup.bash > $ make > $ ./bench_inner testdata/inner-i0.bindat > > To count resource usage: > $ ./csa-op-count.py < extend2inner-hw-csa1.stat > $ ./csa-op-count.py < extend2inner-comp-csa0.stat > > I'll test the loop w/ additional data and update the JLSE wiki > tomorrow. > > - kaz > > _______________________________________________ > anl-intel-csa mailing list > anl-intel-csa(a)lists.jlse.anl.gov > https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa

1 0

[Anl-intel-csa] BWA extend2 asm version
by Kazutomo Yoshii 29 Jun '17

29 Jun '17

Hi, I've written a CSA asm (hybrid dataflow) version of the innerloop of BWA's extend2() and its testbench. only tested it with one input data so far. The asm version is ~6.5 times faster than a compiler-generated version and uses less resources. cycles op_instances CSA-Compiler 4903 148 CSA-ASM 759 107 KNL 7210 3874 -- E5-2660 1094 -- NOTE: - single thread on CPU - comparing CSA sim cycles with CPU time-stamp counters - this is the initial version; there may be some bug or room to improve - hw_extend2inner.s is a hand-written code To build and run: $ git clone /soft/restricted/intel_csa/apps/bwa/bwa.git/ && cd bwa $ git checkout -t origin/csa $ source /soft/restricted/intel_csa/csasetup.bash $ make $ ./bench_inner testdata/inner-i0.bindat To count resource usage: $ ./csa-op-count.py < extend2inner-hw-csa1.stat $ ./csa-op-count.py < extend2inner-comp-csa0.stat I'll test the loop w/ additional data and update the JLSE wiki tomorrow. - kaz

1 0

Re: [Anl-intel-csa] 2D FFTs on CSA
by Vazquez Mayagoitia, Alvaro 28 Jun '17

28 Jun '17

Hi, Adrian I added fftpack to the git repo which uses r2r transformations , I haven't gone further since the code is pure fortran 77. It would be good to know if a good replacement could be tested in CSA. We use a lot FFTs in periodic quantum chemistry codes. Thanks - Alvaro Vazquez-Mayagoitia, PhD Theoretical and computational chemistry Argonne National Laboratory https://web.alcf.anl.gov/~vama/ On Jun 28, 2017 1:33 PM, "Pope, Adrian C." <apope(a)anl.gov> wrote: Update: Christopher Hughes put me in contact with Jongsoo Park for FFTs on CSA - we’ve been chatting over email, and I’m looping in Venkat Vishwanath since he may have some APS use cases. If anyone else is particularly interested this topic just let me know and I’ll include you in discussions. Cheers, Adrian > On Jun 23, 2017, at 12:16 PM, Pope, Adrian C. <apope(a)anl.gov> wrote: > > One of the potential LSST app kernels does many 2D R2R FFTs with sizes ~32x32. They pulled out a simplified version of the code for us and it currently uses FFTW3 interface signatures, so that would be easiest, but we could also adapt to a different interface. I think someone on the call said there was some Intel work on 2D FFT on CSA in the graph(500?) project. If so, could someone send a pointer to that code? > > There may also be another LSST app kernel variant that focuses on 2D FFTs with sizes more like ~4000x4000. I’m still tracking down the details on that, but it would be good to know the size ranges of any 2D FFT work on CSA so far. > > Thanks! > Adrian > > _______________________________________________ > anl-intel-csa mailing list > anl-intel-csa(a)lists.jlse.anl.gov > https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa _______________________________________________ anl-intel-csa mailing list anl-intel-csa(a)lists.jlse.anl.gov https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa

1 0

Re: [Anl-intel-csa] 2D FFTs on CSA
by Pope, Adrian C. 28 Jun '17

28 Jun '17

Update: Christopher Hughes put me in contact with Jongsoo Park for FFTs on CSA - we’ve been chatting over email, and I’m looping in Venkat Vishwanath since he may have some APS use cases. If anyone else is particularly interested this topic just let me know and I’ll include you in discussions. Cheers, Adrian > On Jun 23, 2017, at 12:16 PM, Pope, Adrian C. <apope(a)anl.gov> wrote: > > One of the potential LSST app kernels does many 2D R2R FFTs with sizes ~32x32. They pulled out a simplified version of the code for us and it currently uses FFTW3 interface signatures, so that would be easiest, but we could also adapt to a different interface. I think someone on the call said there was some Intel work on 2D FFT on CSA in the graph(500?) project. If so, could someone send a pointer to that code? > > There may also be another LSST app kernel variant that focuses on 2D FFTs with sizes more like ~4000x4000. I’m still tracking down the details on that, but it would be good to know the size ranges of any 2D FFT work on CSA so far. > > Thanks! > Adrian > > _______________________________________________ > anl-intel-csa mailing list > anl-intel-csa(a)lists.jlse.anl.gov > https://lists.jlse.anl.gov/mailman/listinfo/anl-intel-csa

1 0