From owner-chemistry |-at-| ccl.net Fri Jun 11 13:01:00 2021 From: "Marcos Verissimo Alves marcos_verissimo+/-id.uff.br" To: CCL Subject: CCL: QM codes NOT written in Fortran Message-Id: <-54393-210611113237-31536-HJYa5pDBDK7hI7zDaoB8aA-.-server.ccl.net> X-Original-From: Marcos Verissimo Alves Content-Type: multipart/alternative; boundary="000000000000323fd905c47f3747" Date: Fri, 11 Jun 2021 12:31:53 -0300 MIME-Version: 1.0 Sent to CCL by: Marcos Verissimo Alves [marcos_verissimo,,id.uff.br] --000000000000323fd905c47f3747 Content-Type: text/plain; charset="UTF-8" Content-Transfer-Encoding: quoted-printable Hi everyone, Also, remember that there is a free version of the PGI compiler with CUDA, the PGI Community Edition: https://developer.nvidia.com/openacc-toolkit I'm not too sure about its efficiency, but I suppose it would be comparable to GCC, which is very good indeed. If you need to compile or develop a GPU-accelerated code for NVIDIA GPUs, it's certainly something to try out. Best, Marcos --- Prof. Dr. Marcos Verissimo Alves Prof. Adjunto II, Curso de F=C3=ADsica Computacional Instituto de Ci=C3=AAncias Exatas Universidade Federal Fluminense Volta Redonda - RJ, Brasil Em sex., 11 de jun. de 2021 =C3=A0s 12:08, Joseph Leonard jleonard42||gmail= .com < owner-chemistry:ccl.net> escreveu: > Phil, glad to hear that you had success with gfortran on that platform. > It=E2=80=99s been a long time sopince I worked in fortran and those doing= it more > recently had a very dim view of that suite. I will pass it along. > > Joe > > On Thu, Jun 10, 2021 at 10:36 PM Phil Hasnip phil.hasnip]_[york.ac.uk < > owner-chemistry%%ccl.net > wrote: > >> Dear Joe, >> >> The GNU toolchain works fine (at least gfortran 11) on Mac M1, and we've >> certainly used it to compile and run CASTEP successfully (www.castep.org= ). >> Obviously you need to make sure you also have the correct libraries, e.g= . >> OpenBLAS, FFTW, but MacPorts should take care of that. >> >> We generally find gfortran performance to be very competitive, and of >> course it's value for money is excellent! Having said that, the NAG >> compiler is fantastic for developing software, but not for performance >> (that's not a criticism, that's a deliberate design decision by NAG). >> >> All the best, >> >> Phil >> >> >> ------------------------------------------------------------------------= ------------ >> Dr Phil Hasnip Email: phil.hasnip##york.ac.uk >> EPSRC RSE Fellow Web: www-users.york.ac.uk/~pjh503 >> Dept of Physics >> University of York Tel: +44 (0)1904 322225 >> York YO10 5DD >> >> >> >> Virus-free. >> www.avast.com >> >> <#m_4969243596422521063_m_-4984180958677803491_DAB4FAD8-2DD7-40BB-A1B8-4= E2AA1F9FDF2> >> >> On Thu, 10 Jun 2021 at 05:10, Joe Leonard jleonard42{}gmail.com < >> owner-chemistry##ccl.net> wrote: >> >>> >>> On Jun 9, 2021, at 10:07 PM, Phil Hasnip phil.hasnip[-]york.ac.uk < >>> owner-chemistry__ccl.net> wrote: >>> >>> I'm confused about your Mac M1 chipset reference; how does this relate >>> to your Fortran question? There are good ARM Fortran compilers, if that= 's >>> what you're worried about. >>> >>> >>> Phil, what compilers are up and working on the MacOS/M1 combination? I >>> am aware of the NAG compiler, which has issues for commercial developme= nt >>> (licensing) so I have been told. Also, given the staggeringly greater >>> interest in C/C++ on all sorts of machines, I would think that portabil= ity >>> would suggest alternate languages for new work - Python is a great >>> control-level language, but as you mentioned you need math underneath. >>> >>> Joe >>> =E2=80=94 >>> >> "Peter Thiel was right, we just can't build cool sh*t anymore. I really >>> did want a flying car, and all I have is 140 characters and promises of= AI >>> that never come true." >>> >>> -- > -- > "Peter Thiel was right, we just can't build cool sh*t anymore. I really > did want a flying car, and all I have is 140 characters and promises of A= I > that never come true." > --000000000000323fd905c47f3747 Content-Type: text/html; charset="UTF-8" Content-Transfer-Encoding: quoted-printable
Hi everyone,

Also, remember that = there is a free version of the PGI compiler with CUDA, the PGI Community Ed= ition:=C2=A0https:= //developer.nvidia.com/openacc-toolkit

I'm not t= oo sure about its efficiency, but I suppose it would be comparable to GCC, = which is very good indeed. If you need to compile or develop a GPU-accelera= ted code for NVIDIA GPUs, it's certainly something to try out.

Best,

Marcos=C2=A0

---

Prof. Dr.= Marcos Verissimo Alves
Prof. Adjunto II, Curso de F=C3=ADsica Computac= ional
Instituto de Ci=C3=AAncias= Exatas
Universidade Federal Fluminense
Volt= a Redonda - RJ, Brasil
<= /div>

Em sex., 11 de jun. de 2021 =C3=A0s 12:08, Joseph Leonard j= leonard42||gmail.com <owner-chemistry:ccl.net> escreveu:
Phil,= glad to hear that you had success with gfortran on that platform. It=E2=80= =99s been a long time sopince I worked in fortran and those doing it more r= ecently had a very dim view of that suite.=C2=A0 I will pass it along.=C2= =A0

Joe

On Thu, Jun = 10, 2021 at 10:36 PM Phil Hasnip phil.hasnip]_[york.ac.uk <owner-chemistry%%ccl.net> wrote:
Dear J= oe,

The GNU toolchain works fine (at least gfortran 11) = on Mac M1, and we've certainly used it to compile and run CASTEP succes= sfully (www.castep.org<= /a>). Obviously you need to make sure you also have the correct libraries, = e.g. OpenBLAS, FFTW, but MacPorts should take care of that.

<= /div>

3D"" Virus-free. www.avast.com

=
On Thu, 10= Jun 2021 at 05:10, Joe Leonard jleonard42{}gmail.com <owner-chemistry##ccl.net> wrote:

On Jun 9, 2021, at 10:07 PM, Phil Hasnip phil.hasnip[-]york.ac.uk <owner-chemistry__ccl.net= > wrote:

I'm confused about your Mac M1 chipset reference; how does this rela= te to your Fortran question? There are good ARM Fortran compilers, if that&= #39;s what you're worried about.=C2=A0
<= div>
Phil, what compilers are up and working on the MacOS/M1 combi= nation?=C2=A0 I am aware of the NAG compiler, which has issues for commerci= al development (licensing) so I have been told.=C2=A0 Also, given the stagg= eringly greater interest in C/C++ on all sorts of machines, I would think t= hat portability would suggest alternate languages for new work - Python is = a great control-level language, but as you mentioned you need math undernea= th.

Joe
<= div style=3D"text-align:start;text-indent:0px">
<= div style=3D"text-align:start;text-indent:0px">
=E2=80=94
<= div dir=3D"auto" style=3D"text-align:start;text-indent:0px">
<= div style=3D"text-align:start;text-indent:0px">
<= div style=3D"text-align:start;text-indent:0px">
"Peter Thiel was right, we just can't build= =C2=A0cool sh*t anymore. I really did want a=C2=A0flying car, and all I hav= e is 140=C2=A0characters and promises of AI that never=C2=A0come true."= ;
<= /div>
<= /div>
<= /div>
<= /div>
<= /div>

--
--
"Peter Thiel was right, we just can't build cool sh*t anymore. = I really did want a flying car, and all I have is 140 characters and promis= es of AI that never come true."
--000000000000323fd905c47f3747--