<html xmlns="http://www.w3.org/1999/xhtml" xmlns:v="urn:schemas-microsoft-com:vml" xmlns:o="urn:schemas-microsoft-com:office:office"><head><!--[if gte mso 9]><xml><o:OfficeDocumentSettings><o:AllowPNG/><o:PixelsPerInch>96</o:PixelsPerInch></o:OfficeDocumentSettings></xml><![endif]--></head><body>
Hello Richard Philips well you have mentioned too much details so may be you already know more than me, as I am also just a beginner. But what I learnt is there are two ways to configure correctly. 1: install module in your system, call for nvhpcopenmpi module and configure with normal cuda flags i.e —with-cuda= and —with-cuda-cc=…<div>2. Find correct paths of mpi and compiler libraries and along with cuda flags also give mpif90=path/to/your/mpif90 and cc=nvcc(optional)</div><div>Here is example command that worked for me.</div><div><br><p style="margin: 0px; font-style: normal; font-variant-caps: normal; font-stretch: normal; font-size: 12px; line-height: normal; font-family: system-ui; font-size-adjust: none; font-kerning: auto; font-variant-alternates: normal; font-variant-ligatures: normal; font-variant-numeric: normal; font-variant-east-asian: normal; font-variant-position: normal; font-variant-emoji: normal; font-feature-settings: normal; font-optical-sizing: auto; font-variation-settings: normal;">./configure --with-cuda=$CUDA_HOME --with-cuda-cc=86 --with-cuda-runtime=12.4 --enable-parallel --with-cuda-cc=86 MPIF90=/opt/nvidia/hpc_sdk/Linux_x86_64/24.5/comm_libs/12.4/openmpi4/openmpi-4.1.5/bin/mpif90 </p><br><div class="yahoo-signature"><a style="font-size:12px" href="https://mail.onelink.me/107872968?pid=nativeplacement&c=Global_Acquisition_YMktg_315_Internal_EmailSignature&af_sub1=Acquisition&af_sub2=Global_YMktg&af_sub3=&af_sub4=100000604&af_sub5=EmailSignature__Static_">Sent from Yahoo Mail for iPhone</a><br></div><br><p class="yahoo-quoted-begin" style="font-size: 15px; color: rgb(132, 115, 255); padding-top: 15px; margin-top: 0px;">On Wednesday, February 5, 2025, 8:10 PM, R T Phillips via users <users@lists.quantum-espresso.org> wrote:</p><blockquote class="iosymail"><div dir="ltr">I have been trying to compile Quantum Espresso for use on this system:<br></div><div dir="ltr">Intel i9-14900 (24 core)<br></div><div dir="ltr">Linux Mint 22.1 Cinnamon version 6.4.6<br></div><div dir="ltr">Nvidia GeForce RTX 4070<br></div><div dir="ltr"><br></div><div dir="ltr">with:<br></div><div dir="ltr">nvhpc_2024_2411_Linux_x86_64_cuda_12.6<br></div><div dir="ltr"><br></div><div dir="ltr">The prerequisite gcc has been tested before installing nvhpc as <br></div><div dir="ltr">specified in $1.1 of the nvidia hpc-sdk installation guide <br></div><div dir="ltr">(https://docs.nvidia.com/hpc-sdk/archive/20.9/pdf/hpc209install.pdf).<br></div><div dir="ltr"><br></div><div dir="ltr">The hpc-sdk specification is now said to include cuda and openmpi, so I <br></div><div dir="ltr">have used only the one installation step at the top level of the nvhpc <br></div><div dir="ltr">distribution (adding independent cuda or openmpi installations now seems <br></div><div dir="ltr">to create havoc, though these may have been required in earlier versions <br></div><div dir="ltr">of the hpc-sdk).<br></div><div dir="ltr"><br></div><div dir="ltr">After installation of the nvhpc, nvaccelinfo produces the following <br></div><div dir="ltr">output. Does this demonstrate that a "proper parallel environment" can <br></div><div dir="ltr">be detected, as mentioned in the QE Guide?<br></div><div dir="ltr"><br></div><div dir="ltr">"CUDA Driver Version: 12040<br></div><div dir="ltr">NVRM version: NVIDIA UNIX x86_64 Kernel Module 550.120 <br></div><div dir="ltr">Fri Sep 13 10:10:01 UTC 2024<br></div><div dir="ltr"><br></div><div dir="ltr">Device Number: 0<br></div><div dir="ltr">Device Name: NVIDIA GeForce RTX 4070<br></div><div dir="ltr">Device Revision Number: 8.9<br></div><div dir="ltr">Global Memory Size: 12584878080<br></div><div dir="ltr">Number of Multiprocessors: 46<br></div><div dir="ltr">Concurrent Copy and Execution: Yes<br></div><div dir="ltr">Total Constant Memory: 65536<br></div><div dir="ltr">Total Shared Memory per Block: 49152<br></div><div dir="ltr">Registers per Block: 65536<br></div><div dir="ltr">Warp Size: 32<br></div><div dir="ltr">Maximum Threads per Block: 1024<br></div><div dir="ltr">Maximum Block Dimensions: 1024, 1024, 64<br></div><div dir="ltr">Maximum Grid Dimensions: 2147483647 x 65535 x 65535<br></div><div dir="ltr">Maximum Memory Pitch: 2147483647B<br></div><div dir="ltr">Texture Alignment: 512B<br></div><div dir="ltr">Clock Rate: 2520 MHz<br></div><div dir="ltr">Execution Timeout: Yes<br></div><div dir="ltr">Integrated Device: No<br></div><div dir="ltr">Can Map Host Memory: Yes<br></div><div dir="ltr">Compute Mode: default<br></div><div dir="ltr">Concurrent Kernels: Yes<br></div><div dir="ltr">ECC Enabled: No<br></div><div dir="ltr">Memory Clock Rate: 10501 MHz<br></div><div dir="ltr">Memory Bus Width: 192 bits<br></div><div dir="ltr">L2 Cache Size: 37748736 bytes<br></div><div dir="ltr">Max Threads Per SMP: 1536<br></div><div dir="ltr">Async Engines: 2<br></div><div dir="ltr">Unified Addressing: Yes<br></div><div dir="ltr">Managed Memory: Yes<br></div><div dir="ltr">Concurrent Managed Memory: Yes<br></div><div dir="ltr">Preemption Supported: Yes<br></div><div dir="ltr">Cooperative Launch: Yes<br></div><div dir="ltr">Unified Memory: No<br></div><div dir="ltr">Memory Models Flags: -gpu=mem:separate, -gpu=mem:managed<br></div><div dir="ltr">Default Target: cc89 "<br></div><div dir="ltr"><br></div><div dir="ltr">After the installation of nvhpc, I have tried both the default <br></div><div dir="ltr">.configure, and also a bash script of QE-7.3.1 (see attached), setting <br></div><div dir="ltr">the environment variables specified in $2.4 of the QE Users' Guide, and <br></div><div dir="ltr">nvhpc.<br></div><div dir="ltr"><br></div><div dir="ltr">Both produce config.log files which reveal problems in finding various <br></div><div dir="ltr">components - the Guide warns that this might occur. I have trawled many <br></div><div dir="ltr">websites for a solution but have not resolved this problem. I have <br></div><div dir="ltr">tried some variants of the above, which have led to apparently <br></div><div dir="ltr">successful configuration runs, but 'make' subsequently fails.<br></div><div dir="ltr"><br></div><div dir="ltr">It is pretty obvious that I am completely inexpert at this, and would <br></div><div dir="ltr">greatly appreciate advice if anyone can spot what I am doing incorrectly.<br></div><div dir="ltr"><br></div><div dir="ltr">Richard Phillips, Cavendish Lab. UK<br></div><div dir="ltr"><br></div><div dir="ltr"><br></div>_______________________________________________________________________________<br>The Quantum ESPRESSO Foundation stands in solidarity with all civilians worldwide who are victims of terrorism, military aggression, and indiscriminate warfare.<br>--------------------------------------------------------------------------------<br>Quantum ESPRESSO is supported by MaX (www.max-centre.eu)<br>users mailing list <a ymailto="mailto:users@lists.quantum-espresso.org" href="mailto:users@lists.quantum-espresso.org">users@lists.quantum-espresso.org</a><br><a href="https://lists.quantum-espresso.org/mailman/listinfo/users" target="_blank">https://lists.quantum-espresso.org/mailman/listinfo/users</a><blockquote></blockquote></blockquote></div>
</body></html>