Hi Ding, If you are willing to, you can help us by testing if the CUDA13 branch (https://github.com/projectchrono/DEM-Engine/tree/CUDA13) works on your machine with CUDA13 installed. Just so you know, the performance update I mentioned last time, is postponed. So this branch is DEME with the old performance which is still good. Let me know if you find it usable.
Thank you, Ruochun On Friday, January 9, 2026 at 6:04:00 PM UTC+8 Ruochun Zhang wrote: > Hi Ding, > > CUDA13 removed a bunch of things DEME currently uses. We should be able to > push a new version in a few days which fixes this problem and provides huge > performance boosts. I'll reply here when it is out. So I'd say you either > wait a few days, or use CUDA 12.8 or below. > > Thank you, > Ruochun > > On Friday, January 9, 2026 at 5:46:45 PM UTC+8 [email protected] wrote: > >> Hi Ruochun, I encountered a similar issue when compiling the DEM-engine. >> My CUDA version is 13.0. After generating the .sln file via CMake, I tried >> to build the solution in MSVS 2022, but the build failed with the error >> message shown in the image below. How should I resolve this? Best regards, >> Ding >> >> [image: 屏幕截图 2026-01-09 174529.png] >> 在2025年4月9日星期三 UTC+8 02:31:05<Ruochun Zhang> 写道： >> >>> Hi Mahmoud, >>> >>> With what we have here, I feel both issues are related to multiple >>> installations. >>> >>> For cmake, it could be that you have multiple cmake installations, in >>> which case you should specify the one you want to use in PATH. Or if you >>> are using ccmake, perhaps it's linked against a cmake backend which is old. >>> Reinstalling them should always work... The README document of DEME has a >>> link to a guide on updating cmake to the newest. It's written for WSL, but >>> works for Linux as well. >>> >>> For CUDA, I'm pretty sure 11.3 and 12.1~12.3 do not work, which may >>> include the versions you tried. This is a limitation I cannot overcome and >>> you just have to use a different version. I would just recommend the newest >>> CUDA; or try 12.0, if you want to be safe. The most important thing is to >>> set the CUDA-related environment variables correctly before compilation, if >>> you have multiple installations. The README document of DEME also contains >>> some guidance on how to do that. >>> >>> Ruochun >>> >>> On Wednesday, April 9, 2025 at 1:25:27 AM UTC+8 [email protected] >>> wrote: >>> >>>> Hello All, >>>> >>>> I am trying to run the tutorial BallDrop2D on the github repo, When I >>>> try to compile the .cpp file, I get an error stating that cmake version is >>>> not compatible. Needs minimum version 3.27. The version I have is up to >>>> date 3.27.7. >>>> >>>> Also, I tried to run it PyDEME version, my GPU is recongized and my >>>> cuda version is 12.2. I tried older versions like 11, but when I run the >>>> example I get runtime error and cuda version is defined. My question is >>>> what is the best cuda version suitable for PyDEME that will not cause >>>> problems? >>>> >>>> Thanks >>>> >>> -- You received this message because you are subscribed to the Google Groups "ProjectChrono" group. To unsubscribe from this group and stop receiving emails from it, send an email to [email protected]. To view this discussion visit https://groups.google.com/d/msgid/projectchrono/b7f3f459-f6bd-4262-8f4f-6b05a8027e7an%40googlegroups.com.
