Issues

• On Windows, is the AMGX dynamic library built with VC + CUDA (with a pure C interface) directly usable with the MinGW compiler?

  NVIDIA/AMGX#384 · Filter by author xiaochiyue opened on Sep 28, 2026

• How to utilize AMGX to customize a user-defined solver that acts as a preconditioner for the upper-level solver?

  NVIDIA/AMGX#383 · Filter by author xiaochiyue opened on Sep 24, 2026

• [Issue] Solving the same system on multi GPUs repeatedly gives different results

  bug
  NVIDIA/AMGX#381 · Filter by author stevenliuyi opened on Sep 13, 2026

• [Issue] CLASSICAL AMG fails with cudaErrorInvalidDeviceFunction on CUDA 13 / SM89 while AGGREGATION works

  bug
  NVIDIA/AMGX#380 · Filter by author anme-dhi opened on Sep 11, 2026

• [Issue] Device dotc uses unconjugated cuBLAS dotu and breaks complex GMRES orthogonalization

  bug
  NVIDIA/AMGX#377 · Filter by author 2002WYT opened on Aug 25, 2026

• how to set the min_iters in the configure file.

  NVIDIA/AMGX#376 · Filter by author ztdepztdep opened on Aug 23, 2026

• Feature: Smoothed Aggregation AMG with MIS-k selectors (prototype available)

  NVIDIA/AMGX#373 · Filter by author markadams opened on May 15, 2026
  1

• Question about smoothed aggregation

  NVIDIA/AMGX#369 · Filter by author itk22 opened on Apr 21, 2026

• "optimal" configuration file

  NVIDIA/AMGX#367 · Filter by author RiccardoRossi opened on Mar 27, 2026

• How to set solver's parameters to eliminate isolated equations' effect?

  NVIDIA/AMGX#365 · Filter by author xiaochiyue opened on Feb 2, 2026

• The official version of amgx4foam

  NVIDIA/AMGX#362 · Filter by author jyluan opened on Jan 22, 2026
  2

• AMGX 2.5.0: FGMRES/BiCGStab stagnates near 3e-3 residual on large real general system; many rows have near-zero row-sum; some AMG configs produce NaN residual

  build issues
  NVIDIA/AMGX#361 · Filter by author Guo150-3 opened on Dec 24, 2025

• AMGX 2.5.0 solver stagnation on large system

  build issues
  NVIDIA/AMGX#360 · Filter by author Guo150-3 opened on Dec 24, 2025

• [Build]

  build issues
  NVIDIA/AMGX#358 · Filter by author Liekkas-xlm opened on Dec 21, 2025
  1

• [Issue]I think I might have encountered a bug. When I use the following configuration, the AMGX solver freezes. I'm not even sure if it has entered the solving state—it just stalls.

  bug
  NVIDIA/AMGX#353 · Filter by author Haiqing-Fang opened on Dec 8, 2025

• Are there best practices for AMGX on multiple GPUs? (NVLink)

  NVIDIA/AMGX#352 · Filter by author xiaorong117 opened on Dec 6, 2025

• [Build] Difficulty building with CUDA 12.9+

  build issues
  NVIDIA/AMGX#351 · Filter by author chrisrichardson opened on Nov 28, 2025
  3

• [Issue] When I set the convergence criterion to the absolute residual, the AMGX iteration log reports that “Relative residual has reached machine precision” and then stops further iterations.

  bug
  NVIDIA/AMGX#350 · Filter by author Haiqing-Fang opened on Nov 22, 2025
  1

• Can AMGX implement the additive Schwarz preconditioning method across multiple GPUs?

  new feature
  NVIDIA/AMGX#349 · Filter by author xiaorong117 opened on Nov 16, 2025
  1

• What preprocessing methods are supported for PCG

  NVIDIA/AMGX#347 · Filter by author Yangshen-bot opened on Aug 19, 2025
  1

• [Build] Conda build error, nvtx3 not found

  build issues
  NVIDIA/AMGX#346 · Filter by author jwwood17 opened on Aug 12, 2025

• Compiling AMGX with NVHPC 25.5

  build issues
  Bug

  NVIDIA/AMGX#344 · Filter by author nibblelab opened on Jul 8, 2025
  

• How is the bandwidth of non-symmetric matrices reduced for ILU(0) preconditioner

  NVIDIA/AMGX#343 · Filter by author science-enthusiast opened on Jun 16, 2025

• [Issue] Solver reports wrong vector size in parallel, whereas simple SpMV is correct

  bug
  NVIDIA/AMGX#342 · Filter by author YvanFournier opened on Jun 11, 2025
  1

• Solving Ax=b and A and b are sparse matrix and vector in mtx format - no solution yet.

  NVIDIA/AMGX#341 · Filter by author raavimohindar opened on Apr 7, 2025