aCG/NEWS at main · ParCoreLab/aCG · GitHub

1
2
3
4
5
6
7
8
9
10
11
12
13
14
15
16
17
18
19
20
21
22
23
* Noteworthy changes in release 1.0.0 (2025-04-26)

This is the initial release of aCG, which provides GPU-accelerated
iterative linear solvers based on the conjugate gradient (CG) method.
The solvers support NVIDIA and AMD GPUs, and multi-GPU systems with
GPU-aware MPI, NCCL/RCCL or NVSHMEM.

For NVIDIA GPUs, CUDA implementations of CG and pipelined CG are
provided. Communication is performed by the host CPU using GPU-aware
MPI, NCCL or NVSHMEM, or by the GPU using NVSHMEM device-initiated
communication.

For AMD GPUs, HIP implementations of CG and pipelined CG are provided.
Communication is performed by the host CPU using GPU-aware MPI or RCCL
communication.  A single-GPU HIP version of the monolithic,
device-side CG solver is also provided.

--
Copyright 2025 Koç University and Simula Research Laboratory

Copying and distribution of this file, with or without modification,
are permitted provided the copyright notice and this notice are
preserved.