I’m currently using a GPU to accelerate DMRG computations. My MPS is not large, so the computational scale and memory usage are also small (less than 3GB), which are all within my expectations. However, the GPU utilization is quite low (less than 25%). I’m not familiar with GPU computing in Julia, …

Those bond dimensions are very small, I’m not surprised that you don’t see good GPU utilization in that case. In general we see that bigger tensors get better speedups on GPU (since there are more elements of the tensors for the GPU to parallelize over). Regarding parallelization, for such small bon…

The maximal bond dimension is 7 for the MPO, and is about 20 for the MPS most of the time. I am not using any conserving QNs.

Low GPU utilization (<25%) when accelerating DMRG with GPU

ITensor Julia Questions

mtfishman September 8, 2025, 1:03pm 2

What are the bond dimensions of the MPO and MPS in your DMRG calculation? Additionally, are you conserving QNs? If so, see this post: TEBD time evolution with CUDA backend

Topic		Replies	Views
Memory bottleneck with GPU backend for DMRG ITensor Julia Questions julia , dmrg	1	638	September 12, 2022
GPU is not faster CPU ITensor Julia Questions dmrg	2	156	November 2, 2024
Large RAM and vRAM usage in DMRG DMRG and Numerical Methods	6	192	January 30, 2025
Naive use of CUDA for DMRG leads to MethodError ITensor Julia Questions	3	120	October 29, 2024
Pooling GPU memories with ITensorMPS ITensor Julia Questions	1	83	June 11, 2025

Low GPU utilization (<25%) when accelerating DMRG with GPU

Related topics