Experiments · E5
Can a 128-atom quantum calculation run on this computer?
No. It needs more than 23.45 GB of memory against 15.6 GB available, and was killed after about 35 minutes.
In the log: The 128-atom DFT cell does not fit locally
book/assets/diagrams/exp/E5.svg).Results
EXPERIMENTS.md · line 161Result. 1037 Kohn-Sham states, 1,840,005 G-vectors, FFT grid 160^3.
Estimated total dynamical RAM > 23.45 GB. The container has 15.6 GB; pw.x
exited 137 (SIGKILL, out of memory) after ~35 minutes.
The full record
EXPERIMENTS.md · lines 153–170
E5 — The 128-atom DFT cell does not fit locally
Date 2026-09-12 · Question Can E3's 128-atom conclusion be checked in DFT on this machine? · Provenance same as E4
Method. As E4 at 128 atoms (4x4x4), k-mesh scaled to 2x2x2 to hold k-spacing roughly fixed as the cell grows.
Result. 1037 Kohn-Sham states, 1,840,005 G-vectors, FFT grid 160^3.
Estimated total dynamical RAM > 23.45 GB. The container has 15.6 GB; pw.x
exited 137 (SIGKILL, out of memory) after ~35 minutes.
A 54-atom cell needs 5.68 GB (438 states) and runs locally.
Interpretation. 128-atom DFT for this chemistry is a hard memory limit on a 24 GB machine, not a tuning problem — more MPI ranks reduce per-rank memory but not the total. Until it is run elsewhere, E3's "128 atoms is the minimum" rests on MACE alone and must not be stated as a DFT result.
