jgrusewski 3112e07a5a perf(ci): serialize dqn-pipeline tests to avoid GPU contention ...

4 concurrent DQN trainers on a shared H100 causes severe GPU contention
(28-39% utilization, 42 GB VRAM). Serializing with --test-threads=1
gives each test full GPU access, reducing total wallclock time despite
sequential execution.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>

2026-03-13 19:52:04 +01:00

..

docker

feat(infra): auto-rebuild Docker images in CI pipelines, add build-image webhook

2026-03-12 08:35:55 +01:00

k8s

perf(ci): serialize dqn-pipeline tests to avoid GPU contention

2026-03-13 19:52:04 +01:00

live/production

infra: automate terragrunt via Argo CI, clean up kapsule module, harden PAT rotation

2026-03-10 23:25:33 +01:00

modules

fix(infra): remove stale DNS outputs, add chat/mail FQDNs

2026-03-12 14:47:04 +01:00

scripts

fix(train): correct CLI args for hyperopt vs training binaries

2026-03-09 22:39:02 +01:00