jgrusewski 2d704eaa07 fix: backward cuBLASLt uses heuristic algo — AlgoGetIds selects graph-incompatible algo on H100 ...

AlgoGetIds picks algorithms by ID order. On H100 (SM90), the first
valid algo uses split-K/stream-K with internal workspace allocation
that silently produces zero output when replayed via CUDA Graph.
The heuristic selects graph-safe algorithms by design.

Also removes FOXHUNT_NO_GRAPH temp env var and raw DtoH diagnostic probe.

Co-Authored-By: Claude Opus 4.6 (1M context) <noreply@anthropic.com>

2026-04-14 00:01:34 +02:00

..

argo

fix: backward cuBLASLt uses heuristic algo — AlgoGetIds selects graph-incompatible algo on H100

2026-04-14 00:01:34 +02:00

databases

fix(infra): resolve postgres disk-full cascade, add SPOF mitigations

2026-03-28 18:55:25 +01:00

dns

fix: node-bootstrap resolv.conf keeps VPC upstream as DNS fallback

2026-04-12 10:49:45 +02:00

gitlab

fix(infra): resolve postgres disk-full cascade, add SPOF mitigations

2026-03-28 18:55:25 +01:00

gpu-overlays

fix(infra): replace gitlab.com/prometheus annotations with prometheus.io

2026-03-08 18:39:24 +01:00

jobs

chore: remove old Python-based Databento batch download job

2026-03-31 23:03:03 +02:00

mattermost

feat(infra): deploy Mattermost on platform nodes with CI webhook egress

2026-03-12 09:20:32 +01:00

minio

fix(infra): resolve postgres disk-full cascade, add SPOF mitigations

2026-03-28 18:55:25 +01:00

monitoring

fix(infra): resolve postgres disk-full cascade, add SPOF mitigations

2026-03-28 18:55:25 +01:00

network-policies

fix(infra): resolve postgres disk-full cascade, add SPOF mitigations

2026-03-28 18:55:25 +01:00

secrets

feat(infra): QuestDB metrics sink + monitoring network policies

2026-03-08 02:42:53 +01:00

services

fix: tune log levels across workspace — demote noisy warn to debug/trace

2026-03-14 11:35:15 +01:00

stalwart

feat(stalwart): fallback-admin via K8s secret for reliable auth

2026-03-12 23:01:42 +01:00

storage

feat(infra): replace PVC binary distribution with MinIO S3 fetch

2026-03-04 16:46:57 +01:00

tailscale

feat(infra): migrate container registry from SCW to internal GitLab

2026-03-05 13:52:24 +01:00

training

feat: TOML training profile system — config-driven hyperparameters

2026-03-21 23:35:41 +01:00

gpu-taint.yaml

infra: split GPU pool into training (H100) and inference (L4), add auto-taint

2026-02-24 17:39:56 +01:00

namespace.yaml

fix: comprehensive audit — real brokers, deployment fixes, production safety

2026-02-25 00:32:10 +01:00