From 64a4f4fc0d3784c0375fe9f76112d2b808bafd05 Mon Sep 17 00:00:00 2001
From: jgrusewski <jgrusewski@localhost>
Date: Thu, 26 Feb 2026 12:59:03 +0100
Subject: [PATCH] =?UTF-8?q?fix(ci):=20purge=20sccache=20before=20test=20?=
 =?UTF-8?q?=E2=80=94=20stale=20L4=20PTX=20fails=20on=20H100?=
MIME-Version: 1.0
Content-Type: text/plain; charset=UTF-8
Content-Transfer-Encoding: 8bit

sccache PVC holds CUDA kernel objects compiled for sm_89 (L4).
After moving the runner pool to H100 (sm_90), these produce
CUDA_ERROR_INVALID_PTX at runtime. Nuke the cache so candle
recompiles kernels for the correct compute capability.

Co-Authored-By: Claude Opus 4.6 <noreply@anthropic.com>
---
 .gitlab-ci.yml | 2 ++
 1 file changed, 2 insertions(+)

diff --git a/.gitlab-ci.yml b/.gitlab-ci.yml
index 5dea8e2db..87f073067 100644
--- a/.gitlab-ci.yml
+++ b/.gitlab-ci.yml
@@ -219,6 +219,8 @@ test:
         fi
         echo "Waiting for Redis... ($i/30)"; sleep 1
       done
+    # Purge sccache — stale PTX from L4 (sm_89) causes CUDA_ERROR_INVALID_PTX on H100 (sm_90)
+    - rm -rf "${SCCACHE_DIR:?}"/* || true
     - sccache --zero-stats || true
     - cargo test --workspace --lib -- --skip model_loader::tests
     - sccache --show-stats || true