From 29bddb1b333107ae2b5367843d5e7d182be6fe26 Mon Sep 17 00:00:00 2001
From: Ashutosh Parkhi <86472128+ashutosh-arm@users.noreply.github.com>
Date: Mon, 23 Jan 2023 15:39:04 +0000
Subject: [PATCH] Update manylinux container tags (#156)

Bumping up tlcpack/package-* versions to latest
docker images that include support for Arm(R) Ethos(TM)-N
NPU driver 22.11. This should fix tlcpack nightly and pypi CI.
---
 .github/workflows/conda_gpu_nigthly.yaml       | 10 +++++-----
 .github/workflows/wheel_manylinux_nightly.yaml | 12 ++++++------
 .github/workflows/wheel_manylinux_pypi.yaml    |  2 +-
 3 files changed, 12 insertions(+), 12 deletions(-)

diff --git a/.github/workflows/conda_gpu_nigthly.yaml b/.github/workflows/conda_gpu_nigthly.yaml
index 0e30680..2f0e8d5 100644
--- a/.github/workflows/conda_gpu_nigthly.yaml
+++ b/.github/workflows/conda_gpu_nigthly.yaml
@@ -17,16 +17,16 @@ jobs:
         pkg: ['tlcpack', 'tlcpack-nightly']
         config:
           - cuda: '10.2'
-            image: 'tlcpack/package-cu102:7bfdf7e'
+            image: 'tlcpack/package-cu102:2fddbe0'
           # Conda forge is not shipping a cudnn package with support for CUDA 11.0
           # - cuda: '11.0'
-          #   image: 'tlcpack/package-cu110:7bfdf7e'
+          #   image: 'tlcpack/package-cu110:2fddbe0'
           - cuda: '11.1'
-            image: 'tlcpack/package-cu111:7bfdf7e'
+            image: 'tlcpack/package-cu111:2fddbe0'
           - cuda: '11.3'
-            image: 'tlcpack/package-cu113:7bfdf7e'
+            image: 'tlcpack/package-cu113:2fddbe0'
           - cuda: '11.6'
-            image: 'tlcpack/package-cu116:7bfdf7e'
+            image: 'tlcpack/package-cu116:2fddbe0'
 
     runs-on: ubuntu-latest
 
diff --git a/.github/workflows/wheel_manylinux_nightly.yaml b/.github/workflows/wheel_manylinux_nightly.yaml
index 0b054e8..64fce0c 100644
--- a/.github/workflows/wheel_manylinux_nightly.yaml
+++ b/.github/workflows/wheel_manylinux_nightly.yaml
@@ -19,19 +19,19 @@ jobs:
         # matrix of build configs
         config:
           - cuda: 'none'
-            image: 'tlcpack/package-cpu:7bfdf7e'
+            image: 'tlcpack/package-cpu:2fddbe0'
           - cuda: '10.2'
-            image: 'tlcpack/package-cu102:7bfdf7e'
+            image: 'tlcpack/package-cu102:2fddbe0'
           # CUDA 11.0 is not supported for Conda build,
           # Let's drop support here as well.
           # - cuda: '11.0'
-          #   image: 'tlcpack/package-cu110:7bfdf7e'
+          #   image: 'tlcpack/package-cu110:2fddbe0'
           - cuda: '11.1'
-            image: 'tlcpack/package-cu111:7bfdf7e'
+            image: 'tlcpack/package-cu111:2fddbe0'
           - cuda: '11.3'
-            image: 'tlcpack/package-cu113:7bfdf7e'
+            image: 'tlcpack/package-cu113:2fddbe0'
           - cuda: '11.6'
-            image: 'tlcpack/package-cu116:7bfdf7e'
+            image: 'tlcpack/package-cu116:2fddbe0'
 
     runs-on: ubuntu-latest
 
diff --git a/.github/workflows/wheel_manylinux_pypi.yaml b/.github/workflows/wheel_manylinux_pypi.yaml
index 761a978..a918203 100644
--- a/.github/workflows/wheel_manylinux_pypi.yaml
+++ b/.github/workflows/wheel_manylinux_pypi.yaml
@@ -20,7 +20,7 @@ jobs:
         # matrix of build configs
         config:
           - cuda: 'none'
-            image: 'tlcpack/package-cpu:7bfdf7e'
+            image: 'tlcpack/package-cpu:2fddbe0'
             package_name: 'apache-tvm'
 
     runs-on: ubuntu-latest