Linux CUDA13 Action (#2186)

* Linux CU13 CI

* Bump max CUDA arch

* CUDA13 Linux

* Upload the correct build to rolling (CUDA13)

* Downgrade cuda to get better compatibility

Runpod can't handle 13.1, and if they can't handle it neither can the people with a secondary GPU of an older generation.

* Add support for compute capability 89 in NVCCFLAGS

.github/workflows/kcpp-build-release-linux-cu13.yaml

@@ -0,0 +1,77 @@
+name: Koboldcpp Linux CUDA13
+
+on:
+  workflow_dispatch:
+    inputs:
+      commit_hash:
+        description: 'Optional commit hash to build from'
+        required: false
+        default: ''
+
+env:
+  BRANCH_NAME: ${{ github.head_ref || github.ref_name }}
+  KCPP_CUDA: 13.0.0
+  ARCHES_CU13: 1
+
+jobs:
+  linux:
+    runs-on: ubuntu-22.04
+    permissions: write-all
+    container:
+      image: ubuntu:20.04
+      options: --privileged
+    steps:
+      - name: Clone
+        id: checkout
+        uses: actions/checkout@v5
+        with:
+          ref: ${{ inputs.commit_hash != '' && inputs.commit_hash || github.head_ref || github.ref_name }}
+
+      - name: Show Commit Used
+        run: |
+          echo "Building from ref: ${{ inputs.commit_hash != '' && inputs.commit_hash || github.head_ref || github.ref_name }}"
+
+      - name: Dependencies
+        id: depends
+        run: |
+          apt-get update
+          apt-get install -y sudo
+          export DEBIAN_FRONTEND=noninteractive
+          sudo ln -fs /usr/share/zoneinfo/UTC /etc/localtime
+          echo "tzdata tzdata/Areas select Etc" | sudo debconf-set-selections
+          echo "tzdata tzdata/Zones/Etc select UTC" | sudo debconf-set-selections
+          sudo apt-get -y install git curl bzip2 python3-tk tcl tk
+
+      - name: Set Tcl/Tk Paths
+        run: |
+          echo "TCL_LIBRARY=$(find /usr/lib/ -name 'tcl8*' | head -n 1)" >> $GITHUB_ENV
+          echo "TK_LIBRARY=$(find /usr/lib/ -name 'tk8*' | head -n 1)" >> $GITHUB_ENV
+
+      - name: Build
+        id: make_build
+        run: |
+          ./koboldcpp.sh dist
+
+      - name: Save artifact
+        uses: actions/upload-artifact@v6
+        with:
+          name: kcpp_linux_binary
+          path: dist/
+
+      - name: Install GitHub CLI
+        run: |
+          apt-get update
+          apt-get install -y curl
+          curl -fsSL https://cli.github.com/packages/githubcli-archive-keyring.gpg \
+            | dd of=/usr/share/keyrings/githubcli-archive-keyring.gpg
+          chmod go+r /usr/share/keyrings/githubcli-archive-keyring.gpg
+          echo "deb [arch=$(dpkg --print-architecture) signed-by=/usr/share/keyrings/githubcli-archive-keyring.gpg] https://cli.github.com/packages stable main" \
+            | tee /etc/apt/sources.list.d/github-cli.list > /dev/null
+          apt-get update
+          apt-get install -y gh
+
+      - name: Upload to GitHub Rolling Release
+        env:
+          GH_TOKEN: ${{ secrets.GITHUB_TOKEN }}
+        run: |
+          gh release upload rolling dist/koboldcpp-linux-x64-cuda1300 --clobber --repo ${{ github.repository }}

Makefile

@@ -247,7 +247,9 @@ NVCCFLAGS += -Wno-deprecated-gpu-targets \
              -gencode arch=compute_75,code=compute_75 \
              -gencode arch=compute_80,code=compute_80 \
              -gencode arch=compute_86,code=compute_86 \
-			 -DKCPP_LIMIT_CUDA_MAX_ARCH=860
+             -gencode arch=compute_89,code=compute_89 \
+             -gencode arch=compute_120,code=compute_120 \
+			 -DKCPP_LIMIT_CUDA_MAX_ARCH=1200
 
 else
 NVCCFLAGS += -Wno-deprecated-gpu-targets -arch=all