gpu-compute,tests: Move GPU tests to testlib (#1270)

A new host tag `gcn_gpu` has been added. This allows for selection of
those GPU tests which depend upon the gcn-gpu docker image to run.

In addition to this, the square GPU tests has been moved to the CI
tests. This ensures some GPU code is compiled and run on every PR.

.github/workflows/daily-tests.yaml

@@ -80,6 +80,30 @@ jobs:
                   retention-days: 7
             - run: echo "This job's status is ${{ job.status }}."
 
+    gpu-tests:
+        runs-on: [self-hosted, linux, x64]
+        container: ghcr.io/gem5/gcn-gpu:latest
+        timeout-minutes: 300
+
+        steps:
+            - uses: actions/checkout@v4
+              with:
+                  ref: develop
+            - name: Build VEGA_X86/gem5.opt
+              run: scons build/VEGA_X86/gem5.opt -j`nproc`
+
+            - name: Run Testlib GPU Tests
+              working-directory: ${{ github.workspace }}/tests
+              run: ./main.py run --length=long --skip-build -vvv -t $(nproc) --host gcn_gpu
+
+            - name: Upload results
+              if: success() || failure()
+              uses: actions/upload-artifact@v4
+              with:
+                  name: gpu_COMMIT.${{github.sha}}_RUN.${{github.run_id}}_ATTEMPT.${{github.run_attempt}}
+                  path: tests/testing-results
+                  retention-days: 7
+
   # This runs the SST-gem5 integration compilation and tests it with
   # ext/sst/sst/example.py.
     sst-test:
@@ -124,30 +148,6 @@ jobs:
             - name: Continue gem5 within SystemC test
               run: LD_LIBRARY_PATH=build/ARM/:/opt/systemc/lib-linux64/ ./util/systemc/gem5_within_systemc/gem5.opt.sc m5out/config.ini
 
-  # Runs the gem5 Nighyly GPU tests.
-    gpu-tests:
-        runs-on: [self-hosted, linux, x64]
-        container: ghcr.io/gem5/gcn-gpu:latest
-        timeout-minutes: 720 # 12 hours
-
-        steps:
-            - uses: actions/checkout@v4
-            - name: Compile build/VEGA_X86/gem5.opt
-              run: scons build/VEGA_X86/gem5.opt -j $(nproc)
-            - name: Get Square test-prog from gem5-resources
-              run: build/VEGA_X86/gem5.opt util/obtain-resource.py square-gpu-test -p square
-            - name: Run Square test with VEGA_X86/gem5.opt (SE mode)
-              run: |
-                  mkdir -p tests/testing-results
-                  ./build/VEGA_X86/gem5.opt configs/example/apu_se.py --reg-alloc-policy=dynamic -n3 -c square
-            - name: Get allSyncPrims-1kernel from gem5-resources
-              run: build/VEGA_X86/gem5.opt util/obtain-resource.py allSyncPrims-1kernel -p allSyncPrims-1kernel
-            - name: Run allSyncPrims-1kernel sleepMutex test with VEGA_X86/gem5.opt (SE mode)
-              run: ./build/VEGA_X86/gem5.opt configs/example/apu_se.py --reg-alloc-policy=dynamic -n3 -c allSyncPrims-1kernel --options="sleepMutex 10 16
-                  4"
-            - name: Run allSyncPrims-1kernel lfTreeBarrUsing test with VEGA_X86/gem5.opt (SE mode)
-              run: ./build/VEGA_X86/gem5.opt configs/example/apu_se.py --reg-alloc-policy=dynamic -n3 -c allSyncPrims-1kernel --options="lfTreeBarrUniq
-                  10 16 4"
     daily-tests:
         # The dummy job is used to indicate whether the daily tests have
         # passed or not. This can be used as status check for pull requests.