Merge pull request cupy#8924 from jakirkham/use_nvidia_cccl_2.8_nvrtc…

…_fix

Use NVIDIA CCCL 2.8 latest w/CUDA 12.3 fix

.gitmodules

@@ -1,6 +1,6 @@
 [submodule "third_party/cccl"]
 	path = third_party/cccl
-	url = https://github.com/cupy/cccl.git
+	url = https://github.com/NVIDIA/cccl.git
 [submodule "third_party/jitify"]
 	path = third_party/jitify
 	url = https://github.com/NVIDIA/jitify.git

.pfnci/coverage.rst

@@ -3249,7 +3249,9 @@ CuPy CI Test Coverage
      - 
    * - 
      - cutensor
-     - 2
+     - 🚨
+     - 
+     - 
      - 
      - 
      - 
@@ -3270,8 +3272,6 @@ CuPy CI Test Coverage
      - 
      - 
      - 
-     - ✅
-     - ✅
      - 
      - 
      - 
@@ -3329,7 +3329,7 @@ CuPy CI Test Coverage
      - 
    * - 
      - cutensor,cub
-     - 26
+     - 28
      - 
      - 
      - 
@@ -3350,8 +3350,8 @@ CuPy CI Test Coverage
      - ✅
      - ✅
      - ✅
-     - 
-     - 
+     - ✅
+     - ✅
      - ✅
      - ✅
      - ✅

.pfnci/linux/tests/cuda123.multi.sh

@@ -9,7 +9,7 @@ ACTIONS="$(dirname $0)/actions"
 
 export NVCC="ccache nvcc"
 
-export CUPY_ACCELERATORS="cutensor"
+export CUPY_ACCELERATORS="cutensor,cub"
 
 "$ACTIONS/build.sh"
 export OMPI_ALLOW_RUN_AS_ROOT=1

.pfnci/linux/tests/cuda123.sh

@@ -9,7 +9,7 @@ ACTIONS="$(dirname $0)/actions"
 
 export NVCC="ccache nvcc"
 
-export CUPY_ACCELERATORS="cutensor"
+export CUPY_ACCELERATORS="cutensor,cub"
 
 "$ACTIONS/build.sh"
 "$ACTIONS/unittest.sh" "not slow and not multi_gpu"

.pfnci/matrix.yaml

@@ -322,7 +322,7 @@
   mpi4py: null
   cython: "0.29"
   cuda-python: null
-  env:CUPY_ACCELERATORS: "cutensor"
+  env:CUPY_ACCELERATORS: "cutensor,cub"
   test: "unit"
 
 # CUDA 12.3 (Multi-GPU) | Linux
@@ -332,7 +332,6 @@
   target: "cuda123.multi"
   mpi4py: "3"
   test: "unit-multi"
-  env:CUPY_ACCELERATORS: "cutensor"
 
 # CUDA 12.4 | Linux
 - project: "cupy.linux.cuda124"

.pfnci/windows/_flexci.ps1

@@ -52,7 +52,6 @@ function ActivateCUDA($version) {
         $Env:CUDA_PATH = $Env:CUDA_PATH_V12_2
     } elseif ($version -eq "12.3") {
         $Env:CUDA_PATH = $Env:CUDA_PATH_V12_3
-        $Env:CUPY_ACCELERATORS = ""
     } elseif ($version -eq "12.4") {
         $Env:CUDA_PATH = $Env:CUDA_PATH_V12_4
     } elseif ($version -eq "12.5") {