diff --git a/.github/workflows/ci-dispatch-job.yml b/.github/workflows/ci-dispatch-job.yml
index e487d49fee6..aec1f06dcfa 100644
--- a/.github/workflows/ci-dispatch-job.yml
+++ b/.github/workflows/ci-dispatch-job.yml
@@ -78,7 +78,7 @@ jobs:
             cd ~/cccl
             echo -e "\e[1;34mRunning as 'coder' user in $(pwd):\e[0m"
             echo -e "\e[1;34m${COMMAND}\e[0m"
-            eval '${COMMAND}' || exit_code=$?
+            eval "${COMMAND}" || exit_code=$?
             if [ ! -z "$exit_code" ]; then
               echo -e "::group::️❗ \e[1;31mInstructions to Reproduce CI Failure Locally\e[0m"
               echo "::error:: To replicate this failure locally, follow the steps below:"
diff --git a/ci/inspect_changes.sh b/ci/inspect_changes.sh
index 357257c495a..dce8e2033fd 100755
--- a/ci/inspect_changes.sh
+++ b/ci/inspect_changes.sh
@@ -107,9 +107,10 @@ main() {
   echo
 
   # Print the list of files that have changed:
-  echo "Dirty files:"
+  echo "::group::Dirty files"
   dirty_files | sed 's/^/  - /'
-  echo ""
+  echo "::endgroup::"
+  echo
 
   echo "Modifications in project?"
 
diff --git a/ci/matrix.yaml b/ci/matrix.yaml
index 4848ccb760a..847346b493d 100644
--- a/ci/matrix.yaml
+++ b/ci/matrix.yaml
@@ -36,6 +36,48 @@ msvc2022: &msvc2022 { name: 'cl', version: '14.39', exe: 'cl++' }
 # oneAPI configs
 oneapi: &oneapi { name: 'oneapi', version: '2023.2.0', exe: 'icpc' }
 
+# GHA Workflow job matrices:
+workflows:
+  pull_request:
+    # default_projects: nvcc
+    - {job_types: ['build'], ctk: *ctk_prev_min, host_compiler: *gcc6,                         std: [11, 14]         }
+    - {job_types: ['build'], ctk: *ctk_prev_min, host_compiler: [*gcc7, *gcc8, *gcc9, *llvm9], std: [11, 14, 17]     }
+    - {job_types: ['build'], ctk: *ctk_prev_min, host_compiler: *msvc2017,                     std: 14               }
+    - {job_types: ['build'], ctk: *ctk_prev_max, host_compiler: *gcc11,                        std: [11, 14, 17],     cmake_cuda_arch: '60;70;80;90'}
+    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: [*gcc7, *gcc8, *gcc9],         std: [11, 14, 17]     }
+    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: [*gcc10, *gcc11],              std: [11, 14, 17, 20] }
+    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: [*llvm9, *llvm10],             std: [11, 14, 17]     }
+    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: [*llvm11, *llvm12, *llvm13],   std: [11, 14, 17, 20] }
+    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: [*llvm14, *llvm15],            std: [11, 14, 17, 20] }
+    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: [*gcc12, *llvm16],             std: [11, 14, 17, 20], cpu: 'arm64'}
+    - {job_types: ['test'],  ctk: *ctk_curr,     host_compiler: [*gcc12, *llvm16],             std: [11, 14, 17, 20], cmake_cuda_arch: '60;70;80;90'}
+    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: *oneapi,                       std: [11, 14, 17]     }
+    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: *msvc2019,                     std: [14, 17]         }
+    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: *msvc2022,                     std: [14, 17, 20],     cmake_cuda_arch: '60;70;80;90'}
+    # default_projects: clang-cuda
+    - {job_types: ['build'], device_compiler: *llvm-newest, host_compiler: *llvm-newest, std: [17, 20]}
+    # nvrtc:
+    - {job_types: ['nvrtc'], project: 'libcudacxx', ctk: *ctk_curr, host_compiler: *gcc12, std: [11, 14, 17, 20]}
+    # verify-codegen:
+    - { job_types: ['verify_codegen'], project: 'libcudacxx'}
+    # cccl-infra:
+    - {job_types: ['infra'], project: 'cccl', ctk: *ctk_prev_min, host_compiler: [*gcc-oldest, *llvm-oldest]}
+    - {job_types: ['infra'], project: 'cccl', ctk: *ctk_curr,     host_compiler: [*gcc-newest, *llvm-newest]}
+  nightly:
+    - {job_types: ['test'],  ctk: *ctk_prev_min, gpu: 'v100',     cmake_cuda_arch: '70', host_compiler: *gcc6,   std: [11]             }
+    - {job_types: ['test'],  ctk: *ctk_prev_min, gpu: 't4',       cmake_cuda_arch: '75', host_compiler: *llvm9,  std: [17]             }
+    - {job_types: ['test'],  ctk: *ctk_prev_max, gpu: 'rtx2080',  cmake_cuda_arch: '75', host_compiler: *gcc11,  std: [17]             }
+    - {job_types: ['test'],  ctk: *ctk_curr,     gpu: 'rtxa6000', cmake_cuda_arch: '86', host_compiler: *gcc7,   std: [14]             }
+    - {job_types: ['test'],  ctk: *ctk_curr,     gpu: 'l4',       cmake_cuda_arch: '89', host_compiler: *gcc12,  std: [11, 14, 17, 20] }
+    - {job_types: ['test'],  ctk: *ctk_curr,     gpu: 'rtx4090',  cmake_cuda_arch: '89', host_compiler: *llvm9,  std: [11]             }
+    - {job_types: ['test'],  ctk: *ctk_curr,     gpu: 'h100',     cmake_cuda_arch: '90', host_compiler: *gcc12,  std: [11, 20]         }
+    - {job_types: ['test'],  ctk: *ctk_curr,     gpu: 'h100',     cmake_cuda_arch: '90', host_compiler: *llvm16, std: [17]             }
+    # nvrtc:
+    - {job_types: ['nvrtc'], ctk: *ctk_curr,     gpu: 't4',       cmake_cuda_arch: '75', host_compiler: *gcc12,  std: [20],             project: ['libcudacxx']}
+    - {job_types: ['nvrtc'], ctk: *ctk_curr,     gpu: 'rtxa6000', cmake_cuda_arch: '86', host_compiler: *gcc12,  std: [20],             project: ['libcudacxx']}
+    - {job_types: ['nvrtc'], ctk: *ctk_curr,     gpu: 'l4',       cmake_cuda_arch: '89', host_compiler: *gcc12,  std: [11, 14, 17, 20], project: ['libcudacxx']}
+    - {job_types: ['nvrtc'], ctk: *ctk_curr,     gpu: 'h100',     cmake_cuda_arch: '90', host_compiler: *gcc12,  std: [11, 20],         project: ['libcudacxx']}
+
 #
 # Resources for compute_matrix.py:
 #
@@ -111,14 +153,33 @@ default_os_lookup:
   'ctk12.4-cl14.39': 'windows2022'
   'ctk12.4-oneapi2023.2.0': 'ubuntu22.04'
 
-all_gpus:
-  - 'v100'     # ??x: sm70, 32 GB
-  - 't4'       #  8x: sm75, 16 GB
-  - 'rtx2080'  #  8x: sm75,  8 GB
-  - 'rtxa6000' # 12x: sm86, 48 GB
-  - 'l4'       # 48x: sm89, 24 GB
-  - 'rtx4090'  # 10x: sm89, 24 GB
-  - 'h100'     # 16x: sm90, 80 GB
+gpus:
+  - 'v100'     # ?? runners
+  - 't4'       #  8 runners
+  - 'rtx2080'  #  8 runners
+  - 'rtxa6000' # 12 runners
+  - 'l4'       # 48 runners
+  - 'rtx4090'  # 10 runners
+  - 'h100'     # 16 runners
+
+gpus_sm:
+  'v100':     '70'
+  't4':       '75'
+  'rtx2080':  '75'
+  'rtxa6000': '86'
+  'l4':       '89'
+  'rtx4090':  '89'
+  'h100':     '90'
+
+gpus_mem_gb:
+  'v100':     '32'
+  't4':       '16'
+  'rtx2080':  '8'
+  'rtxa6000': '48'
+  'l4':       '24'
+  'rtx4090':  '24'
+  'h100':     '80'
+
 testing_pool_gpus:
   - 't4'
   - 'rtx2080'
@@ -126,47 +187,3 @@ testing_pool_gpus:
   - 'l4'
   - 'rtx4090'
   - 'h100'
-
-#
-# Workflow matrices:
-#
-workflows:
-  pull_request:
-    # default_projects: nvcc
-    - {job_types: ['build'], ctk: *ctk_prev_min, host_compiler: *gcc6,                         std: [11, 14]         }
-    - {job_types: ['build'], ctk: *ctk_prev_min, host_compiler: [*gcc7, *gcc8, *gcc9, *llvm9], std: [11, 14, 17]     }
-    - {job_types: ['build'], ctk: *ctk_prev_min, host_compiler: *msvc2017,                     std: 14               }
-    - {job_types: ['build'], ctk: *ctk_prev_max, host_compiler: *gcc11,                        std: [11, 14, 17],     cmake_cuda_arch: '60;70;80;90'}
-    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: [*gcc7, *gcc8, *gcc9],         std: [11, 14, 17]     }
-    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: [*gcc10, *gcc11],              std: [11, 14, 17, 20] }
-    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: [*llvm9, *llvm10],             std: [11, 14, 17]     }
-    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: [*llvm11, *llvm12, *llvm13],   std: [11, 14, 17, 20] }
-    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: [*llvm14, *llvm15],            std: [11, 14, 17, 20] }
-    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: [*gcc12, *llvm16],             std: [11, 14, 17, 20], cpu: 'arm64'}
-    - {job_types: ['test'],  ctk: *ctk_curr,     host_compiler: [*gcc12, *llvm16],             std: [11, 14, 17, 20], cmake_cuda_arch: '60;70;80;90'}
-    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: *oneapi,                       std: [11, 14, 17]     }
-    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: *msvc2019,                     std: [14, 17]         }
-    - {job_types: ['build'], ctk: *ctk_curr,     host_compiler: *msvc2022,                     std: [14, 17, 20]     }
-    # default_projects: clang-cuda
-    - {job_types: ['build'], device_compiler: *llvm-newest, host_compiler: *llvm-newest, std: [17, 20]}
-    # nvrtc:
-    - {job_types: ['nvrtc'], project: 'libcudacxx', ctk: *ctk_curr, host_compiler: *gcc12, std: [11, 14, 17, 20]}
-    # verify-codegen:
-    - { job_types: ['verify_codegen'], project: 'libcudacxx'}
-    # cccl-infra:
-    - {job_types: ['infra'], project: 'cccl', ctk: *ctk_prev_min, host_compiler: [*gcc-oldest, *llvm-oldest]}
-    - {job_types: ['infra'], project: 'cccl', ctk: *ctk_curr,     host_compiler: [*gcc-newest, *llvm-newest]}
-  nightly:
-    - {job_types: ['test'],  ctk: *ctk_prev_min, gpu: 'v100',     cmake_cuda_arch: '70', host_compiler: *gcc6,   std: [11]             }
-    - {job_types: ['test'],  ctk: *ctk_prev_min, gpu: 't4',       cmake_cuda_arch: '75', host_compiler: *llvm9,  std: [17]             }
-    - {job_types: ['test'],  ctk: *ctk_prev_max, gpu: 'rtx2080',  cmake_cuda_arch: '75', host_compiler: *gcc11,  std: [17]             }
-    - {job_types: ['test'],  ctk: *ctk_curr,     gpu: 'rtxa6000', cmake_cuda_arch: '86', host_compiler: *gcc7,   std: [14]             }
-    - {job_types: ['test'],  ctk: *ctk_curr,     gpu: 'l4',       cmake_cuda_arch: '89', host_compiler: *gcc12,  std: [11, 14, 17, 20] }
-    - {job_types: ['test'],  ctk: *ctk_curr,     gpu: 'rtx4090',  cmake_cuda_arch: '89', host_compiler: *llvm9,  std: [11]             }
-    - {job_types: ['test'],  ctk: *ctk_curr,     gpu: 'h100',     cmake_cuda_arch: '90', host_compiler: *gcc12,  std: [11, 20]         }
-    - {job_types: ['test'],  ctk: *ctk_curr,     gpu: 'h100',     cmake_cuda_arch: '90', host_compiler: *llvm16, std: [17]             }
-    # nvrtc:
-    - {job_types: ['nvrtc'], ctk: *ctk_curr,     gpu: 't4',       cmake_cuda_arch: '75', host_compiler: *gcc12,  std: [20],             project: ['libcudacxx']}
-    - {job_types: ['nvrtc'], ctk: *ctk_curr,     gpu: 'rtxa6000', cmake_cuda_arch: '86', host_compiler: *gcc12,  std: [20],             project: ['libcudacxx']}
-    - {job_types: ['nvrtc'], ctk: *ctk_curr,     gpu: 'l4',       cmake_cuda_arch: '89', host_compiler: *gcc12,  std: [11, 14, 17, 20], project: ['libcudacxx']}
-    - {job_types: ['nvrtc'], ctk: *ctk_curr,     gpu: 'h100',     cmake_cuda_arch: '90', host_compiler: *gcc12,  std: [11, 20],         project: ['libcudacxx']}