test: add fake NVML backed for GPU dev/testing#2420
test: add fake NVML backed for GPU dev/testing#2420vprashar2929 wants to merge 1 commit intosustainable-computing-io:mainfrom
Conversation
|
|
Codecov Report❌ Patch coverage is
Additional details and impacted files@@ Coverage Diff @@
## main #2420 +/- ##
==========================================
- Coverage 91.91% 91.64% -0.27%
==========================================
Files 54 55 +1
Lines 5365 5449 +84
==========================================
+ Hits 4931 4994 +63
- Misses 313 332 +19
- Partials 121 123 +2 ☔ View full report in Codecov by Sentry. 🚀 New features to boost your workflow:
|
|
📊 Profiling reports are ready to be viewed
💻 CPU Comparison with base Kepler💾 Memory Comparison with base Kepler (Inuse)💾 Memory Comparison with base Kepler (Alloc)⬇️ Download the Profiling artifacts from the Actions Summary page 📦 Artifact name: 🔧 Or use GitHub CLI to download artifacts: gh run download 22039576676 -n profile-artifacts-2420 |
8fa969d to
9dc0f33
Compare
|
📊 Profiling reports are ready to be viewed
💻 CPU Comparison with base Kepler💾 Memory Comparison with base Kepler (Inuse)💾 Memory Comparison with base Kepler (Alloc)⬇️ Download the Profiling artifacts from the Actions Summary page 📦 Artifact name: 🔧 Or use GitHub CLI to download artifacts: gh run download 22039657547 -n profile-artifacts-2420 |
This commit adds a FakeNVMLBackend that implements the NVMLBackend/NVMLDevice interfaces, enabling GPU feature development and e2e testing without real NVIDIA hardware. - Add FakeNVMLBackend and FakeNVMLDevice in nvidia package - Add WithNVMLBackend() collector option for backend injection - Add dev.fake-gpu-meter config (enabled, deviceCount, sharingMode) - Wire fake backend in createGPUMeters() when config is enabled - Add bare-metal e2e GPU tests (metrics presence, labels, conservation) - Add fake-gpu-meter config to all deployment manifests - Update e2e testing docs with gpu_test.go Signed-off-by: vprashar2929 <vibhu.sharma2929@gmail.com>
9dc0f33 to
fd8649e
Compare
|
📊 Profiling reports are ready to be viewed
💻 CPU Comparison with base Kepler💾 Memory Comparison with base Kepler (Inuse)💾 Memory Comparison with base Kepler (Alloc)⬇️ Download the Profiling artifacts from the Actions Summary page 📦 Artifact name: 🔧 Or use GitHub CLI to download artifacts: gh run download 22051110203 -n profile-artifacts-2420 |
This commit adds a FakeNVMLBackend that implements the NVMLBackend/NVMLDevice interfaces, enabling GPU feature development and e2e testing without real NVIDIA hardware.