Conversation
…a/gtc_2025_tutorials
…custom scalar_ops function
…to tylera/gtc_2025_tutorials
|
/build |
Greptile OverviewGreptile SummaryThis PR adds comprehensive GTC 2025 tutorial materials for MatX, including three new lab notebooks covering introduction, kernel fusion, and profiling topics. The changes reorganize the tutorial structure to support in-notebook execution via a new IPython extension ( Major Changes
Issues Found
The tutorial content is well-structured with clear progression from basics to advanced topics, comprehensive NVTX instrumentation for profiling, and practical radar signal processing examples. Confidence Score: 3/5
Important Files Changed
Sequence DiagramsequenceDiagram
participant User
participant JupyterNotebook
participant RunMatxExtension
participant NVCC
participant CompiledBinary
participant GPU
User->>JupyterNotebook: Start GTC Lab Tutorial
JupyterNotebook->>User: Load 01_lab_intro.ipynb
User->>JupyterNotebook: Execute %%run_matx cell
JupyterNotebook->>RunMatxExtension: Pass MatX code
RunMatxExtension->>RunMatxExtension: Wrap code in main()
RunMatxExtension->>RunMatxExtension: Write to /tmp/output.cu
RunMatxExtension->>NVCC: Compile with nvcc
NVCC->>CompiledBinary: Generate /tmp/output
CompiledBinary->>GPU: Execute on GPU
GPU->>CompiledBinary: Return results
CompiledBinary->>RunMatxExtension: stdout/stderr
RunMatxExtension->>JupyterNotebook: Display output
JupyterNotebook->>User: Show results
User->>JupyterNotebook: Progress to 02_lab_fusion.ipynb
User->>JupyterNotebook: Run kernel_fusion sample
JupyterNotebook->>CompiledBinary: Execute pre-built binary
CompiledBinary->>GPU: Run unfused kernels
GPU->>CompiledBinary: Measure performance
CompiledBinary->>GPU: Run fused operation
GPU->>CompiledBinary: Measure performance
CompiledBinary->>User: Compare results
User->>JupyterNotebook: Progress to 03_lab_profiling.ipynb
User->>JupyterNotebook: Profile simple_radar_pipeline
JupyterNotebook->>CompiledBinary: Run with NVTX markers
CompiledBinary->>GPU: Execute radar pipeline stages
Note over CompiledBinary,GPU: PulseCompression<br/>ThreePulseCanceller<br/>DopplerProcessing<br/>CFARDetections
GPU->>CompiledBinary: Performance data
CompiledBinary->>User: Display profiling results
|
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
Additional Comments (2)
|
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
Co-authored-by: greptile-apps[bot] <165735046+greptile-apps[bot]@users.noreply.github.com>
| MATX_EXIT_HANDLER(); | ||
| MATX_EXIT_HANDLER(); |
There was a problem hiding this comment.
duplicate MATX_EXIT_HANDLER() call on lines 44-45
| MATX_EXIT_HANDLER(); | |
| MATX_EXIT_HANDLER(); | |
| MATX_EXIT_HANDLER(); | |
| } |
| MATX_EXIT_HANDLER(); | ||
| MATX_EXIT_HANDLER(); |
There was a problem hiding this comment.
duplicate MATX_EXIT_HANDLER() call on lines 44-45
| MATX_EXIT_HANDLER(); | |
| MATX_EXIT_HANDLER(); | |
| MATX_EXIT_HANDLER(); | |
| } |
|
|
||
|
|
||
| #FROM ghcr.io/nvidia/matx/production:latest AS devel | ||
| FROM gitlab-master.nvidia.com:5005/devtech-compute/sigx-group/container/build:12.8_x86_64_ubuntu22.04-amd64 AS devel |
There was a problem hiding this comment.
using internal NVIDIA GitLab registry instead of public image - verify this is intended for external use
| stderr_thread.join() | ||
| stdout_thread.join() | ||
|
|
||
| cur_line = 0 |
There was a problem hiding this comment.
resetting cur_line = 0 after process completion has no effect - this variable goes out of scope immediately after
Additional Comments (2)
|
No description provided.