Restore resource_cast memory limit check#1231
Open
bdice wants to merge 1 commit into
Open
Conversation
📝 WalkthroughWalkthroughThis PR enhances CUDA device memory reporting by adding support for RMM memory limiting adaptors. The header gains new dependencies for memory resource types, and ChangesDevice memory reporting with RMM limiting adaptors
Estimated code review effort🎯 2 (Simple) | ⏱️ ~10 minutes 🚥 Pre-merge checks | ✅ 5✅ Passed checks (5 passed)
✏️ Tip: You can configure your own custom pre-merge checks in the settings. ✨ Finishing Touches🧪 Generate unit tests (beta)
Comment |
![coderabbitai[bot]](https://avatars.githubusercontent.com/in/347564?s=60&v=4){kind=small}
There was a problem hiding this comment.
Actionable comments posted: 1
🧹 Nitpick comments (1)
cpp/src/utilities/cuda_helpers.cuh (1)
254-259: ⚡ Quick winGate debug prints behind a debug flag.
Unconditional
printfinget_device_memory_size()can spam stdout and add sync overhead when this helper is called frequently. Consider guarding these lines with a compile-time debug flag.Suggested patch
- printf("limiting_adaptor->get_allocation_limit(): %fMiB\n", - limiting_adaptor->get_allocation_limit() / (double)1e6); - printf("used_mem: %fMiB\n", limiting_adaptor->get_allocated_bytes() / (double)1e6); - printf("free_mem: %fMiB\n", - (limiting_adaptor->get_allocation_limit() - limiting_adaptor->get_allocated_bytes()) / - (double)1e6); +#ifdef CUOPT_ENABLE_RESOURCE_LIMIT_DEBUG + printf("limiting_adaptor->get_allocation_limit(): %fMiB\n", + limiting_adaptor->get_allocation_limit() / (double)1e6); + printf("used_mem: %fMiB\n", limiting_adaptor->get_allocated_bytes() / (double)1e6); + printf("free_mem: %fMiB\n", + (limiting_adaptor->get_allocation_limit() - limiting_adaptor->get_allocated_bytes()) / + (double)1e6); +#endif🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the rest with a brief reason, keep changes minimal, and validate. In `@cpp/src/utilities/cuda_helpers.cuh` around lines 254 - 259, The three unconditional printf calls inside get_device_memory_size() (printing limiting_adaptor->get_allocation_limit(), get_allocated_bytes(), and free memory) should be wrapped in a compile-time debug macro so they don't run in production; add a macro guard (e.g., CUDA_HELPERS_DEBUG or similar) around those prints in cpp/src/utilities/cuda_helpers.cuh and only emit the diagnostics when the macro is defined, keeping the rest of get_device_memory_size() unchanged. Ensure the macro name is documented in the header comment so callers can enable it at build time.
🤖 Prompt for all review comments with AI agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
Inline comments:
In `@cpp/src/utilities/cuda_helpers.cuh`:
- Around line 254-259: The printed memory diagnostics use the label "MiB" but
divide by 1e6 (decimal MB); update the three printf calls that reference
limiting_adaptor->get_allocation_limit() and
limiting_adaptor->get_allocated_bytes() to divide by 1024.0 * 1024.0 (or use a
named constant like BYTES_PER_MIB) so the units correctly reflect mebibytes, and
keep the "MiB" labels unchanged.
---
Nitpick comments:
In `@cpp/src/utilities/cuda_helpers.cuh`:
- Around line 254-259: The three unconditional printf calls inside
get_device_memory_size() (printing limiting_adaptor->get_allocation_limit(),
get_allocated_bytes(), and free memory) should be wrapped in a compile-time
debug macro so they don't run in production; add a macro guard (e.g.,
CUDA_HELPERS_DEBUG or similar) around those prints in
cpp/src/utilities/cuda_helpers.cuh and only emit the diagnostics when the macro
is defined, keeping the rest of get_device_memory_size() unchanged. Ensure the
macro name is documented in the header comment so callers can enable it at build
time.
🪄 Autofix (Beta)
Fix all unresolved CodeRabbit comments on this PR:
- Push a commit to this branch (recommended)
- Create a new PR with the fixes
ℹ️ Review info
⚙️ Run configuration
Configuration used: Path: .coderabbit.yaml
Review profile: CHILL
Plan: Enterprise
Run ID: 430867ae-8a10-47b2-948f-2993d59e032c
📒 Files selected for processing (1)
cpp/src/utilities/cuda_helpers.cuh
Comment on lines
+254
to
+259
| printf("limiting_adaptor->get_allocation_limit(): %fMiB\n", | ||
| limiting_adaptor->get_allocation_limit() / (double)1e6); | ||
| printf("used_mem: %fMiB\n", limiting_adaptor->get_allocated_bytes() / (double)1e6); | ||
| printf("free_mem: %fMiB\n", | ||
| (limiting_adaptor->get_allocation_limit() - limiting_adaptor->get_allocated_bytes()) / | ||
| (double)1e6); |
There was a problem hiding this comment.
Fix unit mismatch in memory diagnostics.
Line 254, Line 256, and Line 257 label values as MiB but divide by 1e6 (MB). Use 1024.0 * 1024.0 or relabel to MB.
Suggested patch
- printf("limiting_adaptor->get_allocation_limit(): %fMiB\n",
- limiting_adaptor->get_allocation_limit() / (double)1e6);
- printf("used_mem: %fMiB\n", limiting_adaptor->get_allocated_bytes() / (double)1e6);
+ constexpr double bytes_per_mib = 1024.0 * 1024.0;
+ printf("limiting_adaptor->get_allocation_limit(): %fMiB\n",
+ limiting_adaptor->get_allocation_limit() / bytes_per_mib);
+ printf("used_mem: %fMiB\n", limiting_adaptor->get_allocated_bytes() / bytes_per_mib);
printf("free_mem: %fMiB\n",
- (limiting_adaptor->get_allocation_limit() - limiting_adaptor->get_allocated_bytes()) /
- (double)1e6);
+ (limiting_adaptor->get_allocation_limit() - limiting_adaptor->get_allocated_bytes()) /
+ bytes_per_mib);📝 Committable suggestion
‼️ IMPORTANT
Carefully review the code before committing. Ensure that it accurately replaces the highlighted code, contains no missing lines, and has no issues with indentation. Thoroughly test & benchmark the code to ensure it meets the requirements.
Suggested change
| printf("limiting_adaptor->get_allocation_limit(): %fMiB\n", | |
| limiting_adaptor->get_allocation_limit() / (double)1e6); | |
| printf("used_mem: %fMiB\n", limiting_adaptor->get_allocated_bytes() / (double)1e6); | |
| printf("free_mem: %fMiB\n", | |
| (limiting_adaptor->get_allocation_limit() - limiting_adaptor->get_allocated_bytes()) / | |
| (double)1e6); | |
| constexpr double bytes_per_mib = 1024.0 * 1024.0; | |
| printf("limiting_adaptor->get_allocation_limit(): %fMiB\n", | |
| limiting_adaptor->get_allocation_limit() / bytes_per_mib); | |
| printf("used_mem: %fMiB\n", limiting_adaptor->get_allocated_bytes() / bytes_per_mib); | |
| printf("free_mem: %fMiB\n", | |
| (limiting_adaptor->get_allocation_limit() - limiting_adaptor->get_allocated_bytes()) / | |
| bytes_per_mib); |
🤖 Prompt for AI Agents
Verify each finding against current code. Fix only still-valid issues, skip the
rest with a brief reason, keep changes minimal, and validate.
In `@cpp/src/utilities/cuda_helpers.cuh` around lines 254 - 259, The printed
memory diagnostics use the label "MiB" but divide by 1e6 (decimal MB); update
the three printf calls that reference limiting_adaptor->get_allocation_limit()
and limiting_adaptor->get_allocated_bytes() to divide by 1024.0 * 1024.0 (or use
a named constant like BYTES_PER_MIB) so the units correctly reflect mebibytes,
and keep the "MiB" labels unchanged.
bdice
added
non-breaking
51 tasks
Contributor
|
Requesting @rg20 's review |
This file contains hidden or bidirectional Unicode text that may be interpreted or compiled differently than what appears below. To review, open the file in an editor that reveals hidden Unicode characters.
Learn more about bidirectional Unicode characters
Sign up for free
to join this conversation on GitHub.
Already have an account?
Sign in to comment
2 participants
Add this suggestion to a batch that can be applied as a single commit.This suggestion is invalid because no changes were made to the code.Suggestions cannot be applied while the pull request is closed.Suggestions cannot be applied while viewing a subset of changes.Only one suggestion per line can be applied in a batch.Add this suggestion to a batch that can be applied as a single commit.Applying suggestions on deleted lines is not supported.You must change the existing code in this line in order to create a valid suggestion.Outdated suggestions cannot be applied.This suggestion has been applied or marked resolved.Suggestions cannot be applied from pending reviews.Suggestions cannot be applied on multi-line comments.Suggestions cannot be applied while the pull request is queued to merge.Suggestion cannot be applied right now. Please check back later.
Description
Restore the memory-limit-aware branch in
get_device_memory_size()that was temporarily removed during the CCCL/RMM memory-resource migration in #1035.The function now uses
cuda::mr::resource_castonrmm::mr::get_current_device_resource_ref()to detect an activermm::mr::limiting_resource_adaptor. When present, it reports the limit/usage/free values and returns the smaller of CUDA total memory and the adaptor allocation limit, matching the previous behavior.