Added the material for XGBoost optimization - #30
Conversation
|
Since the latest changes still have perf data, it cannot be approved till we get perf claim pre-requisites fulfilled. |
david-cortes-intel
left a comment
There was a problem hiding this comment.
General comment: this guide says 'xgboost', but it is limited to predictions/inference, while a similar guide could also be done for training, covering details like threading, hyperparameters to try, and similar.
razdoburdin
left a comment
There was a problem hiding this comment.
please update installation instructions and consider switching to the actual versions of the software.
…-learn, removing memory allocator section, and clarifying the scope to include all 3 methods
@rsiyer-intel Updated the doc with PDT approved data |
Done and done |
…e mention of undefined default, removed unnecessary symmetric tree mention
| daal4py's vectorized tree traversal is [backend-bound](https://www.intel.com/content/www/us/en/docs/vtune-profiler/cookbook/2023-0/top-down-microarchitecture-analysis-method.html) — whether the bottleneck is core execution units or memory bandwidth, adding hyperthreads increases resource contention on the shared physical core, harming performance. | ||
|
|
||
| > **Cloud instance note:** On AWS and GCP, each vCPU does not necessarily map to a hyperthread. Smaller instance sizes use soft partitioning, so you may not know how many physical cores vs. hyperthreads you are getting. The guidance below applies most directly to bare-metal or dedicated-host instances where the physical topology is known. On shared instances, benchmark with your specific instance size to determine whether pinning provides a benefit. | ||
| > **Cloud instance note:** Depending on CSP deployment choices, smaller instance sizes may not expose how many physical cores vs. hyperthreads are provided. The guidance below applies most directly to bare-metal or dedicated-host instances where the physical topology is known. On shared instances, benchmark with your specific instance size to determine whether pinning provides a benefit. |
There was a problem hiding this comment.
That information is always retrievable at runtime, through tools like lscpu or python-level tools like joblib.
adgubrud
left a comment
There was a problem hiding this comment.
Thank you for making the requested changes! A couple of small details to resolve.
| - [MySQL & PostgreSQL](software/mysql-postgresql/README.md) | ||
| - [R (Rlang / Rstats)](software/R/README.md) | ||
| - [scikit-learn](software/scikit-learn/README.md) | ||
| - [XGBoost](software/xgboost/README.ms) |
There was a problem hiding this comment.
We just recently alphabetized the TOC. Can you please move this entry to be in alphabetical order?
| - [MySQL & PostgreSQL](software/mysql-postgresql/README.md) | ||
| - [R (Rlang / Rstats)](software/R/README.md) | ||
| - [scikit-learn](software/scikit-learn/README.md) | ||
| - [XGBoost](software/xgboost/README.ms) |
There was a problem hiding this comment.
Link is broken
| - [XGBoost](software/xgboost/README.ms) | |
| - [XGBoost](software/xgboost/README.md) |
Added the materials for XGBoost optimization. Please review and give me your feedback.