andygrove opened a new pull request, #4873:
URL: https://github.com/apache/datafusion-comet/pull/4873

   ## Which issue does this PR close?
   
   N/A. Documentation update.
   
   ## Rationale for this change
   
   The Comet vs Gluten comparison document referenced an older single-node 
TPC-H benchmark for performance and did not surface several of Comet's 
differentiating strengths. This updates the performance comparison to cite 
independent third-party benchmarking and adds sections covering areas where 
Comet has meaningful advantages.
   
   ## What changes are included in this PR?
   
   - **Performance**: Replace the single-node TPC-H numbers with a reference to 
the independent AWS Labs benchmark comparing Comet 0.16.0 with Gluten 1.6.0 on 
TPC-DS 3TB, which found similar overall performance with Gluten roughly 9% 
faster, and note the wide per-query variation and our expectation that the gap 
closes over time.
   - **Codegen Dispatch**: New section describing Comet's codegen dispatcher, 
which runs Spark's own generated code inside the native pipeline for 
correctness-by-default and to avoid full query-stage fallbacks. Gluten has no 
direct equivalent.
   - **Regular Expression Compatibility**: New section explaining that Comet is 
100% compatible with Spark regex by default (via the codegen dispatcher running 
the JVM `java.util.regex` engine), whereas Velox's C++ RE2 engine cannot match 
Java regex semantics such as backreferences and lookaround.
   - **Scala and Java UDF Support**: New section noting that Comet runs 
existing JVM Scala/Java scalar UDFs inside the native pipeline without rewrites 
and without forcing the enclosing operator back to Spark.
   - **Underlying Execution Engine**: Expand the Rust vs C++ discussion to 
cover Rust's compile-time memory-safety guarantees.
   - **Community and Governance**: New section on Comet being built on Apache 
DataFusion, a vendor-neutral ASF community, versus Velox being primarily 
Meta-driven.
   - Update the Summary to reflect the revised performance framing and the 
additional strengths.
   
   ## How are these changes tested?
   
   Documentation only. No tests required.
   


-- 
This is an automated message from the Apache Git Service.
To respond to the message, please log on to GitHub and use the
URL above to go to the specific comment.

To unsubscribe, e-mail: [email protected]

For queries about this service, please contact Infrastructure at:
[email protected]


---------------------------------------------------------------------
To unsubscribe, e-mail: [email protected]
For additional commands, e-mail: [email protected]

Reply via email to