Implemented QuART_tail, QuART_xtail, QuART_lil and refactored codebase

[cangokmen]

Changed files: ART.h, ArtNode.h, .gitignore, CMakeLists.txt, README.md
Created files: ArtNodeNewMethods.cpp, QuARTVariants/QuART_tail.h, QuARTVariants/QuART_xtail.h, QuARTVariants/QuART_lil.h, benchmarks/profile_inserts.cpp, benchmarks/art.cpp, benchmarks/quart_tail.cpp, benchmarks/quart_xtail.cpp, benchmarks/quart_lil.cpp, run_experiments.sh

• ART.h: Modified insert function names to make them distinguishable when profiling. Added printTree method that pretty-prints the tree. Added verifyTailPath method that returns whether the tail path is true (can be used in QuART_tail and QuART_xtail trees for verification).
• ArtNode.h: Added declarations of insertNodeX methods used in fast path insertion. Added printFpPath method that pretty-prints a node array.
• ArtNodeNewMethods.cpp: Implemented insertNodeX functions used in fast path insertion. Follow the comments to see which part does what.
• QuARTVariants/QuART_tail.h: Implemented the generic tail insert.
• QuARTVariants/QuART_xtail.h: Implemented a fp method similar to tail insert but faster (name recommendations are greatly appreciated).
• QuARTVariants/QuART_lil.h: Implemented the generic lil insert.
• benchmarks/profile_inserts.cpp: Fragment with only tree inserts, use this during profiling. It doesn't read input file, only inserts integers in sorted order from 1 to 10000000. To profile, run perf record ./profile_inserts -v -N - 10000000 (N < 10000000), then perf report.
• benchmarks/art.cpp: Created executable that will run the base ART algorithm.
• benchmarks/quart_tail.cpp: Created executable that will run the tail insert algorithm.
• benchmarks/quart_xtail.cpp: Created executable that will run the xtail insert algorithm.
• benchmarks/quart_lil.cpp: Created executable that will run the lil insert algorithm.
• run_experiments.sh: Runs the benchmark experiments.
• .gitignore: Updated gitignore so that it ignores the results folder.
• CMakeList.txt: Added necessary lines to use perf, and added the executables for benchmark files.
• README.md: Updated README with the new changes. It explains how to run the experiments too.

[Copilot]

Pull Request Overview

This PR introduces three new QuART variants (tail, xtail, lil) with fast-path insertion strategies, refactors the core ART implementation to support path printing/verification, and adds benchmark tooling along with documentation updates.

• Implemented QuART_tail, QuART_xtail, and QuART_lil insert methods in the QuARTVariants directory
• Extended ART core (ART.h, ArtNode.h, ArtNodeNewMethods.cpp) to initialize and track fast-path (fp), printing, and verification
• Added per-variant benchmark executables, an experiment script (run_experiments.sh), updated CMake targets, and refreshed README.md

Reviewed Changes

Copilot reviewed 15 out of 17 changed files in this pull request and generated 4 comments.

Show a summary per file

File	Description
run_experiments.sh	Script to automate running all variants over workloads
benchmarks/quart_*.cpp	Added benchmark drivers for xtail, tail, and lil variants
benchmarks/profile_inserts.cpp	Benchmark-only insert driver; loads sequential keys
benchmarks/art.cpp	Updated ART benchmark to default N=10M and removed verbose logic
CMakeLists.txt	New targets for all benchmarks; adjusted compiler flags
README.md	Documented new executables, script usage, and variant descriptions
QuARTVariants/*.h	Core implementations for QuART_tail, QuART_xtail, QuART_lil
ART.h, ArtNode.h	Initialized members for fast-path, added printTree & verifyTailPath
ArtNodeNewMethods.cpp	Implemented fast-path insertNodeX methods and debugging helpers

Comments suppressed due to low confidence (3)

benchmarks/profile_inserts.cpp:39

• [nitpick] The comment above mentions changing the tree to ART::ART, but the code still instantiates QuART_xtail. Update the comment or the code to match the intended tree type.

    ART::QuART_xtail* tree = new ART::QuART_xtail();

README.md:52

• [nitpick] The README still documents a '-v' flag for executables, but the benchmark code (e.g., art.cpp) no longer supports verbose mode. Remove or update references to '-v'.

- `-v`: Verbose mode (optional, default = false)

run_experiments.sh:1

• [nitpick] Add a shebang (e.g., #!/usr/bin/env bash) at the top of the script and ensure that variables like LOGDIR, RESULTS, SUFFIX, and REPEAT are defined or passed in before use.

for FILE in ../bods/workloads/workload_N*_K*_L*.bin; do

[ramananeesh]

please address the changes. Also, make sure you use .clang-format to format each of your files.

[ramananeesh]

get rid of this line. Make sure you are formatting the files using the clang-format file at the base directory in the repo

[cangokmen]

I'll do the formatting on Thursday

[ramananeesh]

remove empty line

[cangokmen]

I'll do the formatting on Thursday

[ramananeesh]

remove empty line.

[cangokmen]

I'll do the formatting on Thursday

[ramananeesh]

why do we have QuART_lil in this PR?

[cangokmen]

I can put it in a seperate PR, it was easy to implement after having the tail implementation

[ramananeesh]

Always, keep PRs atomic so that it is easy to track changes. Coupling multiple features will only make the codebase messy if we need to rollback.

Also, since this was something @NoahP-K was trying to implement, move this to a separate branch and collaborate with him on this. We can make a new PR.

[ramananeesh]

we can use a single benchmark file that can compile different trees, instead of duplicated benchmark code.

[cangokmen]

Should I do this in this PR or the next PR?

[ramananeesh]

Up to you. Technically, if these files are created in this PR, you can fix it here. However, if you feel that getting the rest of the features merged into main will be beneficial for the whole group, you can choose to do it in a separate PR as well.