Paper on CuriousCoding

QuadRank: Engineering a High Throughput Rank

Thu, 29 Jan 2026 00:00:00 +0100

Table of Contents

Abstract
1 Introduction
2 Background
- 2.1 Further implementations
3 BiRank
- 3.1 Variants
4 QuadRank
- 4.1 Variants
5 Results
- 5.1 BiRank
- 5.2 QuadRank
6 Conclusion
7 Acknowledgements
8 Code snippets
9 Pairing superblocks
10 Additional results
- 10.1 Cache misses per query
- 10.2 Throughput for small inputs
- 10.3 AMD EPYC evals
11 QuadFm: A Batching FM-index
- 11.1 Results
12 Further code optimization ideas

\[\newcommand{\rank}{\mathsf{rank}} \newcommand{\rankone}{\mathsf{rank}} \newcommand{\rankall}{\mathsf{rank_4}}\]

QuickHeap: the fastest priority queue

Mon, 11 Aug 2025 00:00:00 +0200

Table of Contents

1 Background
2 Literature on Quickheaps
3 Bucket-based implementation
- Data structure
- Push
- Pop
- Partition
4 Results
5 Conclusion

Backlinks: bsky, X, hacker news, lobste.rs

A preprint on the SimdQuickHeap (github:RagnarGrootKoerkamp/QuickHeap) with the latest results can now be found on arXiv (DOI, PDF):

PtrHash: Minimal Perfect Hashing at RAM Throughput

Mon, 03 Feb 2025 00:00:00 +0100

Table of Contents

Abstract
1 Introduction
2 Related work
3 PtrHash
- 3.1 Overview
- 3.2 Construction
- 3.3 Bucket Assignment Functions
- 3.4 Remapping using CacheLineEF
4 Results
- 4.1 Construction
  - 4.1.1 Bucket Functions
  - 4.1.2 Tuning Parameters for Construction
  - 4.1.3 Remap
- 4.2 Comparison to Other Methods
5 Conclusions and Future Work
Appendix A: Query Throughput
Appendix B: Sharding
- Evaluation

This is the online version of my SEA 2025 paper on PtrHash (DOI, PDF). The original development-log can be found here.

SimdMinimizers: Computing random minimizers, fast

Fri, 12 Jul 2024 00:00:00 +0200

Table of Contents

1 Introduction
- 1.1 Results
2 Random minimizers
3 Algorithms
4 Analysing what we have so far
5 Rolling our own hash
6 SIMD sliding window
- 6.1 Results
  - Human genome results
7 Extending into something useful
- 7.1 Collecting minimizer positions
- 7.2 Deduplicating the minimizer positions
- 7.3 Super-k-mers
- 7.4 Canonical k-mers
- 7.5 AntiLex hash
8 Conclusion
- 8.1 Future work

SimdMinimizers has been published as a paper: DOI, PDF:

A near-tight lower bound on minimizer density

Tue, 25 Jun 2024 00:00:00 +0200

Table of Contents

Succinct background
- Definitions
- Lower bounds
A new lower bound
Discussion
Post scriptum
Acknowledgement

The results of this post are now published in Bioinformatics: DOI, PDF:

Kille, Bryce, Ragnar Groot Koerkamp, Drake McAdams, Alan Liu, and Todd J Treangen. 2024. “A near-tight Lower Bound on the Density of Forward Sampling Schemes.” Edited by Yann Ponty. Bioinformatics, December. https://doi.org/10.1093/bioinformatics/btae736.

This content has also been absorbed into my thesis chapter on minimizers.

A*PA2: Up to 19x faster exact global alignment

Sat, 23 Mar 2024 00:00:00 +0100

Table of Contents

Abstract
1 Introduction
- 1.1 Contributions
- 1.2 Previous work
  - 1.2.1 Needleman-Wunsch
  - 1.2.2 Graph algorithms
  - 1.2.3 Computational volumes
  - 1.2.4 Parallelism
  - 1.2.5 Tools
2 Preliminaries
3 Methods
- 3.1 Band-doubling
- 3.2 Blocks
- 3.3 Memory
- 3.4 SIMD
- 3.5 SIMD-friendly sequence profile
- 3.6 Traceback
- 3.7 A*
  - 3.7.1 Bulk-contours update
  - 3.7.2 Pre-pruning
- 3.8 Determining the rows to compute
  - 3.8.1 Sparse heuristic invocation
- 3.9 Incremental doubling
4 Results
- 4.1 Setup
- 4.2 Comparison with other aligners
- 4.3 Effects of methods
5 Discussion
Acknowledgements
Conflict of interest
6 Appendix
- 6.1 Bitpacking
- 6.2 Comparison with other aligners
- 6.3 Effects of methods

\begin{equation*} \newcommand{\g}{g^*} \newcommand{\h}{h^*} \newcommand{\f}{f^*} \newcommand{\cgap}{c_{\textrm{gap}}} \newcommand{\xor}{\ \mathrm{xor}\ } \newcommand{\and}{\ \mathrm{and}\ } \newcommand{\st}[2]{\langle #1, #2\rangle} \newcommand{\matches}{\mathcal M} \end{equation*}