GPU-accelerated TLS, optimized BLS, cuFINUFFT LS, and multi-GPU benchmarks#56

Open

johnh2o2 wants to merge 124 commits intomasterfrom

testing/runpod-benchmarks

Commits on Oct 9, 2025

Commits on Oct 10, 2025

Merge pull request #38 from xiaziyna/copilot/add-nufft-lrt-feature
johnh2o2
authored

Commits on Oct 14, 2025

Initial plan
Copilot
committed
Add comprehensive PyCUDA technology assessment and findings

Copilot
and
johnh2o2
committed
Add getting started guide and visual summary for assessment

Copilot
and
johnh2o2
committed
Add comprehensive index for all assessment documents

Copilot
and
johnh2o2
committed
Phase 1: Drop Python 2.7 support and modernize codebase

Copilot
and
johnh2o2
committed
Phase 2: Add CI/CD, documentation, and version bump to 0.4.0

Copilot
and
johnh2o2
committed
Add comprehensive migration guide and documentation index

Copilot
and
johnh2o2
committed
Make changes compatible with v1.0 base branch

Copilot
and
johnh2o2
committed

Commits on Oct 17, 2025

Commits on Oct 25, 2025

Improve sparse BLS implementation and add RunPod development support

authored and
John Hoffman
committed
Add GPU-accelerated sparse BLS implementation

authored and
John Hoffman
committed
Update scripts/sync-to-runpod.sh

johnh2o2
and
Copilot
authored
Update scripts/test-remote.sh

johnh2o2
and
Copilot
authored
Update scripts/setup-remote.sh

johnh2o2
and
Copilot
authored
Merge pull request #45 from johnh2o2/fix/sparse-bls-wrapped-transits
johnh2o2
authored
Merge branch 'v1.0' into feature/sparse-bls-gpu

John Hoffman
authored and
John Hoffman
committed
Update minimum Python version from 3.7 to 3.8

authored and
John Hoffman
committed
Update cuvarbase/kernels/sparse_bls_simple.cu

johnh2o2
and
Copilot
authored
Update cuvarbase/kernels/sparse_bls_simple.cu

johnh2o2
and
Copilot
authored
Update cuvarbase/kernels/sparse_bls.cu

johnh2o2
and
Copilot
authored
Update cuvarbase/kernels/sparse_bls.cu

johnh2o2
and
Copilot
authored
Update cuvarbase/bls.py

johnh2o2
and
Copilot
authored
Merge pull request #46 from johnh2o2/feature/sparse-bls-gpu
johnh2o2
authored
Add comprehensive algorithm benchmarking suite

authored and
John Hoffman
committed
Add persistent benchmark runner and time estimator

authored and
John Hoffman
committed
Add example benchmark results showing 315x speedup

authored and
John Hoffman
committed
Add comprehensive TESS catalog BLS cost analysis

authored and
John Hoffman
committed
WIP: BLS kernel optimization - baseline and analysis

authored and
John Hoffman
committed
Add optimized BLS kernel with bank conflict fixes and warp shuffles

John Hoffman
authored and
John Hoffman
committed
Fix warp shuffle reduction bug in optimized BLS kernel

John Hoffman
authored and
John Hoffman
committed
Complete BLS kernel optimization work with results documentation

John Hoffman
authored and
John Hoffman
committed
Add adaptive BLS with dynamic block sizing

John Hoffman
authored and
John Hoffman
committed
Add realistic batch Keplerian BLS benchmark

John Hoffman
authored and
John Hoffman
committed
Complete adaptive BLS implementation with dramatic results

John Hoffman
authored and
John Hoffman
committed
Add baseline verification script

John Hoffman
authored and
John Hoffman
committed
Add GPU utilization analysis and architecture comparison

John Hoffman
authored and
John Hoffman
committed
Add comprehensive PR summary

John Hoffman
authored and
John Hoffman
committed
Fix adaptive benchmark to use Keplerian frequency grids

John Hoffman
authored and
John Hoffman
committed
Update PR summary to reflect Keplerian grids

John Hoffman
authored and
John Hoffman
committed
Add thread-safety and LRU eviction to kernel cache

authored and
John Hoffman
committed
Update PR summary with code quality improvements

authored and
John Hoffman
committed
Add comprehensive documentation for code quality fixes

authored and
John Hoffman
committed
Remove PR_SUMMARY.md (moved to GitHub PR description)

authored and
John Hoffman
committed

Commits on Oct 26, 2025

Merge pull request #47 from johnh2o2/feature/benchmark-suite
johnh2o2
authored
Merge pull request #49 from johnh2o2/feature/optimize-bls-kernel
johnh2o2
authored
Improve README: highlight BLS optimization and credit Jamila Taaki

authored and
John Hoffman
committed
Correct sparse BLS citation and remove cost implications

authored and
John Hoffman
committed
Merge pull request #50 from johnh2o2/readme-improvements
johnh2o2
authored
Enable GPU sparse BLS by default in eebls_transit

authored and
John Hoffman
committed
Update cuvarbase/bls.py

johnh2o2
and
Copilot
authored
Update cuvarbase/bls.py

johnh2o2
and
Copilot
authored
Merge pull request #51 from johnh2o2/sparse-bls-gpu-default
johnh2o2
authored

Commits on Oct 27, 2025

Repository cleanup: consolidate docs and organize tests

authored and
John Hoffman
committed
Move JSON files to analysis/ and remove cleanup history doc

authored and
John Hoffman
committed
Merge pull request #52 from johnh2o2/repository-cleanup
johnh2o2
authored
Phase 1: TLS GPU implementation - Core infrastructure

authored and
John Hoffman
committed
Phase 2: TLS GPU optimization - Advanced features

authored and
John Hoffman
committed
Phase 3: TLS production features - Statistics & usability

authored and
John Hoffman
committed
Fix TLS GPU kernel compilation and add GPU test script

authored and
John Hoffman
committed
Document RunPod GPU testing issues and solutions

authored and
John Hoffman
committed
Fix period grid generation in tls_grids.py

authored and
John Hoffman
committed
Fix critical Ofir period grid generation bugs

authored and
John Hoffman
committed
Fix critical TLS GPU bugs: Ofir grid, duration scaling, and Thrust sorting

authored and
John Hoffman
committed
Consolidate TLS to single performant kernel

authored and
John Hoffman
committed
Add Keplerian-aware duration constraints for TLS

authored and
John Hoffman
committed
Wire up Keplerian TLS Python API

authored and
John Hoffman
committed
Add PR description markdown file for easy copying

authored and
John Hoffman
committed
Clean up TLS test files and update README

authored and
John Hoffman
committed
Reorganize TLS documentation and test files

authored and
John Hoffman
committed
Update cuvarbase/tls_models.py

johnh2o2
and
Copilot
authored
Address PR review comments for TLS implementation

authored and
John Hoffman
committed
Replace insertion sort with bitonic sort for scalability

authored and
John Hoffman
committed

Commits on Feb 7, 2026

Fix TLS to use limb-darkened transit template instead of box model

authored and
John Hoffman
committed
Add automated RunPod pod lifecycle scripts

authored and
John Hoffman
committed

Commits on Feb 8, 2026

Fix sparse BLS correctness: CPU j-loop, GPU kernels, shared memory

authored and
John Hoffman
committed
Fix mod1_fast overflow and compile_bls function name filtering

authored and
John Hoffman
committed
Add bounds guard to max reductions and block_size validation

authored and
John Hoffman
committed
Fix stale docstring referencing __float2int_rd

authored and
John Hoffman
committed
Replace circular sparse BLS tests with ground-truth correctness tests

authored and
John Hoffman
committed
Fix sparse BLS test compatibility and robustness

authored and
John Hoffman
committed
Strengthen sparse BLS tests per code review

authored and
John Hoffman
committed
Merge fix/sparse-bls-tests: Fix sparse BLS correctness + ground-truth tests

John Hoffman
authored and
John Hoffman
committed
Merge fix/bls-optimized-minor: Fix mod1_fast overflow + compile_bls filtering

John Hoffman
authored and
John Hoffman
committed
Fix documentation errors, citations, and remove fabricated benchmarks

authored and
John Hoffman
committed
Merge tls-gpu-implementation: Add GPU-accelerated Transit Least Squares

authored and
John Hoffman
committed
Rewrite benchmark framework: all algorithms, cost-per-lightcurve, GPU comparison

authored and
John Hoffman
committed
Add multi-GPU benchmark results across 7 GPU architectures

authored and
John Hoffman
committed
Add multi-lightcurve batch BLS kernel and Keplerian frequency grid

authored and
John Hoffman
committed
Add cuFINUFFT backend for Lomb-Scargle periodogram

authored and
John Hoffman
committed
Merge branch 'feature/bls-batch' into testing/runpod-benchmarks

John Hoffman
authored and
John Hoffman
committed
Fix bugs found during GPU testing, add benchmark suite

authored and
John Hoffman
committed

Commits on Feb 10, 2026

Add benchmark docs, fBLS spec, and fix survey-scale benchmarks

authored and
John Hoffman
committed