diff --git a/examples/cuda-fanout/CATALOG.md b/examples/cuda-fanout/CATALOG.md index 00bef02..6369801 100644 --- a/examples/cuda-fanout/CATALOG.md +++ b/examples/cuda-fanout/CATALOG.md @@ -221,6 +221,47 @@ work where H + S exist; that's a future workload. Progress doc: examples/cuda-fanout/plans/form-D-build-progress.md +## Wave 4 — surveyed 2026-06-05 + +20 additional forms covering lattice cryptography, lattice +reduction (LLL/BKZ/sieve), privacy/MPC primitives, ZK prover +internals beyond MSM/NTT, sparse direct solvers, real-time DSP, +computational biology beyond pairwise alignment, erasure coding, +computational geometry, ODE/PDE, time series, quantum chemistry. + +| form | speedup / throughput | hardware | relevance | +|-----------------------------------|------------------------------------------------|---------------------|-----------| +| cuda-kalman-batched | 1386x for 5000-component measurements | various | unsandbox HIGH (RSSI tracking, punters surface) | +| cuda-g6k-tensor-sieve | 1230x vs G6K CPU at dim 120; SVP record dim 180 | 4 Turing | ECDSA HIGH (lattice attacks on biased nonces) | +| cuda-zkspeed-sumcheck-hyperplonk | 801x geomean; sumcheck 8.4s → 9.5ms | accelerator | undefect HIGH (verifiable-defect-scan provenance) | +| cuda-kyber-batched-ntt | ~451x batched (B=65k); HI-Kyber 6.47x prior SOTA | RTX 3080 | undefect HIGH (PQ migration); unsandbox HIGH (proxy TLS) | +| cuda-ironman-ote | 237x OT throughput vs full-thread CPU | near-memory | unsandbox HIGH (cross-fleet PSI/MPC) | +| cuda-cudss-cholesky | >100x vs QDLDL; 20x vs CHOLMOD factor | NVIDIA | undefect HIGH (defect-graph eigenproblems) | +| cuda-particle-filter | ~150x absolute (5000 particles @ 170 Hz) | GPGPU | unsandbox MEDIUM | +| cuda-rk-stiff-chemkin | 126x single-core; 25x 6-core hydrogen RKCK 524k ODEs | GPGPU | ECDSA MEDIUM (batched ODE maps to candidate sweep) | +| cuda-fem-assembly-jit | 87x assembly serial CPU; 126x peak num integration | GPGPU | LOW | +| cuda-cufalcon-sign | 201k sig/s Falcon-512 A100; verify 2.72M sig/s, 29.5x AVX2 | A100 | undefect HIGH; unsandbox HIGH (PQ disclosure signing) | +| cuda-cudahull-3d | 30-40x over Qhull CPU | NVIDIA | LOW | +| cuda-fastplay-garbled | 35-40x over serial garbling | GPU cluster | unsandbox MEDIUM | +| cuda-air-fri | ~22.8x avg end-to-end ZK speedup | GPGPU | undefect HIGH | +| cuda-rabin-fingerprint | 16x over single-thread CPU; 40 Gbps absolute | GTX 780 | unsandbox HIGH (CDN/proxy cache dedup) | +| cuda-gdel3d | 10x over CGAL 3D Delaunay; 70x Voronoi 10M | NVIDIA | LOW | +| cuda-perasure-crs | 10x vs Jerasure; 10 GB/s GTX780 absolute | GTX 780 | unsandbox HIGH (object-storage erasure) | +| cuda-piranha-mpc | 4x vs CryptGPU VGG16 private inference | single GPU | unsandbox HIGH (MPC layer for untrusted containers) | +| cuda-scamp-matrix-profile | quintillion pairwise comparisons / day | GPGPU | undefect HIGH; unsandbox HIGH (journald intrusion detection) | +| cuda-cudtw-subseq | 2-3 orders of magnitude vs UCR-Suite CPU | Volta | unsandbox HIGH (audio diff for bluetooth/sbc forensics) | +| cuda-terachem-dft | 1-2 orders of magnitude over CPU | 4x Tesla | LOW (catalog completeness) | + +### Wave 4 filtered out (below threshold) + +- MAFFT MSA — 11-20x peak on old NVIDIA hardware; largely surpassed +- RAxML likelihood — 32x kernel only, ~3-10x end-to-end +- AmgX CG — 3-4x on solve-time when AmgX is the baseline +- BVH Karras LBVH — 2-3x over prior GPU LBVH; fails strict 10x +- LDPC decode — 40-160 Mbps; not >=10x over modern SIMD CPU +- Mesh decimation — application-dependent, borderline +- Discrete Gaussian sampler — single-digit %; fold into cuda-kyber-batched-ntt + ## Build order — recommended 3-form runway 1. **A: `cuda-secp256k1-batched-mul`** — biggest immediate win. Drop VanitySearch's CUDA secp256k1 kernel as a BSHK worker; lumbda gets a 6.5 Gkeys/s point-mul oracle. Every candidate validation rides this form. Hardware already owned (3090 + 4090). diff --git a/www/bend.html b/www/bend.html index fc2a9bd..f909363 100644 --- a/www/bend.html +++ b/www/bend.html @@ -217,6 +217,34 @@ make gpu-worker LUMBDA=asm # smallest footprint +

Surveyed forms (Wave 4)

+ + + + + + + + + + + + + + + + + + + + + + + + +
formthroughputhardwareref
cuda-kalman-batched1386× for 5000-component measurementsvariousCUDAkalmanFilter
cuda-g6k-tensor-sieve1230× vs G6K CPU sieve at dim 120; SVP record dim 180 on 4 Turing GPUs4 TuringDucas/Stevens/van Woerden EC 2021
cuda-zkspeed-sumcheck-hyperplonk801× geomean over CPU; sumcheck 8.4 s → 9.5 msfull-chip acceleratorzkSpeed HPCA 2025
cuda-kyber-batched-ntt~451× batched (B=65k); HI-Kyber 6.47× over prior GPU SOTARTX 3080HI-Kyber
cuda-ironman-ote237× OT throughput vs full-thread CPUnear-memory variantIronman arXiv 2507.16391
cuda-cudss-cholesky>100× vs QDLDL; 20× vs CHOLMOD factorNVIDIANVIDIA cuDSS
cuda-particle-filter~150× absolute (5000 particles @ 170 Hz)GPGPUEURASIP J ASP 2013
cuda-rk-stiff-chemkin126× vs single-core; 25× vs 6-core for hydrogen RKCK 524k ODEsGPGPUNiemeyer & Sung
cuda-fem-assembly-jit87× assembly vs serial CPU; 126× peak numerical integrationGPGPUMironov et al.
cuda-cufalcon-sign201k sig/s Falcon-512 on A100; verify 2.72M sig/s, 29.5× vs AVX2A100cuFalcon eprint 2025/249
cuda-cudahull-3d30–40× over Qhull CPUNVIDIACudaHull CAG 2012
cuda-fastplay-garbled35–40× over serial garbling on GPU clusterGPU clusterFastplay eprint 2011/097
cuda-air-fri~22.8× avg end-to-end ZK speedup; FRI commitmentGPGPUAir-FRI SAC 2025
cuda-rabin-fingerprint16× over single-thread CPU; 40 Gbps absoluteGTX 780 (HARENS)HARENS CloudCom 2016
cuda-gdel3d10× over CGAL 3D Delaunay; 70× Voronoi/jump-flood at 10M pointsNVIDIAgDel3D I3D 2014
cuda-perasure-crs10× vs multithread Jerasure; 10 GB/s on GTX780 absoluteGTX 780PErasure IEEE Cluster 2015
cuda-piranha-mpc4× vs CryptGPU on VGG16 private inference; full 3/4-party stacks single-GPUsingle GPUPiranha USENIX Sec 2022
cuda-scamp-matrix-profilequintillion pairwise comparisons / day (absolute)GPGPUSCAMP
cuda-cudtw-subseq2–3 orders of magnitude over UCR-Suite CPU; soft-DTW up to 5000×VoltacuDTW++ Euro-Par 2020
cuda-terachem-dft1–2 orders of magnitude over CPU; 8–50× vs GAMESS on 256-core cluster4× TeslaTeraChem
+

Wave 4 filter-outs: MAFFT MSA (11–20×, surpassed), RAxML likelihood (32× kernel only, ~3–10× end-to-end), AmgX CG (3–4× vs AmgX baseline), BVH Karras LBVH (2–3× over prior GPU LBVH), LDPC decode (40–160 Mbps, not ≥10× over modern SIMD CPU), mesh decimation (application-dependent), discrete Gaussian sampler (single-digit % gains; fold into Kyber NTT). Revisit when the published number changes.

+

Why a form earns its slot

A form is GPU-worth-it when at least one of: