Operational bounded quant pipeline; CPU reference measured, GPU comparison requires a kernel receipt
Three orthogonal risk signals per bar — PCA-Risk · TDA-Fracture · HJB-Kelly — with receipt state SIGNED_VERIFIED, honestly labeled SAMPLE_SIGNAL · NOT_LIVE · NO_BACKTEST_VALIDATED. Not a trading instruction.
backend=CPU pure-Python reference · gpu_reachable=False · scenario=calm
CPU pure-Python referenceLedoit-Wolf shrinkage intensity ρ and the MP noise edge λ⁺ are computed exactly per the cited formulas on a SAMPLE synthetic universe. cuML/cuPy GPU acceleration is ROADMAP (CPU fallback ran here). NO backtest.
β0 (components) is exact via union-find; β1 is the 1-skeleton cycle rank (E−V+C), an honest fast proxy for the genuine Vietoris-Rips H1 that the GPU giotto-tda/Ripser++ path (ROADMAP) computes exactly. Synthetic SAMPLE windows; z-score baseline is illustrative. NOT calibrated on real data.
Weights auto-compress when fracture/anomaly fire (derisk_ratio<1) — the elegant property of the TDA-Kelly channel. BUT γ and κ are free, UNcalibrated hyperparameters: this is MODELED architecture, NOT a backtested strategy. NEVER a live-trading instruction.
szl-gpu-quant-v0.1253df7441469f15d…DSSE signature verified against the configured SZL public key.
vLLM --tensor-parallel-size 2 shards ONE larger model across a-11-oy.com GPU + RTX 4000
e.g. Qwen3-32B comfortably, or a quantized Nemotron-3-Super across combined VRAM
main GPU = primary agent model; RTX 4000 = dedicated governance/draft GPU: Auto-Review CLASSIFIER + speculative-decode DRAFT (Qwen2.5-Coder-1.5B) + embeddings
keeps the main GPU from stalling on inline review/draft — best fit for our agent+Auto-Review arch
Route via build.nvidia.com NIM through our LiteLLM/RouteLLM gateway
Ultra (550B-A55B) needs ~768GB VRAM (4×GB200-class) — CANNOT run on 2 local GPUs. NEVER claim local Ultra. Verify NVIDIA claims on OUR τ-bench+J/token harness.
| Claim | NVIDIA published | SZL comparison | Separate local evidence |
|---|---|---|---|
| Nemotron speedup vs prior frontier | up to 5x | — HISTORICAL | 0.5217x local Ollama wall-speed ratio vs qwen2.5:3b |
| Reasoning/accuracy uplift | +30% | 16.667 percentage points vs local baseline HISTORICAL | — |
| Benchmark accuracy | 91% | 100.0% (6/6 exact match) HISTORICAL | — |
| Long-context retrieval | 1M-token retrieval | 1/2 needles; max 2050 evaluated prompt tokens HISTORICAL | — |
| cuML PCA speedup (quant Layer 1) | 10-50x (S&P 500 scale); about 100x genomic | — HISTORICAL | CPU reference p50 11.09 ms; GPU comparison UNAVAILABLE |
| Ripser++ persistence (quant Layer 2) | up to 30x vs CPU Ripser | — HISTORICAL | CPU stress reference p50 10.378 ms; GPU comparison UNAVAILABLE |
loaded=True · completed_at=2026-07-18T17:44:02.420257+00:00 · signature=UNSIGNED_CONTENT_ADDRESSED · content_sha256=8ede48b23df3063247c0…
The NVIDIA column is a cited vendor claim, not an endorsement. The SZL column comes from a bounded local execution receipt and is not presented as a vendor-scale replication. Signature state is derived from cryptographic verification; unsigned content-addressed evidence is never displayed as signed.