Frozen targets scored
8 / 8
Every Phase 1A target produced a prediction and score packet.
Prospective targets
4 / 4
The after-freeze targets completed without OOM or route timeout.
Median wall clock
396 s
All-target median from generated summary cards.
Median peak RSS
34.08 GB
Approximate process RSS, not full unified-memory profiling.
Result Summary
| Stratum | Targets | Scored | Median wall clock | Median peak RSS | Median lDDT | Median TM-score |
|---|---|---|---|---|---|---|
| All Phase 1A | 8 | 8 | 395.96 s | 34.08 GB | 0.8597 | 0.9218 |
| Pre-freeze evidence | 4 | 4 | 384.60 s | 35.27 GB | 0.8597 | 0.9218 |
| Prospective after freeze | 4 | 4 | 726.63 s | 34.08 GB | 0.8618 | 0.9095 |
Target-Level Results
| Target | Stratum | Bin | Length | pLDDT | Wall clock | Peak RSS | lDDT | TM-score | RMSD | Aligned residues | Card |
|---|---|---|---|---|---|---|---|---|---|---|---|
13CZ:A | pre-freeze | mid | 161 | 84.1826 | 378.916 s | 31.9900 GB | 0.8613 | 0.9224 | 1.65 | 150 | summary |
13EE:A | pre-freeze | mid | 161 | 81.6716 | 401.635 s | 38.5429 GB | 0.8582 | 0.9212 | 1.68 | 148 | card |
13FT:A | pre-freeze | mid | 228 | 92.9916 | 390.276 s | 31.2522 GB | 0.8915 | 0.9588 | 1.23 | 212 | card |
11BC:C | pre-freeze | hard | 116 | 81.2384 | 337.455 s | 41.4060 GB | 0.7492 | 0.6569 | 1.97 | 63 | card |
12OY:A | prospective | hard | 304 | 87.7506 | 308.066 s | 33.2323 GB | 0.8135 | 0.8724 | 2.84 | 285 | card |
9H8N:A | prospective | long | 604 | 88.1646 | 739.974 s | 34.6408 GB | 0.8420 | 0.8491 | 3.68 | 568 | card |
9K9X:A | prospective | long | 565 | 95.8087 | 713.290 s | 34.0239 GB | 0.8815 | 0.9467 | 1.96 | 522 | card |
9K9Y:A | prospective | long | 565 | 96.2908 | 753.399 s | 34.1274 GB | 0.9157 | 0.9826 | 1.07 | 487 | card |
Figures
Interpretation
SimpleFold-3B is operationally viable on this M3 Max setup for the frozen Phase 1A set. The long-bin targets completed without OOM, clustering near 34-35 GB peak RSS. This supports a narrow Apple Silicon viability claim for the stated model, route, hardware, and targets.
This is not a SimpleFold paper reproduction, a biological generality claim, or a cross-model leaderboard. The membrane/complex targets need domain review, and the 11BC:C score must be interpreted with its 63-residue alignment coverage caveat.
Artifacts
- Phase 1A summary CSV
- Phase 1A summary JSON
- SimpleFold-3B install and runtime card
- Runtime environment JSON
- Phase 1A biological review packet
- Biological review CSV
- Biological review JSON
- 12OY:A result directory
- 9H8N:A result directory
- 9K9X:A result directory
- 9K9Y:A result directory
- 9K9X terminal log
- 9K9Y terminal log
- Results page screenshot
- Public results page screenshot
- Protocol page screenshot
- Model/runtime card screenshot
- Workshop draft screenshot
Source documents: docs/34_simplefold3b_phase1a_results.md and pilot/data/simplefold3b_phase1a_summary.json.