Origin. PLI Labs research archive, file
kernel-research/UPSCAN1105-SUPPLEMENT.md. Published 2026-11-18 in the Volta V100 research repository. Edited for publication. Machine names, private file paths, network addresses, session names and the operator’s name were replaced with generic terms (the V100 server,[private path],[address],the owner). Links to files that are not published were turned into plain text. No figure and no finding was changed. Dates inside the note are its own; later results are in the build pages of this repository.
This is a metadata watch alongside the three sources required by BRIEF.md, not a replacement for their full inventories. Public GitHub reads only, 2026-11-05. Raw evidence is in evidence/.
- 0rand: public repository list captured; all 15 branches of
deepseek-v41-tensorfold-sparkand all 15 branches ofDeepSeek-V4.1-Flash-1017-EXL3-TensorFold-2x-DGX-Sparkexamined for commits since 2026-10-28. Only the former’smain-0randhas new commits in this date window (2), heada903cb5fe3b4c91b2b17e2a1aaf5dc80a045ced0; this is the head already reviewed in an earlier private review (2026-11-02). Recipe default ispr65-flash17-servingat305659970ae63729b0753b27dea85a7ac4b5a93b. No unreviewed branch update after the last reviewed head was found. jayleaton/deepseek-v41-tensorfold-spark: 10 main commits since date; source of earlier reviewed batch/row-invariance design, not new Volta kernels. Saved metadata.ashhart/TensorFold: 1,450 main commits returned across all pages since date, latestcde4fe7d9959; saved complete metadata. Title inventory only here, not a full 1,450-change implementation review.turboderp-org/exllamav3: 20 master commits since date, saved metadata.HeNryous/deepseek-v41-1017-dgx-spark: zero default-branch commits since date. Initial guessed repoHeNryous/DeepSeek-V4.1-Flashreturned 404; resolved actual name from the owner’s upstream-watch script and checked successfully. The 404 is not an access blocker.vllm-project/vllm: 726 main commits since date, saved complete metadata and inspected selected primary PR descriptions below. Hardware-specific numbers remain author claims on their stated equipment.
| Primary change | Potential use / KCAMP mapping | Scope and exclusion |
|---|---|---|
| #61464 MRV2 DBO with speculative decoding | C3: request-boundary microbatch split, full-batch metadata rebuild after target forward so speculator FULL graphs see valid buffers | Engine design reference, not proof of PP2/DSpark/SM70 support or measured benefit here. DBO communication overlap is not automatically PP-stage interleaving. |
| #64458 MRV2 PP broadcast fixes | C0/C2/C3/S6: avoid non-last-stage host sync; gather on main stream before broadcast stream reads; skip absent drafts; size broadcast per actual step | Correctness/ownership audit before greater concurrency; compare with our already ported PP hooks. No extra speed credit for duplicate code. |
| #60618 fused prefill score + top-k | New/S10-related prefill selection design: remove logits materialization, one persistent launch for <=16,384 query rows / <=1M keys; 296 MiB workspace stated vs 2.6 GiB LiteTopK | Their gated 24-bit ranking permits boundary ties and has a rescanning path; our exactness must be preserved. Modern FP8 scoring implementation needs a Volta FP16/FP32 replacement. No portable whole-model multiplier. |
| #59131 gather compressed KV once | S1/S2/S4 context/memory traffic: 38 compressed-layer gathers become 4 source-group gathers per chunk on the described path | Mega-attention B200 path; our packed ring and fused sparse attention do not necessarily gather whole history. Audit actual calls first; no forecast without traffic count. |
| #63956 token-sharded sparse MLA prefill | New/S1/S4: sequence-parallel prefill, preserve head-sharded decode in mixed steps | Author claims GLM-5.3 NVFP4 TP4 GB200 TTFT 1.24x/1.29x/1.40x at 32K/64K/128K. Those are neither V4.1 nor Volta data. Would change communication and GEMM shapes; no exactness assumption. |
| #64442 DSpark inverse-CDF Markov fusion | C1/S6 implementation pattern for fused sampler; author GB200 one-request Markov 111.9→71.6 µs, not whole round | Our coupled keyed-Gumbel sampler is already fused/sharded. Switching random transforms would lose seeded byte identity/coupling unless redesigned, so do not adopt as a drop-in speed win. |
All six selected PR descriptions and merge SHAs are captured as evidence/vllm-pr-*.json. None of these supplementary reads authorize deployment. The ranked report is grounded in the requested 1Cat/Maya/DSX inventories and our measured baseline.