All issues

docs(readme): refresh the benchmark table — accumulated drift (stale numbers + descriptions) found during #440

#442

· pekkah opened

on Jul 13, 2026

perf(cuda): DSpark GPU draft round runs ~3× the bandwidth floor — small-M (B=7) projection GEMMs → weight-stationary matvec (#428 lever 1)

#441

· pekkah opened

on Jul 13, 2026

Centralize the KVarN support matrix + extend it to the server (TurboQuant head-dim gate)

#437

· pekkah opened

on Jul 11, 2026

GPU Lloyd-Max TqAttention writes compressed-region V aggregate in the rotated basis (no inverse WHT)

#435

· pekkah opened

on Jul 11, 2026

KVarN follow-ups: int8-TC prefill GEMM, split-tile decode kernel, minors (#180 successor)

#433

· pekkah opened

on Jul 11, 2026

perf(cuda): CUDA-graph-captured k-token spec verify — the ~10.6 ms fixed verify overhead is now DSpark's whole gap (#428 lever 2)

#430

· pekkah opened

on Jul 9, 2026

feat(engine): DSpark under continuous batching + load-aware verify length (spec Phases 5/6 remainder)

#427

· pekkah opened

on Jul 9, 2026

perf(cuda): dense int8-MMQ prefill ~2–2.6× behind llama.cpp (cp.async MMQ + flash attn) — Qwen3-8B / Gemma4

#409

· pekkah opened

on Jun 25, 2026

perf(cuda): full-offload MoE decode ~3.4× behind llama.cpp — OLMoE 126 vs 426 t/s (on-GPU experts, kernel-bound)

#408

· pekkah opened

on Jun 25, 2026

perf(cuda): batched int8-MMQ trunk prefill gated off for dense NORM-RoPE & QKV-bias models — SmolLM2 105×, VibeThinker/Qwen2 159×

#407

· pekkah opened

on Jun 25, 2026

perf(cuda): batched int8-MMQ trunk prefill gated off for MoE — OLMoE prefill 97 vs llama.cpp 17,137 t/s (176×)

#406

· pekkah opened

on Jun 25, 2026

perf(cuda): README vs llama.cpp CUDA benchmark gaps — full results + tracking

#405

· pekkah opened

on Jun 25, 2026