RTX 3060
Small-card capacity and 32k RULER parity.
GPU-grouped public template for Hydra kernel packaging, benchmark, capacity, quality, and claim-boundary evidence.
Public framing: Hydra is a bounded-residency decode attention kernel. These rows support package coverage, selected capacity unlocks, tested quality parity, and exact-Qwen fit/usability evidence. They do not support universal speedup or universal FlashAttention replacement claims.
Small-card capacity and 32k RULER parity.
Kernel package and HF benchmark coverage.
Package coverage plus negative revolver boundary.
Capacity row and exact-Qwen FP8 evidence through 14k.
4070 capacity/quality rows; 4070 Ti package rows.
Qwen3-4B 32k/65k RULER quality parity rows.
80GB capacity row through bounded T=327,680.
HF benchmark and final kernel-builder gate.
Blackwell package/HF rows and exact-Qwen frontier data.