strata generate: PCIe probe: 28.7 GB/s host->device -> pcie_frac 0.55 (default 0.55)
strata generate: this CPU has no AVX-512: the expert kernels run on AVX-2 (multi-token for the i-quant gate/up rows)
strata generate: native pack: /data/strata-bench/packs/iq3_s experts (largest blob 2.66 MB), token embedding type 23 in mapped host memory (322 MiB)
strata generate: 1466 MiB of weights loaded from /data/strata-bench/packs/iq3_s (302 canonical tensors skipped: served natively)
strata generate: 300 native projection matrices, 2018.88 MiB of weights
strata generate: PLE on, table 320001536 rows of /data/strata-bench/models/ISTA-ed59f920/IQ3_S/Qwen3.8-Flash-Next-GSQ-RCO-IQ3_S-00002-of-00002.gguf
strata mtp: draft layer loaded, 950 MiB of VRAM (experts 675, dense 111), files read in 0.73 s (1084 MiB/s)
strata generate: expert arena: cudaHostRegister PORTABLE ok; MAP_HUGETLB unavailable (no hugetlb pool configured?); using 4 KB pages
strata generate: loaded 46.84 GiB at 5.71 GiB/s
strata generate: profile /data/strata-bench/src/strata-8fc40dde/data/expert-profile.bin: 24576 ranked pairs, built for 24576 slots
strata generate: experimental native Q5_K head, 521472000 bytes
strata generate: expert cache auto: 16.08 GiB free, 700 MiB reserved -> 6210 slots
strata generate: expert cache 8120 slots, 15.40 GiB of VRAM; policy is
strata generate: the GPU computes the experts in the cache; it rounds differently from the CPU,
                 so a reply can differ slightly from a run without the cache (same quality:
                 bench/results/2026-09-27-cache-parity).
                 PROFILE, ranked by routing frequency, no eviction.
strata generate: pre-filled 8120 of 8120 slots from the profile; slot 0 verified
strata generate: R4 hit path ON - resident experts are computed on the GPU
strata generate: 23 expert-pool workers + the host thread
strata generate: session is up (engine 0.1.25)
strata generate: token graph hit path: 8120 resident experts, decided on the device
strata serve: prompt chunk auto: 8192 tokens
strata serve: the prompt path borrows 2349 cache slots (4.43 GiB)
strata verify: window up to 6 tokens, 74.7 MiB of device buffers
strata mtp: draft head over 40525 tokens (81.2 MiB)
strata serve: 233 MiB of VRAM free with everything loaded - LOW: requests may stall; add --vram-reserve-mib 979 to the config's args (or lower --max-context)
strata verify: captured the 6-token window (upload no error, sync no error)
strata verify: captured the 1-token window (upload no error, sync no error)
strata verify: captured the 4-token window (upload no error, sync no error)
strata verify: captured the 3-token window (upload no error, sync no error)
strata verify: captured the 2-token window (upload no error, sync no error)
strata serve: prompt 4038 tokens = 0 reused + 4038 read in 2157 ms (1871.9 tok/s), 256 generated in 3289 ms (77.8 tok/s), drafts accepted 156 of 236, 1 checkpoints
strata serve: decode expert cache hit rate: 87.1% (128839 hits / 147989 lookups)
strata serve: suffix drafts: 5 windows, 14 of 25 drafts accepted
strata serve: prompt 4038 tokens = 0 reused + 4038 read in 1872 ms (2157.3 tok/s), 256 generated in 2805 ms (91.3 tok/s), drafts accepted 162 of 234, 1 checkpoints
strata serve: decode expert cache hit rate: 91.7% (137700 hits / 150161 lookups)
strata serve: suffix drafts: 1 windows, 1 of 2 drafts accepted
strata serve: prompt 4038 tokens = 0 reused + 4038 read in 1880 ms (2147.5 tok/s), 256 generated in 2996 ms (85.4 tok/s), drafts accepted 161 of 230, 1 checkpoints
strata serve: decode expert cache hit rate: 92.5% (138741 hits / 149963 lookups)
strata serve: prompt 32038 tokens = 0 reused + 32038 read in 12087 ms (2650.5 tok/s), 256 generated in 2938 ms (87.1 tok/s), drafts accepted 150 of 214, 2 checkpoints
strata serve: decode expert cache hit rate: 91.9% (134042 hits / 145842 lookups)
strata serve: suffix drafts: 1 windows, 2 of 2 drafts accepted
strata serve: prompt 32038 tokens = 0 reused + 32038 read in 12067 ms (2655.1 tok/s), 256 generated in 2996 ms (85.5 tok/s), drafts accepted 156 of 215, 2 checkpoints
strata serve: decode expert cache hit rate: 93.2% (135449 hits / 145279 lookups)
strata serve: prompt 32038 tokens = 0 reused + 32038 read in 12024 ms (2664.5 tok/s), 256 generated in 2946 ms (86.9 tok/s), drafts accepted 158 of 219, 2 checkpoints
strata serve: decode expert cache hit rate: 93.5% (136960 hits / 146485 lookups)
strata serve: suffix drafts: 3 windows, 5 of 6 drafts accepted
strata serve: prompt 128037 tokens = 0 reused + 128037 read in 47142 ms (2716.0 tok/s), 256 generated in 3564 ms (71.8 tok/s), drafts accepted 155 of 219, 6 checkpoints
strata serve: decode expert cache hit rate: 91.7% (133204 hits / 145239 lookups)
strata serve: suffix drafts: 1 windows, 1 of 1 drafts accepted
strata serve: prompt 128037 tokens = 0 reused + 128037 read in 47334 ms (2705.0 tok/s), 256 generated in 3675 ms (69.7 tok/s), drafts accepted 142 of 233, 6 checkpoints
strata serve: decode expert cache hit rate: 92.4% (146228 hits / 158295 lookups)
strata serve: suffix drafts: 7 windows, 12 of 19 drafts accepted
strata serve: prompt 128037 tokens = 0 reused + 128037 read in 47465 ms (2697.5 tok/s), 256 generated in 3958 ms (64.7 tok/s), drafts accepted 142 of 253, 6 checkpoints
strata serve: decode expert cache hit rate: 92.3% (154468 hits / 167425 lookups)
strata serve: suffix drafts: 12 windows, 24 of 38 drafts accepted