strata generate: PCIe probe: 28.7 GB/s host->device (best of 28.2 28.6 28.7 28.7) -> 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-6f32ec07/iq3_s experts (largest blob 2.66 MB), token embedding IQ4_XS in mapped host memory (322 MiB)
strata generate: 1466 MiB of weights loaded from /data/strata-bench/packs-6f32ec07/iq3_s (302 canonical tensors skipped: served natively)
strata generate: 300 native projection matrices, 2018.88 MiB of weights
strata generate: the SSD is kept awake while rows are read: one page of the table after 100 ms without a read, until 60 s after the last request (STRATA_SSD_KEEPALIVE=0 turns it off)
strata generate: profile /data/strata-bench/src/strata-6f32ec07/data/expert-profile.bin: 24576 ranked pairs, built for 24576 slots
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 0.64 s (1226 MiB/s)
strata generate: experimental native Q5_K head, 521472000 bytes
strata generate: GPU 0: NVIDIA GeForce RTX 5090 Laptop GPU, compute capability 12.0
strata generate: expert arena read through the file cache (buffered (not Windows))
strata generate: expert arena: cudaHostRegister PORTABLE ok; large pages skipped (STRATA_NO_LARGEPAGES); using 4 KB pages
strata generate: loaded 46.84 GiB at 5.53 GiB/s
strata generate: expert cache auto: 16.08 GiB free, 700 MiB reserved (+86 MiB for the draft head) -> 6174 slots
strata generate: expert cache 8073 slots, 15.31 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 8073 of 8073 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 0.1.39)
strata generate: token graph hit path: 8073 resident experts, decided on the device
strata serve: prompt chunk auto: 8192 tokens, a 384-slot ring
strata serve: the prompt path borrows 2183 CUDA0 cache slots (4.12 GiB)
strata hc: CUDA0: the hyper-connection read runs as staged (the norm per token and stream, the down projection's activations staged ahead by cp.async); checked bit for bit against the plain read on this card (STRATA_HC_SPLIT=1 or 0 for the earlier ones)
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: 323 MiB of VRAM free with everything loaded
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 1996 ms (2023.0 tok/s), 256 generated in 3532 ms (72.5 tok/s), drafts accepted 160 of 237, 1 checkpoints
strata serve: decode expert cache hit rate: 87.3% (127882 hits / 146540 lookups); 13300 more read by the GPU over PCIe or from another GPU (8.3% of all 159840 routed)
strata serve: suffix drafts: 4 windows, 9 of 20 drafts accepted
strata serve: prompt 4038 tokens = 0 reused + 4038 read in 1763 ms (2290.2 tok/s), 256 generated in 3138 ms (81.6 tok/s), drafts accepted 160 of 210, 1 checkpoints
strata serve: decode expert cache hit rate: 92.3% (128923 hits / 139655 lookups); 7225 more read by the GPU over PCIe or from another GPU (4.9% of all 146880 routed)
strata serve: suffix drafts: 1 windows, 2 of 3 drafts accepted
strata serve: prompt 4038 tokens = 0 reused + 4038 read in 1794 ms (2250.8 tok/s), 256 generated in 3164 ms (80.9 tok/s), drafts accepted 161 of 226, 1 checkpoints
strata serve: decode expert cache hit rate: 92.3% (135658 hits / 147005 lookups); 7555 more read by the GPU over PCIe or from another GPU (4.9% of all 154560 routed)
strata serve: suffix drafts: 1 windows, 4 of 5 drafts accepted