strata serve: prompt 19 tokens = 0 reused + 19 read in 437 ms (43.5 tok/s), 2 generated in 110 ms (18.3 tok/s), drafts accepted 3 of 3, 1 checkpoints
strata serve: decode expert cache hit rate: 48.4% (1082 hits / 2236 lookups)
strata serve: KV streaming: 92.31% of 468 block reads hit VRAM, 0.1 MiB read from RAM
strata serve: prompt 4096 tokens = 0 reused + 4096 read in 5216 ms (785.3 tok/s), 256 generated in 3839 ms (66.7 tok/s), drafts accepted 142 of 168, 1 checkpoints
strata serve: decode expert cache hit rate: 87.4% (118856 hits / 135923 lookups)
strata serve: KV streaming: 99.28% of 896574 block reads hit VRAM, 26.2 MiB read from RAM
strata serve: prompt 4096 tokens = 0 reused + 4096 read in 5067 ms (808.3 tok/s), 256 generated in 3285 ms (77.9 tok/s), drafts accepted 138 of 165, 1 checkpoints
strata serve: decode expert cache hit rate: 96.8% (131390 hits / 135796 lookups)
strata serve: KV streaming: 99.27% of 890388 block reads hit VRAM, 26.1 MiB read from RAM
strata serve: prompt 4096 tokens = 0 reused + 4096 read in 5059 ms (809.6 tok/s), 256 generated in 3303 ms (77.5 tok/s), drafts accepted 140 of 177, 1 checkpoints
strata serve: decode expert cache hit rate: 97.4% (137888 hits / 141543 lookups)
strata serve: KV streaming: 99.30% of 927396 block reads hit VRAM, 26.2 MiB read from RAM
strata serve: prompt 32768 tokens = 0 reused + 32768 read in 16972 ms (1930.7 tok/s), 256 generated in 3521 ms (72.7 tok/s), drafts accepted 135 of 164, 4 checkpoints
strata serve: decode expert cache hit rate: 98.0% (134487 hits / 137264 lookups)
strata serve: KV streaming: 97.10% of 899658 block reads hit VRAM, 105.3 MiB read from RAM
strata serve: prompt 32768 tokens = 0 reused + 32768 read in 16993 ms (1928.3 tok/s), 256 generated in 3611 ms (70.9 tok/s), drafts accepted 128 of 163, 4 checkpoints
strata serve: decode expert cache hit rate: 98.5% (137601 hits / 139675 lookups)
strata serve: KV streaming: 97.10% of 915036 block reads hit VRAM, 106.9 MiB read from RAM
strata serve: prompt 32768 tokens = 0 reused + 32768 read in 16966 ms (1931.3 tok/s), 256 generated in 3483 ms (73.5 tok/s), drafts accepted 139 of 175, 4 checkpoints
strata serve: decode expert cache hit rate: 98.4% (138858 hits / 141109 lookups)
strata serve: KV streaming: 97.16% of 924324 block reads hit VRAM, 105.7 MiB read from RAM
strata serve: prompt 128000 tokens = 0 reused + 128000 read in 58205 ms (2199.1 tok/s), 256 generated in 3538 ms (72.4 tok/s), drafts accepted 142 of 162, 6 checkpoints
strata serve: decode expert cache hit rate: 98.5% (130884 hits / 132932 lookups)
strata serve: KV streaming: 94.90% of 871932 block reads hit VRAM, 179.0 MiB read from RAM
strata serve: prompt 128000 tokens = 0 reused + 128000 read in 58181 ms (2200.0 tok/s), 256 generated in 3551 ms (72.1 tok/s), drafts accepted 144 of 181, 6 checkpoints
strata serve: decode expert cache hit rate: 98.7% (138787 hits / 140621 lookups)
strata serve: KV streaming: 95.26% of 921222 block reads hit VRAM, 175.8 MiB read from RAM
strata serve: prompt 128000 tokens = 0 reused + 128000 read in 58268 ms (2196.7 tok/s), 256 generated in 3376 ms (75.8 tok/s), drafts accepted 147 of 166, 6 checkpoints
strata serve: decode expert cache hit rate: 99.0% (131123 hits / 132474 lookups)
strata serve: KV streaming: 94.78% of 868860 block reads hit VRAM, 182.8 MiB read from RAM
