strata serve: prompt chunk auto: 8192 tokens, a 384-slot ring
strata prefill timing: 799 tokens, GPU timeline 3329 ms, wall 3339 ms, host staging 55 ms: embed+steps 5 (0.2%) hc read 270 (8.1%) gdn 443 (13.3%) qsa proj 91 (2.7%) qsa indexer 0 (0.0%) qsa select 1 (0.0%) qsa attn 50 (1.5%) router+shared 44 (1.3%) host grouping 12 (0.3%) gather 11 (0.3%) wait copy 1376 (41.3%) dequant 415 (12.5%) gemm gate/up 311 (9.3%) gemm down 155 (4.6%) combine 18 (0.5%) ple 10 (0.3%) kv stage 0 (0.0%) gdn conv+gates 5 (0.1%) gdn recurrence 35 (1.0%) gdn out proj 78 (2.3%)
strata serve: prompt 806 tokens = 0 reused + 806 read in 3635 ms (221.7 tok/s), 2 generated in 140 ms (14.3 tok/s), drafts accepted 3 of 3, 1 checkpoints
strata prefill timing: 495 tokens, GPU timeline 2477 ms, wall 2479 ms, host staging 107 ms: embed+steps 0 (0.0%) hc read 81 (3.3%) gdn 365 (14.7%) qsa proj 83 (3.4%) qsa indexer 0 (0.0%) qsa select 0 (0.0%) qsa attn 24 (1.0%) router+shared 77 (3.1%) host grouping 15 (0.6%) gather 3 (0.1%) wait copy 1077 (43.5%) dequant 319 (12.9%) gemm gate/up 223 (9.0%) gemm down 109 (4.4%) combine 10 (0.4%) ple 6 (0.2%) kv stage 0 (0.0%) gdn conv+gates 3 (0.1%) gdn recurrence 20 (0.8%) gdn out proj 61 (2.5%)
strata serve: prompt 502 tokens = 0 reused + 502 read in 2667 ms (188.2 tok/s), 2 generated in 114 ms (17.5 tok/s), drafts accepted 3 of 3, 1 checkpoints
strata prefill timing: 1108 tokens, GPU timeline 3391 ms, wall 3396 ms, host staging 184 ms: embed+steps 3 (0.1%) hc read 158 (4.7%) gdn 598 (17.6%) qsa proj 117 (3.5%) qsa indexer 0 (0.0%) qsa select 2 (0.0%) qsa attn 88 (2.6%) router+shared 53 (1.6%) host grouping 14 (0.4%) gather 7 (0.2%) wait copy 1144 (33.7%) dequant 454 (13.4%) gemm gate/up 358 (10.6%) gemm down 179 (5.3%) combine 26 (0.8%) ple 14 (0.4%) kv stage 0 (0.0%) gdn conv+gates 12 (0.3%) gdn recurrence 55 (1.6%) gdn out proj 109 (3.2%)
strata serve: prompt 1115 tokens = 0 reused + 1115 read in 3676 ms (303.3 tok/s), 2 generated in 115 ms (17.4 tok/s), drafts accepted 3 of 3, 1 checkpoints
strata prefill timing: 8874 tokens, GPU timeline 11978 ms, wall 11981 ms, host staging 336 ms: embed+steps 38 (0.3%) hc read 1200 (10.0%) gdn 1346 (11.2%) qsa proj 597 (5.0%) qsa indexer 1 (0.0%) qsa select 62 (0.5%) qsa attn 1897 (15.8%) router+shared 228 (1.9%) host grouping 42 (0.3%) gather 56 (0.5%) wait copy 1683 (14.1%) dequant 854 (7.1%) gemm gate/up 1428 (11.9%) gemm down 706 (5.9%) combine 768 (6.4%) ple 34 (0.3%) kv stage 8 (0.1%) gdn conv+gates 92 (0.8%) gdn recurrence 491 (4.1%) gdn out proj 447 (3.7%)
strata serve: prompt 8881 tokens = 0 reused + 8881 read in 12470 ms (712.2 tok/s), 3 generated in 117 ms (25.7 tok/s), drafts accepted 3 of 3, 1 checkpoints
