Raise ubatch benchmark context to 32K

docs/LLM/ubatch-benchmark-large.md

@@ -13,7 +13,7 @@ timestamp: 2026-09-01
 # frank-default-large ubatch sweep
 
 This benchmark sweeps the active `frank-default-large` model on Dankclaw using
-scratch ports and a 16K-context workload. The production service is stopped
+scratch ports and a 32K-context workload. The production service is stopped
 before the sweep and restarted by the benchmark's exit handler.
 
 ## Sweep
@@ -30,7 +30,7 @@ STOPSVC=1 ./scripts/bench-ubatch-rocmi4.sh
 ```
 
 The script uses the active Q6 AGENT model, standalone MTP draft, a fixed
-16K batch, and the tracked Sharp
+32K batch, and the tracked Sharp
 template, and `--reasoning-format deepseek`. It reports server timing lines and
 per-run generation throughput. Run only when the production service can be
 interrupted; the GPU is shared with live workloads.

scripts/bench-ubatch-rocmi4.sh

@@ -3,7 +3,7 @@
 # frank-default-large ROCmFPX standalone-MTP stack (dankclaw, gfx1151).
 #
 # Boots one llama-server instance per ubatch on a scratch port with a fixed
-# -b and -c 16384, runs a standardized ~75%-of-ctx prose workload, and reads
+# -b and -c 32768, runs a standardized ~75%-of-ctx prose workload, and reads
 # authoritative prefill/generation tok/s from the server's own print_timing
 # lines. Steps from UBATCH_MIN to UBATCH_MAX inclusive by UBATCH_STEP.
 #
@@ -30,8 +30,8 @@ fi
 : "${UBATCH_MAX:=6144}"
 : "${UBATCH_STEP:=256}"
 : "${UBATCH_VALUES:=256 512 768 1024 1280 1536 2048 3072 4096 5120 6144}"
-: "${CTX:=16384}"                # 16k contexts
-: "${BENCH_BATCH:=16384}"         # batch is fixed to the 16k context size
+: "${CTX:=32768}"                # 32k contexts
+: "${BENCH_BATCH:=32768}"         # batch is fixed to the 32k context size
 : "${OUT_TOKENS:=256}"
 : "${RUNS:=3}"                   # decode runs per ubatch
 : "${PORT_BASE:=7894}"           # scratch ports 7894..

tests/frank_large_config.bats

@@ -69,8 +69,8 @@ TEMPLATE_FILE="$REPO_ROOT/scripts/chat-template.jinja"
     grep -q ': "${UBATCH_MAX:=6144}"' "$bench"
     grep -q 'UBATCH_VALUES.*256 512 768 1024 1280 1536 2048 3072 4096 5120 6144' "$bench"
     grep -q ': "${STOPSVC:=1}"' "$bench"
-    grep -q ': "${CTX:=16384}"' "$bench"
-    grep -q ': "${BENCH_BATCH:=16384}"' "$bench"
+    grep -q ': "${CTX:=32768}"' "$bench"
+    grep -q ': "${BENCH_BATCH:=32768}"' "$bench"
     grep -q 'MODEL_FILE' "$bench"
     grep -q 'REASONING_FORMAT' "$bench"
 }