large: ubatch 4096 (match v3), add --no-mmap; sync bench harness prod config

scripts/bench-kairic-ab.sh

@@ -7,9 +7,10 @@
 #                 greedy argmax fast path ON)
 #   nommap        runner + --no-mmap
 #   nofast        runner + fast path OFF
-#   prod          production profile (b4096/ub2048, reasoning on/deepseek,
+#   prod          production profile (b4096/ub4096, --no-mmap, reasoning on/deepseek,
 #                 fast path OFF, preserve_thinking)
-#   prodnommap    prod + --no-mmap
+#   prodnommap    (obsolete: identical to prod since ubatch 4096 + --no-mmap
+#                 moved into the production profile)
 #
 # Usage: ./scripts/bench-kairic-ab.sh <cfg> [ctx] [prompt_chars]
 # Example: ./scripts/bench-kairic-ab.sh prod 16384 53000
@@ -61,7 +62,7 @@ case "$CFG" in
   runner)  export LLAMA_TARGET_GREEDY_ARGMAX_FASTPATH=1; B=2048; UB=512;  REA=off;  RFMT=none;     EXTRA="" ;;
   nommap)  export LLAMA_TARGET_GREEDY_ARGMAX_FASTPATH=1; B=2048; UB=512;  REA=off;  RFMT=none;     EXTRA=""; MMAP="--no-mmap" ;;
   nofast)  export LLAMA_TARGET_GREEDY_ARGMAX_FASTPATH=0; B=2048; UB=512;  REA=off;  RFMT=none;     EXTRA="" ;;
-  prod)    export LLAMA_TARGET_GREEDY_ARGMAX_FASTPATH=0; B=4096; UB=2048; REA=on;   RFMT=deepseek; EXTRA="--chat-template-kwargs {\"preserve_thinking\":\"true\"}" ;;
+  prod)    export LLAMA_TARGET_GREEDY_ARGMAX_FASTPATH=0; B=4096; UB=4096; REA=on;   RFMT=deepseek; EXTRA="--chat-template-kwargs {\"preserve_thinking\":\"true\"}"; MMAP="--no-mmap" ;;
   prodnommap) export LLAMA_TARGET_GREEDY_ARGMAX_FASTPATH=0; B=4096; UB=2048; REA=on; RFMT=deepseek; EXTRA="--chat-template-kwargs {\"preserve_thinking\":\"true\"}"; MMAP="--no-mmap" ;;
 esac
 

scripts/frank-default-large.env

@@ -51,9 +51,10 @@ LISTEN_PORT=52625
 CTX_SIZE=262144
 PARALLEL=1
 # Local override: the pinned runner uses 2048/512; production chat uses a
-# larger 4096/2048 batch (unified memory has headroom).
+# larger 4096/4096 batch (unified memory has headroom; matches the retired
+# v3 profile).
 BATCH_SIZE=4096
-UBATCH_SIZE=2048
+UBATCH_SIZE=4096
 TARGET_THREADS=16
 BATCH_THREADS=32
 CACHE_TYPE_K=f16

scripts/serve-frank-large.sh

@@ -12,11 +12,13 @@
 # See docs/kairic-edge.md for the build runbook and binary identity.
 #
 # This script replicates the release pinned runner
-# (ROCmFPX-kairic/scripts/run-kairic-edge-gfx1151.sh) verbatim. The only
+# (ROCmFPX-kairic/scripts/run-kairic-edge-gfx1151.sh). The deliberate
-# deliberate differences: --host/--port come from BIND_ADDR/LISTEN_PORT
+# differences: --host/--port come from BIND_ADDR/LISTEN_PORT (safe names;
-# (safe names; the ROCm HIP runtime reads a bare HOST env var, see the
+# the ROCm HIP runtime reads a bare HOST env var, see the 2026-08-13 TEMP
-# 2026-08-13 TEMP incident) and LD_LIBRARY_PATH uses this machine's split
+# incident), LD_LIBRARY_PATH uses this machine's split 7.15 SDK layout,
-# 7.15 SDK layout.
+# the profile runs --no-mmap, batch 4096 / ubatch 4096, reasoning on with
+# deepseek format and preserve_thinking, and the target greedy argmax fast
+# path is disabled for non-greedy chat sampling.
 #
 # Tunables come from frank-default-large.env in this directory unless
 # FRANK_LARGE_ENV_FILE overrides that path.
@@ -48,7 +50,7 @@ fi
 : "${CTX_SIZE:=262144}"
 : "${PARALLEL:=1}"
 : "${BATCH_SIZE:=4096}"
-: "${UBATCH_SIZE:=2048}"
+: "${UBATCH_SIZE:=4096}"
 : "${TARGET_THREADS:=16}"
 : "${BATCH_THREADS:=32}"
 : "${CACHE_TYPE_K:=f16}"
@@ -126,7 +128,9 @@ export PROMPTFORGE_OUTPUT_K8_STRICT_GREEDY=0
 # 4, deterministic sampler). Local overrides for production chat:
 #   - reasoning on + deepseek format (thinking goes to message.reasoning_content,
 #     matching the retired v3 service so the fleet's reasoning clients work)
-#   - batch 4096 / ubatch 2048 (larger than the runner's 2048/512)
+#   - batch 4096 / ubatch 4096 (larger than the runner's 2048/512; matches v3)
+#   - --no-mmap (local choice; benchmarks show no perf difference versus
+#     mmap, kept per operator preference)
 #   - chat_template_kwargs preserve_thinking=true (preserve thinking tokens
 #     in the conversation history across turns)
 # Sampling is overridden per request by clients; the card recommends Qwen
@@ -152,6 +156,7 @@ exec "$LLAMA_BIN" \
     --cache-ram "$CACHE_RAM" \
     --cache-prompt \
     --cache-idle-slots \
+    --no-mmap \
     --metrics \
     --kairic-edge \
     --spec-type draft-mtp \