revert clamp for Run B (without fix) benchmark+eval at conc-64

Run B of A/B test: benchmark + eval WITHOUT dispatch token clamp.
MORI_MAX_DISPATCH_TOKENS_DECODE will be 32 (<256 threshold).
Expected: corrupted output, inflated AL, ~0% gsm8k.

Author: Oseltamivir

benchmarks/multi_node/amd_utils/server_sglang.sh

@@ -248,15 +248,6 @@ if [[ "$DECODE_MTP_SIZE" -gt 0 ]]; then
     MORI_MOE_MAX_INPUT_TOKENS_DECODE=$((MORI_MOE_MAX_INPUT_TOKENS_DECODE * (DECODE_MTP_SIZE + 1)))
 fi
 
-# Clamp dispatch tokens to >= 256 to avoid the low-latency All2All kernel
-# variant in MoRI which silently corrupts outputs at small buffer sizes.
-if [[ "$DECODE_ENABLE_DP" == "true" ]] && [[ "$DECODE_ENABLE_EP" == "true" ]]; then
-    if [[ $MORI_MAX_DISPATCH_TOKENS_DECODE -lt 256 ]]; then
-        echo "[WARN] Clamping MORI_MAX_DISPATCH_TOKENS_DECODE from $MORI_MAX_DISPATCH_TOKENS_DECODE to 256 (All2All kernel threshold)"
-        MORI_MAX_DISPATCH_TOKENS_DECODE=256
-    fi
-fi
-
 # =============================================================================
 # Cluster Topology Configuration
 # =============================================================================

perf-changelog.yaml

@@ -3448,5 +3448,5 @@
 - config-keys:
     - dsr1-fp4-mi355x-sglang-disagg-8k1k-mtp
   description:
-    - "Throwaway: conc-64 DEP8+MTP3 benchmark+eval WITH dispatch token clamp (MORI_MAX_DISPATCH_TOKENS_DECODE >= 256). A/B test for All2All kernel corruption fix."
+    - "Throwaway: conc-64 DEP8+MTP3 benchmark+eval WITHOUT dispatch token clamp (Run B of A/B test). Expect corrupted output / 0pct gsm8k."
   pr-link: https://github.com/SemiAnalysisAI/InferenceX/pull/1659