Bake RMSE-winning recipe into feedsim_dlrm + feedsim_autoscale_dlrm

Differential Revision: D115652574

Author: excelle08

benchpress/config/jobs.yml

@@ -513,7 +513,12 @@
     - 'io_mean=200'
     - 'workload=dlrm'
     - 'dlrm_model=models/dlrm_small.pt'
-    - 'dlrm_batch_size=64'
+    # RMSE-winning recipe (2026-08, supersedes t43 c7): minimizes all-metrics
+    # RMSE vs prod multifeed_aggregator (uArch + instruction mix + hot-function)
+    # on BGM & Grace, and preserves gen-over-gen ranking across all 6 LSSTs.
+    # Pairs with the integer-domain feature-extractor helpers + memory-streaming
+    # stride-sweep (N=16, code default) earlier in this stack.
+    - 'dlrm_batch_size=32'
     - 'dlrm_threads=1'
     - 'dlrm_inferences=1'
     - 'client_side_features=0'
@@ -522,18 +527,18 @@
     - 'client_feature_seed=42'
     - 'client_num_dense=13'
     - 'client_num_sparse=26'
-    - 'feature_complexity=5'
+    - 'feature_complexity=8'
     - 'num_stories=400'
-    - 'extractors_per_story=280'
+    - 'extractors_per_story=240'
     - 'story_processors_per_story=2'
-    - 'stories_per_processor_pass=100'
+    - 'stories_per_processor_pass=150'
     - 'silesia_dir=silesia'
     - 'stories_per_request=10'
     - 'mock_tls=1'
-    - 'mock_zstd_frac=0.75'
+    - 'mock_zstd_frac=0.9'
     - 'mock_keepalive_interval_ms=200'
-    - 'rpc_fanout_scale=0.05'
-    - 'server_zstd=0'
+    - 'rpc_fanout_scale=0.10'
+    - 'server_zstd=1'
     - 'sla_p95_ms=700'
     # Driver pipeline depth (max outstanding requests per driver connection).
     # Default 1. Raise (e.g. 2) if the final phase saturates neither CPU nor SLA
@@ -607,11 +612,12 @@
     - 'io_mean=200'
     - 'workload=dlrm'
     - 'dlrm_model=models/dlrm_small.pt'
-    # t43 c7: bs=64 is the balanced choice across both gen-over-gen
-    # correlation (BGM/CPL = 4.25× → 2nd best of 12 cells) and prod
-    # hot-function gap (CPL 4th, BGM 5th). bs=32 wins peak QPS alone but
-    # breaks correlation (BGM stretches to 4.75× CPL).
-    - 'dlrm_batch_size=64'
+    # RMSE-winning recipe (2026-08, supersedes t43 c7): minimizes all-metrics
+    # RMSE vs prod (uArch + instruction mix + hot-function) on BGM & Grace.
+    # bs=32 (with the integer-domain extractor helpers + stride-sweep earlier in
+    # this stack) raises SIMD toward prod and, at eps=240, preserves gen-over-gen
+    # ranking across all 6 LSSTs (validated 2026-08-11).
+    - 'dlrm_batch_size=32'
     - 'dlrm_threads=1'
     - 'dlrm_inferences=1'
     - 'client_side_features=0'
@@ -620,28 +626,27 @@
     - 'client_feature_seed=42'
     - 'client_num_dense=13'
     - 'client_num_sparse=26'
-    - 'feature_complexity=5'
-    # 400 x 280 = 112K extractor calls/req. Combined with the scaled-up
-    # story-processor pipeline (below) this matches prod's FE:story CPU
-    # ratio of ~1.3:1 (multifeed: FE 29%, story 22% on BGM Strobelight).
+    - 'feature_complexity=8'
+    # 400 x 240 = 96K extractor calls/req. eps=240 lands Ranking-FeatureExtraction
+    # near prod on BGM (~22% vs 22.8%); the integer-domain helpers made each call
+    # cheaper, so eps rose from 280 to hold the FE share after batch32/zstd/rpc grew.
     - 'num_stories=400'
-    - 'extractors_per_story=280'
-    # Story-processor pipeline scaled 8x from t32 first attempt (was
-    # 1 pass/story * 25 stories/pass = 25 cmpns/story → 1.1% Ranking-Story
-    # on BGM, vs prod 9.3%). Bumped to 2 passes * 100 stories = 200
-    # comparisons/story, target ~9% Ranking-Story share.
+    - 'extractors_per_story=240'
+    # Story-processor pipeline: 2 passes * 150 stories = 300 comparisons/story,
+    # target ~9-10% Ranking-Story share (≈ prod 9.3% on BGM).
     - 'story_processors_per_story=2'
-    - 'stories_per_processor_pass=100'
+    - 'stories_per_processor_pass=150'
     - 'silesia_dir=silesia'
     - 'stories_per_request=10'
-    # t43 c7: TLS+keepalive on mock_services, 75% of channels compress,
-    # rpc_fanout_scale=0.05 (~188 RPCs/session). Server-side ZSTD off
-    # (mock-side ZSTD provides the prod-realistic Compression share).
+    # TLS+keepalive on mock_services; mock_zstd_frac=0.9 + server_zstd=1 land
+    # Compression ≈ prod (BGM ~10% vs 10.7%); rpc_fanout_scale=0.10 lands
+    # RPC-AsyncIO ≈ prod (~15%). DRIVER_TLS stays off (default): it overshoots
+    # Encryption without moving throughput.
     - 'mock_tls=1'
-    - 'mock_zstd_frac=0.75'
+    - 'mock_zstd_frac=0.9'
     - 'mock_keepalive_interval_ms=200'
-    - 'rpc_fanout_scale=0.05'
-    - 'server_zstd=0'
+    - 'rpc_fanout_scale=0.10'
+    - 'server_zstd=1'
     - 'sla_p95_ms=700'
     # Driver pipeline depth (max outstanding requests per driver connection).
     # Default 1. Raise (e.g. 2) if the final phase saturates neither CPU nor SLA