diff --git a/tests/vault/gear/scheduler-stress-test.cpp b/tests/vault/gear/scheduler-stress-test.cpp
index 974672ab7..003ea90a4 100644
--- a/tests/vault/gear/scheduler-stress-test.cpp
+++ b/tests/vault/gear/scheduler-stress-test.cpp
@@ -296,6 +296,7 @@ SHOW_EXPR(micros);
               {
                 usec LOAD_BASE = 500us;
                 uint CONCURRENCY = 4;
+                bool SCHED_DEPENDS = true;
                 bool showRuns = true;
                 
                 auto testLoad() { return TestChainLoad<>{64}.configureShape_chain_loadBursts(); }
@@ -306,7 +307,7 @@ SHOW_EXPR(micros);
 SHOW_EXPR(stress)
 SHOW_EXPR(delta)
 SHOW_EXPR(time)
-          CHECK (delta > 4.0);
+          CHECK (delta > 2.0);
           CHECK (0.55 > stress and stress > 0.4);
         }
       
diff --git a/tests/vault/gear/stress-test-rig.hpp b/tests/vault/gear/stress-test-rig.hpp
index 60d6496f1..f125755cd 100644
--- a/tests/vault/gear/stress-test-rig.hpp
+++ b/tests/vault/gear/stress-test-rig.hpp
@@ -159,7 +159,9 @@ namespace test {
         void
         configureTest (TestSetup& testSetup, double stressFac)
           {
-            testSetup.withLoadTimeBase (CONF::LOAD_BASE)
+            testSetup.withLoadTimeBase(CONF::LOAD_BASE)
+                     .withBaseExpense (CONF::BASE_EXPENSE)
+                     .withSchedDepends(CONF::SCHED_DEPENDS)
                      .withAdaptedSchedule(stressFac, CONF::CONCURRENCY);
           }
         
@@ -328,6 +330,8 @@ namespace test {
       using usec = std::chrono::microseconds;
       
       usec LOAD_BASE = 500us;
+      usec BASE_EXPENSE = 0us;
+      bool SCHED_DEPENDS = false;
       uint CONCURRENCY = work::Config::getDefaultComputationCapacity();
       double EPSILON      = 0.01;          ///< error bound to abort binary search
       double UPPER_STRESS = 0.6;           ///< starting point for the upper limit, likely to fail
diff --git a/tests/vault/gear/test-chain-load.hpp b/tests/vault/gear/test-chain-load.hpp
index 4126b20cc..cbb4ac354 100644
--- a/tests/vault/gear/test-chain-load.hpp
+++ b/tests/vault/gear/test-chain-load.hpp
@@ -193,8 +193,8 @@ namespace test {
     const Duration SCHEDULE_LEVEL_STEP{_uTicks(1ms)};   ///< time budget to plan for the calculation of each »time level« of jobs
     const Duration SCHEDULE_NODE_STEP{Duration::NIL};   ///< additional time step to include in the plan for each job (node).
     const Duration SCHEDULE_PLAN_STEP{_uTicks(100us)};  ///< time budget to reserve for each node to be planned and scheduled
-    const bool     SCHEDULE_DEPENDENCY = false;         ///< explicitly schedule a dependent job (or rely on NOTIFY)
-    const bool     SCHEDULE_NOTIFY     = true;          ///< explicitly set notify dispatch time to the dependencie's start time. 
+    const bool     SCHED_DEPENDS = false;               ///< explicitly schedule a dependent job (or rely on NOTIFY)
+    const bool     SCHED_NOTIFY  = true;                ///< explicitly set notify dispatch time to the dependency's start time.
     
     inline uint
     defaultConcurr()
@@ -1358,7 +1358,7 @@ namespace test {
         {
           auto round = roundsNeeded (scaleStep);
           Sink sink;
-          size_t scree{0x55DEAD55};
+          size_t scree{sink};
           for ( ; 0 < round; --round)
             boost::hash_combine (scree,scree);
           sink = scree;
@@ -1369,12 +1369,14 @@ namespace test {
       causeMemProcessLoad (uint scaleStep)
         {
           auto [siz,round] = allocNeeded (scaleStep);
-          lib::UninitialisedDynBlock<Sink> memBlock{siz};
-          ++*memBlock.front();
+          lib::UninitialisedDynBlock<size_t> memBlock{siz};
+          Sink sink;
+          *memBlock.front() = sink+1;
           for ( ; 0 < round; --round)
             for (size_t i=0; i<memBlock.size()-1; ++i)
               memBlock[i+1] += memBlock[i];
-          ++*memBlock.back();
+          sink = *memBlock.back();
+          sink++;
         }
       
       double
@@ -1671,8 +1673,8 @@ namespace test {
       FrameRate  levelSpeed_{1, SCHEDULE_LEVEL_STEP};
       FrameRate   planSpeed_{1, SCHEDULE_PLAN_STEP};
       TimeVar   nodeExpense_{SCHEDULE_NODE_STEP};
-      double    schedNotify_{SCHEDULE_NOTIFY? 1.0:0.0};
-      bool     schedDepends_{SCHEDULE_DEPENDENCY};
+      double    schedNotify_{SCHED_NOTIFY? 1.0:0.0};
+      bool     schedDepends_{SCHED_DEPENDS};
       uint  blockLoadFactor_{2};
       size_t      chunkSize_{DEFAULT_CHUNKSIZE};
       TimeVar     startTime_{Time::ANYTIME};
@@ -1850,14 +1852,14 @@ namespace test {
         }
       
       ScheduleCtx&&
-      withScheduleDependency (bool explicitly)
+      withSchedDepends (bool explicitly)
         {
           schedDepends_ = explicitly;
           return move(*this);
         }
       
       ScheduleCtx&&
-      withScheduleNotify (double degree)
+      withSchedNotify (double degree)
         {
           schedNotify_ = degree;
           return move(*this);
diff --git a/wiki/thinkPad.ichthyo.mm b/wiki/thinkPad.ichthyo.mm
index 3fecb6a2e..33a3390e4 100644
--- a/wiki/thinkPad.ichthyo.mm
+++ b/wiki/thinkPad.ichthyo.mm
@@ -107996,16 +107996,13 @@ Date:&#160;&#160;&#160;Thu Apr 20 18:53:17 2023 +0200<br/>
 <node CREATED="1704508460471" ID="ID_1158750541" MODIFIED="1704508465339" TEXT="das ist wirklich verbl&#xfc;ffend"/>
 <node CREATED="1704508466387" ID="ID_1330141096" MODIFIED="1704508564637" TEXT="liegt wohl an der fokussierten Kapazit&#xe4;tssteuerung">
 <richcontent TYPE="NOTE"><html>
-  <head>
-    
-  </head>
+  <head/>
   <body>
     <p>
       anders kann ich mir das nicht erkl&#228;ren, denn es ist ja nun immer nur eine kleine Zahl an Eintr&#228;gen in der Queue, d.h. es besteht durchaus die Gefahr, da&#223; sich Worker schlafen legen, weil die Queue scheinbar leer ist. Tats&#228;chlich passiert das aber nur in den Abschnitten mit geringer Parallelit&#228;t (ja der Doppel-Strang wird nun nur von einem Worker bearbeitet)
     </p>
   </body>
-</html>
-</richcontent>
+</html></richcontent>
 </node>
 </node>
 </node>
@@ -108026,16 +108023,13 @@ Date:&#160;&#160;&#160;Thu Apr 20 18:53:17 2023 +0200<br/>
 <icon BUILTIN="flag-yellow"/>
 <node CREATED="1704506817515" ID="ID_170171194" MODIFIED="1704506848881" TEXT="geht das sinnvollerweise?">
 <richcontent TYPE="NOTE"><html>
-  <head>
-    
-  </head>
+  <head/>
   <body>
     <p>
       mu&#223; dann die Info &#252;ber zwei Ebenen durchreichen...?
     </p>
   </body>
-</html>
-</richcontent>
+</html></richcontent>
 </node>
 </node>
 <node BACKGROUND_COLOR="#eee5c3" COLOR="#990000" CREATED="1704497136967" ID="ID_273051296" MODIFIED="1704497151404" TEXT="Zeitlimit im NOTIFY auf korrekte Handhabung pr&#xfc;fen">
@@ -110653,22 +110647,17 @@ Date:&#160;&#160;&#160;Thu Apr 20 18:53:17 2023 +0200<br/>
 </node>
 <node CREATED="1704503038702" ID="ID_1005585769" MODIFIED="1704503122646">
 <richcontent TYPE="NODE"><html>
-  <head>
-    
-  </head>
+  <head/>
   <body>
     <p>
       die typische Test-Load ~ 500&#181;s braucht im Scheduler-Kontext oft doppelt so lange als kalibriert
     </p>
   </body>
-</html>
-</richcontent>
+</html></richcontent>
 <icon BUILTIN="messagebox_warning"/>
 <node CREATED="1704508579832" ID="ID_940093755" MODIFIED="1704508637214">
 <richcontent TYPE="NODE"><html>
-  <head>
-    
-  </head>
+  <head/>
   <body>
     <p>
       nachdem ich aber nur noch per NOTIFY triggere,
@@ -110680,11 +110669,58 @@ Date:&#160;&#160;&#160;Thu Apr 20 18:53:17 2023 +0200<br/>
       Nur an einer Stelle mit vielen aktiven Workern, treten mal wieder 2ms auf
     </p>
   </body>
-</html>
-</richcontent>
+</html></richcontent>
 <icon BUILTIN="messagebox_warning"/>
+<node CREATED="1704677773725" HGAP="37" ID="ID_1327394317" MODIFIED="1704677791533" TEXT="die stress-test-Messung scheint dem aber zu widersprechen" VSHIFT="29"/>
+<node CREATED="1704677792419" ID="ID_1310162809" MODIFIED="1704677884361" TEXT="in beiden F&#xe4;llen bekommen wir &lt;0.5, aber ohne schedule-depends sogar ehr weniger">
+<richcontent TYPE="NOTE"><html>
+  <head/>
+  <body>
+    <p>
+      0.42 statt 0.56
+    </p>
+  </body>
+</html></richcontent>
+<arrowlink COLOR="#a65a70" DESTINATION="ID_276645813" ENDARROW="Default" ENDINCLINATION="27;-139;" ID="Arrow_ID_316226697" STARTARROW="None" STARTINCLINATION="-474;15;"/>
 </node>
 </node>
+<node CREATED="1704675385917" ID="ID_1698920120" MODIFIED="1704675407523" TEXT="stellt sich die Frage: warum spricht die Test-Load so stark auf concurrency an?"/>
+<node COLOR="#5b280f" CREATED="1704675447594" ID="ID_1820148966" MODIFIED="1704678619828" TEXT="Verdacht: sie arbeiten auf volatile data...">
+<icon BUILTIN="idea"/>
+<icon BUILTIN="button_cancel"/>
+<node CREATED="1704675485440" ID="ID_1790139504" MODIFIED="1704675510440" TEXT="eigentlich w&#xfc;rde ein einziger volatile als &#xbb;Sink&#xab; am Ende gen&#xfc;gen"/>
+<node CREATED="1704675515508" ID="ID_366712394" MODIFIED="1704675555243" TEXT="da wir aber auf Volatiles arbeiten, k&#xf6;nnte das zu geh&#xe4;uften Cache-flushes f&#xfc;hren"/>
+<node CREATED="1704675929596" ID="ID_140425262" MODIFIED="1704675935783" TEXT="Untersuchung...">
+<node CREATED="1704675936723" ID="ID_1880610455" MODIFIED="1704675940590" TEXT="Zustand vorher">
+<node CREATED="1704677929504" ID="ID_1565180226" MODIFIED="1704677937299" TEXT="kalibriert auf 100&#xb5;s:  cpu=41.5947 mem=55.316"/>
+<node CREATED="1704677938619" ID="ID_601391343" MODIFIED="1704677951777" TEXT="kalibriert auf 1ms: cpu=42.3137 mem=60.5797"/>
+</node>
+<node CREATED="1704675941122" ID="ID_161446280" MODIFIED="1704675954731" TEXT="Volatiles aus der Berechnung selber entfernen"/>
+<node CREATED="1704675955736" ID="ID_416913619" MODIFIED="1704675958336" TEXT="Zustand nachher">
+<node CREATED="1704678313156" ID="ID_1131252174" MODIFIED="1704678325038" TEXT="Kein Einflu&#xdf; auf die Kalbrierung feststellbar"/>
+<node CREATED="1704678604021" ID="ID_1814241983" MODIFIED="1704678617576" TEXT="kein Einflu&#xdf; auf das Verhalten im Stre&#xdf;-Test feststellbar"/>
+</node>
+</node>
+<node CREATED="1704678622568" ID="ID_1060142265" MODIFIED="1704678646340" TEXT="These widerlegt">
+<icon BUILTIN="button_cancel"/>
+</node>
+</node>
+</node>
+<node CREATED="1704677622809" ID="ID_276645813" MODIFIED="1704680965552">
+<richcontent TYPE="NODE"><html>
+  <head/>
+  <body>
+    <p>
+      <i>ohne </i>schedule-dependency ist die Performance aber<i>&#160;minimal schlechter</i>
+    </p>
+  </body>
+</html></richcontent>
+<linktarget COLOR="#a65a70" DESTINATION="ID_276645813" ENDARROW="Default" ENDINCLINATION="27;-139;" ID="Arrow_ID_316226697" SOURCE="ID_1310162809" STARTARROW="None" STARTINCLINATION="-474;15;"/>
+<icon BUILTIN="messagebox_warning"/>
+<node CREATED="1704677674802" ID="ID_1422232610" MODIFIED="1704677710441" TEXT="obwohl man dann in Einzelbeobachtung einen glatteren Ablauf mit weniger Contention sieht"/>
+<node CREATED="1704677712357" ID="ID_730665682" MODIFIED="1704677732239" TEXT="und wir die h&#xe4;lfte der gescheduleten Jobs einsparen"/>
+<node CREATED="1704677733514" ID="ID_568047047" MODIFIED="1704677752491" TEXT="k&#xf6;nnte vielleicht daran liegen, da&#xdf; wir dadurch auch Kapazit&#xe4;t in den Schlaf schicken"/>
+</node>
 </node>
 </node>
 </node>