diff --git a/compiler/bnode.nim b/compiler/bnode.nim
index bbb7151d5b..f8884d51cc 100644
--- a/compiler/bnode.nim
+++ b/compiler/bnode.nim
@@ -129,6 +129,26 @@
 ##   processes are 9.9s, and the 28 backend ones are 6.8s of which 6.7s is
 ##   inside the stage bodies — `.t.bif` writing 1.68s and cg's demand-driven
 ##   generation 1.91s are the two largest single items.
+##
+##   The frontend splits (same run, `Stage`/`WriteNif` + the loading slots):
+##
+##       startup (exec+runtime+config)    0.17s    2%
+##       loading imported `.s.bif`        4.57s   46%
+##       writing this module's `.s.bif`   1.84s   18%
+##       sem + parse                      3.41s   34%
+##
+##   So two thirds of the frontend is artifact I/O, not compilation. Within the
+##   loading, `interfHidden` is 1.05s of it: 1.70M hidden-symbol stubs against
+##   0.29M exported ones, built by every `nim m` for every module it imports.
+##   That table is reached ONLY through `modulegraphs.interfSelect` when
+##   `optImportHidden` is set, which happens in exactly one place — an
+##   `import x {.all.}`. Skipping it outright (measured with a probe, not a
+##   guess) takes `InterfTables` 1161ms -> 80ms, the frontend 9.98s -> 8.93s and
+##   the whole Atlas build 22.19s -> 20.47s. Doing it CORRECTLY means populating
+##   the table lazily on first `interfSelect(true)` rather than deciding
+##   up front — a macro-generated `{.all.}` import cannot be seen syntactically,
+##   and guessing wrong loses symbols silently. `loaderCtx` is the hook: the
+##   module index is still in the DecodeContext after loading.
 ## * `-d:icBridgeOnly` builds the buffer but generates off the tree, which
 ##   separates the ENCODER's cost from the READER's. Encoding is free — it does
 ##   not show in wall time at all.
diff --git a/compiler/icprof.nim b/compiler/icprof.nim
index f9ce4ddc50..fd19a45124 100644
--- a/compiler/icprof.nim
+++ b/compiler/icprof.nim
@@ -44,7 +44,10 @@ when defined(icBNodeProf):
       tStage,
       tLowerOwned, tLowerHooks, tLowerWrite,
       tCgGen, tCgInit, tCgFinish, tCgWrite,
-      tMergeStage, tEmitRender, tLinkStage
+      tMergeStage, tEmitRender, tLinkStage,
+      # `nim m` (the frontend): the sem pass as a whole, and writing the module's
+      # `.s.bif`. `Stage - WriteNif - <the loading slots>` is then sem proper.
+      tWriteNif
 
   let procStart = getMonoTime()
     ## Set when this module initialises, i.e. essentially at process start, so
diff --git a/compiler/main.nim b/compiler/main.nim
index cddf2fc96b..b589cd62de 100644
--- a/compiler/main.nim
+++ b/compiler/main.nim
@@ -29,6 +29,7 @@ when defined(nimPreviewSlimSystem):
 import ../dist/checksums/src/checksums/sha1
 
 import pipelines
+import icprof
 from icconfig import produceIcConfig, ensureIcConfig
 
 when not defined(nimKochBootstrap):
@@ -445,7 +446,9 @@ proc mainCommand*(graph: ModuleGraph) =
     # per-module compilation model cannot provide (yet); methods dispatch
     # through the classic if-chain dispatchers instead
     excl conf.features, Feature.vtables
-    commandCheck(graph)
+    # `tStage` for a `nim m` process, so `Process - Stage` is its real startup
+    # (exec, runtime init, config replay) rather than its whole runtime.
+    timed tStage: commandCheck(graph)
   of cmdNifC:
     setUseIc(true)
     excl conf.features, Feature.vtables
diff --git a/compiler/pipelines.nim b/compiler/pipelines.nim
index 0837355467..e155ac2f98 100644
--- a/compiler/pipelines.nim
+++ b/compiler/pipelines.nim
@@ -10,6 +10,7 @@ when not defined(nimKochBootstrap):
   import "../dist/nimony/src/lib" / bitabs
 
 import pipelineutils
+import icprof
 
 import ../dist/checksums/src/checksums/sha1
 
@@ -336,11 +337,12 @@ proc processPipelineModule*(graph: ModuleGraph; module: PSym; idgen: IdGenerator
       # `injectDestructorCalls` and top-level locals were never destroyed.
       let moduleFlags =
         if sfInjectDestructors in module.flags: ModFlagInjectDestructors else: 0'i32
-      writeNifModule(graph.config, module.position.int32, topLevelStmts, graph.opsLog,
-                     replayActions, implDeps, reexportedModuleSyms(graph, module),
-                     genericOffers, typeOffers, resolvedImportDeps, firstUnusedId,
-                     expansions, moduleFlags,
-                     reexportedLocalSyms(graph, module))
+      timed tWriteNif:
+        writeNifModule(graph.config, module.position.int32, topLevelStmts, graph.opsLog,
+                       replayActions, implDeps, reexportedModuleSyms(graph, module),
+                       genericOffers, typeOffers, resolvedImportDeps, firstUnusedId,
+                       expansions, moduleFlags,
+                       reexportedLocalSyms(graph, module))
       # The module's REAL direct imports (incl. macro-generated) for `nim ic`'s
       # graph re-derivation; see ast2nif.writeSemDeps / semdata.addImportFileDep.
       var semDepPaths: seq[string] = @[]