Increase -A default to 4MB.

This gives a small increase in performance under most circumstances.

For single threaded GC the improvement is on the order of 1-2%.

For multi threaded GC the results are quite noisy but seem to
fall into the same ballpark.

Fixes #16499

4 files changed, 23 insertions, 15 deletions

diff --git a/docs/users_guide/runtime_control.rst b/docs/users_guide/runtime_control.rst
index 5d3b24adac..49dc6a5441 100644
--- a/docs/users_guide/runtime_control.rst
+++ b/docs/users_guide/runtime_control.rst
@@ -420,7 +420,7 @@ performance.
 
 .. rts-flag:: -A ⟨size⟩
 
-    :default: 1MB
+    :default: 4MB
 
     .. index::
        single: allocation area, size
@@ -429,15 +429,22 @@ performance.
     collector. The allocation area (actually generation 0 step 0) is
     fixed and is never resized (unless you use :rts-flag:`-H [⟨size⟩]`, below).
 
-    Increasing the allocation area size may or may not give better
-    performance (a bigger allocation area means worse cache behaviour
-    but fewer garbage collections and less promotion).
+    Optimal settings depend on the actual machine, program, and other RTS options.
+    Increasing the allocation area size means worse cache behaviour
+    but fewer garbage collections and less promotion.
+
+    In general settings >= 4MB can reduce performance in some cases, in particular for single
+    threaded operation. However in a parallel setting increasing the allocation area
+    to ``16MB``, or even ``64MB`` can increase gc throughput significantly.
 
     With only 1 generation (e.g. ``-G1``, see :rts-flag:`-G ⟨generations⟩`) the
     ``-A`` option specifies the minimum allocation area, since the actual size
     of the allocation area will be resized according to the amount of data in
     the heap (see :rts-flag:`-F ⟨factor⟩`, below).
 
+    When heap profiling using a smaller allocation area can increase accuracy as more frequent
+    major garbage collections also results in more frequent heap snapshots
+
 .. rts-flag:: -AL ⟨size⟩
 
     :default: :rts-flag:`-A <-A ⟨size⟩>` value
diff --git a/rts/RtsFlags.c b/rts/RtsFlags.c
index b23b19752b..044e7742c4 100644
--- a/rts/RtsFlags.c
+++ b/rts/RtsFlags.c
@@ -153,10 +153,11 @@ void initRtsFlagsDefaults(void)
     RtsFlags.GcFlags.stkChunkSize       = (32 * 1024) / sizeof(W_);
     RtsFlags.GcFlags.stkChunkBufferSize = (1 * 1024) / sizeof(W_);
 
-    RtsFlags.GcFlags.minAllocAreaSize   = (1024 * 1024)       / BLOCK_SIZE;
+    /* -A default. See #16499 for a discussion about the tradeoffs */
+    RtsFlags.GcFlags.minAllocAreaSize   = (4 * 1024 * 1024)       / BLOCK_SIZE;
     RtsFlags.GcFlags.largeAllocLim      = 0; /* defaults to minAllocAreasize */
     RtsFlags.GcFlags.nurseryChunkSize   = 0;
-    RtsFlags.GcFlags.minOldGenSize      = (1024 * 1024)       / BLOCK_SIZE;
+    RtsFlags.GcFlags.minOldGenSize      = (1024 * 1024)       / BLOCK_SIZE; /* -O default */
     RtsFlags.GcFlags.maxHeapSize        = 0;    /* off by default */
     RtsFlags.GcFlags.heapLimitGrace     = (1024 * 1024);
     RtsFlags.GcFlags.heapSizeSuggestion = 0;    /* none */
diff --git a/testsuite/tests/profiling/should_run/all.T b/testsuite/tests/profiling/should_run/all.T
index dec77add28..ad10baac13 100644
--- a/testsuite/tests/profiling/should_run/all.T
+++ b/testsuite/tests/profiling/should_run/all.T
@@ -28,7 +28,7 @@ test('heapprof001',
      compile_and_run, [''])
 
 test('T2592',
-     [only_ways(['profasm']), extra_run_opts('+RTS -M1m -RTS'), exit_code(251)],
+     [only_ways(['profasm']), extra_run_opts('+RTS -M1m -A1m -RTS'), exit_code(251)],
      compile_and_run, [''])
 
 test('T3001', [only_ways(['prof_hb']), extra_ways(['prof_hb'])],
diff --git a/testsuite/tests/rts/T9579/Makefile b/testsuite/tests/rts/T9579/Makefile
index 23177ee125..e1bf2d3079 100644
--- a/testsuite/tests/rts/T9579/Makefile
+++ b/testsuite/tests/rts/T9579/Makefile
@@ -3,43 +3,43 @@ include $(TOP)/mk/boilerplate.mk
 include $(TOP)/mk/test.mk
 
 T9579_stackoverflow_rtsnone:
-	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=none -fforce-recomp -with-rtsopts -K1m \
+	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=none -fforce-recomp -with-rtsopts "-A1M -K1m" \
 		-outputdir tmp_T9579_stackoverflow_rtsnone \
 		StackOverflow.hs -o T9579_stackoverflow_rtsnone
 
 T9579_stackoverflow_rtssome:
-	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=some -fforce-recomp -with-rtsopts -K1m \
+	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=some -fforce-recomp -with-rtsopts "-A1M -K1m" \
 		-outputdir tmp_T9579_stackoverflow_rtssome \
 		StackOverflow.hs -o T9579_stackoverflow_rtssome
 
 T9579_stackoverflow_rtsall:
-	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=all  -fforce-recomp -with-rtsopts -K1m \
+	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=all  -fforce-recomp -with-rtsopts "-A1M -K1m" \
 		-outputdir tmp_T9579_stackoverflow_rtsall \
 		StackOverflow.hs -o T9579_stackoverflow_rtsall
 
 T9579_stackoverflow_rtsall_no_suggestions:
-	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=all -fforce-recomp -with-rtsopts -K1m \
+	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=all -fforce-recomp -with-rtsopts "-A1M -K1m" \
 		-no-rtsopts-suggestions \
 		-outputdir tmp_T9579_stackoverflow_rtsall_no_suggestions \
 		StackOverflow.hs -o T9579_stackoverflow_rtsall_no_suggestions
 
 T9579_outofheap_rtsnone:
-	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=none -fforce-recomp -with-rtsopts -M1m \
+	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=none -fforce-recomp -with-rtsopts "-A1M -M1m" \
 		-outputdir tmp_T9579_outofheap_rtsnone \
 		OutOfHeap.hs -o T9579_outofheap_rtsnone
 
 T9579_outofheap_rtssome:
-	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=some -fforce-recomp -with-rtsopts -M1m \
+	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=some -fforce-recomp -with-rtsopts "-A1M -M1m" \
 		-outputdir tmp_T9579_outofheap_rtssome \
 		OutOfHeap.hs -o T9579_outofheap_rtssome
 
 T9579_outofheap_rtsall:
-	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=all -fforce-recomp -with-rtsopts -M1m \
+	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=all -fforce-recomp -with-rtsopts "-A1M -M1m" \
 		-outputdir tmp_T9579_outofheap_rtsall \
 		OutOfHeap.hs -o T9579_outofheap_rtsall
 
 T9579_outofheap_rtsall_no_suggestions:
-	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=all -fforce-recomp -with-rtsopts -M1m \
+	'$(TEST_HC)' $(TEST_HC_OPTS) -v0 -rtsopts=all -fforce-recomp -with-rtsopts "-A1M -M1m" \
 		-no-rtsopts-suggestions \
 		-outputdir tmp_T9579_outofheap_rtsall_no_suggestions \
 		OutOfHeap.hs -o T9579_outofheap_rtsall_no_suggestions