[llvm-profdata] Add block percent to detailed summary #105915

ellishg · 2024-08-24T02:03:38Z

In the detailed summary, we want to report the fraction of total blocks that account for each percentile cutoff. Also, use ProfileSummary::printSummary() in Instr profiles for consistancy.

llvmbot · 2024-08-24T02:04:07Z

@llvm/pr-subscribers-pgo

@llvm/pr-subscribers-llvm-ir

Author: Ellis Hoag (ellishg)

Changes

In the detailed summary, we want to report the fraction of total blocks that account for each percentile cutoff. Also, use ProfileSummary::printSummary() in Instr profiles for consistancy.

Patch is 25.84 KiB, truncated to 20.00 KiB below, full version: https://github.com/llvm/llvm-project/pull/105915.diff

10 Files Affected:

(modified) compiler-rt/test/profile/Linux/Inputs/instrprof-value-merge.c (+2)
(modified) compiler-rt/test/profile/Linux/binary-id.c (+10)
(modified) compiler-rt/test/profile/Linux/profile-version.c (+4)
(modified) llvm/lib/IR/ProfileSummary.cpp (+6-5)
(modified) llvm/test/tools/llvm-profdata/cs-sample-nested-profile.test (+34-34)
(modified) llvm/test/tools/llvm-profdata/general.proftext (+11-11)
(modified) llvm/test/tools/llvm-profdata/sample-summary.test (+17-17)
(modified) llvm/test/tools/llvm-profdata/suppl-instr-with-sample.test (+16-16)
(modified) llvm/test/tools/llvm-profdata/vtable-value-prof.test (+8-6)
(modified) llvm/tools/llvm-profdata/llvm-profdata.cpp (+2-7)

diff --git a/compiler-rt/test/profile/Linux/Inputs/instrprof-value-merge.c b/compiler-rt/test/profile/Linux/Inputs/instrprof-value-merge.c
index 6f5b4790cca564..3efbcf26e81ff7 100644
--- a/compiler-rt/test/profile/Linux/Inputs/instrprof-value-merge.c
+++ b/compiler-rt/test/profile/Linux/Inputs/instrprof-value-merge.c
@@ -54,6 +54,8 @@ int main(int argc, char *argv[]) {
 // CHECK-NEXT: Total functions: 3
 // CHECK-NEXT: Maximum function count: 327
 // CHECK-NEXT: Maximum internal block count: 297
+// CHECK-NEXT: Total number of blocks: 8
+// CHECK-NEXT: Total count: 805
 // CHECK-NEXT: Statistics for indirect call sites profile:
 // CHECK-NEXT:   Total number of sites: 3
 // CHECK-NEXT:   Total number of sites with values: 3
diff --git a/compiler-rt/test/profile/Linux/binary-id.c b/compiler-rt/test/profile/Linux/binary-id.c
index 9bff9004e97a7e..93a6852e626eb1 100644
--- a/compiler-rt/test/profile/Linux/binary-id.c
+++ b/compiler-rt/test/profile/Linux/binary-id.c
@@ -62,6 +62,8 @@ int main() {
 // BINARY-ID-RAW-PROF-NEXT: Total functions: 3
 // BINARY-ID-RAW-PROF-NEXT: Maximum function count: 1
 // BINARY-ID-RAW-PROF-NEXT: Maximum internal block count: 0
+// BINARY-ID-RAW-PROF-NEXT: Total number of blocks: 3
+// BINARY-ID-RAW-PROF-NEXT: Total count: 3
 // BINARY-ID-RAW-PROF-NEXT: Binary IDs:
 // BINARY-ID-RAW-PROF-NEXT: {{[0-9a-f]+}}
 
@@ -69,6 +71,8 @@ int main() {
 // BINARY-ID-MERGE-PROF-NEXT: Total functions: 3
 // BINARY-ID-MERGE-PROF-NEXT: Maximum function count: 3
 // BINARY-ID-MERGE-PROF-NEXT: Maximum internal block count: 0
+// BINARY-ID-MERGE-PROF-NEXT: Total number of blocks: 3
+// BINARY-ID-MERGE-PROF-NEXT: Total count: 9
 // BINARY-ID-MERGE-PROF-NEXT: Binary IDs:
 // BINARY-ID-MERGE-PROF-NEXT: {{[0-9a-f]+}}
 
@@ -76,6 +80,8 @@ int main() {
 // BINARY-ID-INDEXED-PROF-NEXT: Total functions: 3
 // BINARY-ID-INDEXED-PROF-NEXT: Maximum function count: 3
 // BINARY-ID-INDEXED-PROF-NEXT: Maximum internal block count: 0
+// BINARY-ID-INDEXED-PROF-NEXT: Total number of blocks: 3
+// BINARY-ID-INDEXED-PROF-NEXT: Total count: 9
 // BINARY-ID-INDEXED-PROF-NEXT: Binary IDs:
 // BINARY-ID-INDEXED-PROF-NEXT: {{[0-9a-f]+}}
 
@@ -83,6 +89,8 @@ int main() {
 // BINARY-ID-SHARE-RAW-PROF-NEXT: Total functions: 3
 // BINARY-ID-SHARE-RAW-PROF-NEXT: Maximum function count: 1
 // BINARY-ID-SHARE-RAW-PROF-NEXT: Maximum internal block count: 0
+// BINARY-ID-SHARE-RAW-PROF-NEXT: Total number of blocks: 3
+// BINARY-ID-SHARE-RAW-PROF-NEXT: Total count: 3
 // BINARY-ID-SHARE-RAW-PROF-NEXT: Binary IDs:
 // BINARY-ID-SHARE-RAW-PROF-NEXT: {{[0-9a-f]+}}
 // BINARY-ID-SHARE-RAW-PROF-NEXT: {{[0-9a-f]+}}
@@ -92,6 +100,8 @@ int main() {
 // BINARY-ID-SHARE-INDEXED-PROF-NEXT: Total functions: 3
 // BINARY-ID-SHARE-INDEXED-PROF-NEXT: Maximum function count: 1
 // BINARY-ID-SHARE-INDEXED-PROF-NEXT: Maximum internal block count: 0
+// BINARY-ID-SHARE-INDEXED-PROF-NEXT: Total number of blocks: 3
+// BINARY-ID-SHARE-INDEXED-PROF-NEXT: Total count: 3
 // BINARY-ID-SHARE-INDEXED-PROF-NEXT: Binary IDs:
 // BINARY-ID-SHARE-INDEXED-PROF-NEXT: {{[0-9a-f]+}}
 // BINARY-ID-SHARE-INDEXED-PROF-NEXT: {{[0-9a-f]+}}
diff --git a/compiler-rt/test/profile/Linux/profile-version.c b/compiler-rt/test/profile/Linux/profile-version.c
index bffc602387ad50..7e20d1bee6cc5e 100644
--- a/compiler-rt/test/profile/Linux/profile-version.c
+++ b/compiler-rt/test/profile/Linux/profile-version.c
@@ -25,10 +25,14 @@ int main() {
 // RAW-PROF-NEXT: Total functions: 3
 // RAW-PROF-NEXT: Maximum function count: 1
 // RAW-PROF-NEXT: Maximum internal block count: 0
+// RAW-PROF-NEXT: Total number of blocks: 3
+// RAW-PROF-NEXT: Total count: 3
 // RAW-PROF-NEXT: Profile version: {{[0-9]+}}
 
 // INDEXED-PROF: Instrumentation level: Front-end
 // INDEXED-PROF-NEXT: Total functions: 3
 // INDEXED-PROF-NEXT: Maximum function count: 3
 // INDEXED-PROF-NEXT: Maximum internal block count: 0
+// INDEXED-PROF-NEXT: Total number of blocks: 3
+// INDEXED-PROF-NEXT: Total count: 9
 // INDEXED-PROF-NEXT: Profile version: {{[0-9]+}}
diff --git a/llvm/lib/IR/ProfileSummary.cpp b/llvm/lib/IR/ProfileSummary.cpp
index acb4c52e8918fc..59f29e5982dfb0 100644
--- a/llvm/lib/IR/ProfileSummary.cpp
+++ b/llvm/lib/IR/ProfileSummary.cpp
@@ -251,7 +251,7 @@ ProfileSummary *ProfileSummary::getFromMD(Metadata *MD) {
 void ProfileSummary::printSummary(raw_ostream &OS) const {
   OS << "Total functions: " << NumFunctions << "\n";
   OS << "Maximum function count: " << MaxFunctionCount << "\n";
-  OS << "Maximum block count: " << MaxCount << "\n";
+  OS << "Maximum internal block count: " << MaxInternalCount << "\n";
   OS << "Total number of blocks: " << NumCounts << "\n";
   OS << "Total count: " << TotalCount << "\n";
 }
@@ -259,9 +259,10 @@ void ProfileSummary::printSummary(raw_ostream &OS) const {
 void ProfileSummary::printDetailedSummary(raw_ostream &OS) const {
   OS << "Detailed summary:\n";
   for (const auto &Entry : DetailedSummary) {
-    OS << Entry.NumCounts << " blocks with count >= " << Entry.MinCount
-       << " account for "
-       << format("%0.6g", (float)Entry.Cutoff / Scale * 100)
-       << " percentage of the total counts.\n";
+    OS << format("%lu blocks (%.2f%%) with count >= %lu account for %0.6g%% of "
+                 "the total counts.\n",
+                 Entry.NumCounts,
+                 NumCounts ? (100.f * Entry.NumCounts / NumCounts) : 0,
+                 Entry.MinCount, 100.f * Entry.Cutoff / Scale);
   }
 }
diff --git a/llvm/test/tools/llvm-profdata/cs-sample-nested-profile.test b/llvm/test/tools/llvm-profdata/cs-sample-nested-profile.test
index 7b01324219115c..d2b07cf05fd240 100644
--- a/llvm/test/tools/llvm-profdata/cs-sample-nested-profile.test
+++ b/llvm/test/tools/llvm-profdata/cs-sample-nested-profile.test
@@ -153,47 +153,47 @@ RUN: llvm-profdata show -sample -detailed-summary %t3.proftext | FileCheck %s -c
 
 ; SUMMARY:      Total functions: 4
 ; SUMMARY-NEXT: Maximum function count: 32
-; SUMMARY-NEXT: Maximum block count: 362830
+; SUMMARY-NEXT: Maximum internal block count: 0
 ; SUMMARY-NEXT: Total number of blocks: 16
 ; SUMMARY-NEXT: Total count: 772562
 ; SUMMARY-NEXT: Detailed summary:
-; SUMMARY-NEXT: 1 blocks with count >= 362830 account for 1 percentage of the total counts.
-; SUMMARY-NEXT: 1 blocks with count >= 362830 account for 10 percentage of the total counts.
-; SUMMARY-NEXT: 1 blocks with count >= 362830 account for 20 percentage of the total counts.
-; SUMMARY-NEXT: 1 blocks with count >= 362830 account for 30 percentage of the total counts.
-; SUMMARY-NEXT: 1 blocks with count >= 362830 account for 40 percentage of the total counts.
-; SUMMARY-NEXT: 2 blocks with count >= 362805 account for 50 percentage of the total counts.
-; SUMMARY-NEXT: 2 blocks with count >= 362805 account for 60 percentage of the total counts.
-; SUMMARY-NEXT: 2 blocks with count >= 362805 account for 70 percentage of the total counts.
-; SUMMARY-NEXT: 2 blocks with count >= 362805 account for 80 percentage of the total counts.
-; SUMMARY-NEXT: 2 blocks with count >= 362805 account for 90 percentage of the total counts.
-; SUMMARY-NEXT: 3 blocks with count >= 23327 account for 95 percentage of the total counts.
-; SUMMARY-NEXT: 4 blocks with count >= 23324 account for 99 percentage of the total counts.
-; SUMMARY-NEXT: 4 blocks with count >= 23324 account for 99.9 percentage of the total counts.
-; SUMMARY-NEXT: 11 blocks with count >= 24 account for 99.99 percentage of the total counts.
-; SUMMARY-NEXT: 16 blocks with count >= 10 account for 99.999 percentage of the total counts.
-; SUMMARY-NEXT: 16 blocks with count >= 10 account for 99.9999 percentage of the total counts.
+; SUMMARY-NEXT: 1 blocks (6.25%) with count >= 362830 account for 1% of the total counts.
+; SUMMARY-NEXT: 1 blocks (6.25%) with count >= 362830 account for 10% of the total counts.
+; SUMMARY-NEXT: 1 blocks (6.25%) with count >= 362830 account for 20% of the total counts.
+; SUMMARY-NEXT: 1 blocks (6.25%) with count >= 362830 account for 30% of the total counts.
+; SUMMARY-NEXT: 1 blocks (6.25%) with count >= 362830 account for 40% of the total counts.
+; SUMMARY-NEXT: 2 blocks (12.50%) with count >= 362805 account for 50% of the total counts.
+; SUMMARY-NEXT: 2 blocks (12.50%) with count >= 362805 account for 60% of the total counts.
+; SUMMARY-NEXT: 2 blocks (12.50%) with count >= 362805 account for 70% of the total counts.
+; SUMMARY-NEXT: 2 blocks (12.50%) with count >= 362805 account for 80% of the total counts.
+; SUMMARY-NEXT: 2 blocks (12.50%) with count >= 362805 account for 90% of the total counts.
+; SUMMARY-NEXT: 3 blocks (18.75%) with count >= 23327 account for 95% of the total counts.
+; SUMMARY-NEXT: 4 blocks (25.00%) with count >= 23324 account for 99% of the total counts.
+; SUMMARY-NEXT: 4 blocks (25.00%) with count >= 23324 account for 99.9% of the total counts.
+; SUMMARY-NEXT: 11 blocks (68.75%) with count >= 24 account for 99.99% of the total counts.
+; SUMMARY-NEXT: 16 blocks (100.00%) with count >= 10 account for 99.999% of the total counts.
+; SUMMARY-NEXT: 16 blocks (100.00%) with count >= 10 account for 99.9999% of the total counts.
 
 
 ; SUMMARY-NEST:      Total functions: 4
 ; SUMMARY-NEST-NEXT: Maximum function count: 32
-; SUMMARY-NEST-NEXT: Maximum block count: 362830
+; SUMMARY-NEST-NEXT: Maximum internal block count: 0
 ; SUMMARY-NEST-NEXT: Total number of blocks: 15
 ; SUMMARY-NEST-NEXT: Total count: 772504
 ; SUMMARY-NEST-NEXT: Detailed summary:
-; SUMMARY-NEST-NEXT: 1 blocks with count >= 362830 account for 1 percentage of the total counts.
-; SUMMARY-NEST-NEXT: 1 blocks with count >= 362830 account for 10 percentage of the total counts.
-; SUMMARY-NEST-NEXT: 1 blocks with count >= 362830 account for 20 percentage of the total counts.
-; SUMMARY-NEST-NEXT: 1 blocks with count >= 362830 account for 30 percentage of the total counts.
-; SUMMARY-NEST-NEXT: 1 blocks with count >= 362830 account for 40 percentage of the total counts.
-; SUMMARY-NEST-NEXT: 2 blocks with count >= 362805 account for 50 percentage of the total counts.
-; SUMMARY-NEST-NEXT: 2 blocks with count >= 362805 account for 60 percentage of the total counts.
-; SUMMARY-NEST-NEXT: 2 blocks with count >= 362805 account for 70 percentage of the total counts.
-; SUMMARY-NEST-NEXT: 2 blocks with count >= 362805 account for 80 percentage of the total counts.
-; SUMMARY-NEST-NEXT: 2 blocks with count >= 362805 account for 90 percentage of the total counts.
-; SUMMARY-NEST-NEXT: 3 blocks with count >= 23327 account for 95 percentage of the total counts.
-; SUMMARY-NEST-NEXT: 4 blocks with count >= 23324 account for 99 percentage of the total counts.
-; SUMMARY-NEST-NEXT: 4 blocks with count >= 23324 account for 99.9 percentage of the total counts.
-; SUMMARY-NEST-NEXT: 10 blocks with count >= 21 account for 99.99 percentage of the total counts.
-; SUMMARY-NEST-NEXT: 15 blocks with count >= 10 account for 99.999 percentage of the total counts.
-; SUMMARY-NEST-NEXT: 15 blocks with count >= 10 account for 99.9999 percentage of the total counts.
+; SUMMARY-NEST-NEXT: 1 blocks (6.67%) with count >= 362830 account for 1% of the total counts.
+; SUMMARY-NEST-NEXT: 1 blocks (6.67%) with count >= 362830 account for 10% of the total counts.
+; SUMMARY-NEST-NEXT: 1 blocks (6.67%) with count >= 362830 account for 20% of the total counts.
+; SUMMARY-NEST-NEXT: 1 blocks (6.67%) with count >= 362830 account for 30% of the total counts.
+; SUMMARY-NEST-NEXT: 1 blocks (6.67%) with count >= 362830 account for 40% of the total counts.
+; SUMMARY-NEST-NEXT: 2 blocks (13.33%) with count >= 362805 account for 50% of the total counts.
+; SUMMARY-NEST-NEXT: 2 blocks (13.33%) with count >= 362805 account for 60% of the total counts.
+; SUMMARY-NEST-NEXT: 2 blocks (13.33%) with count >= 362805 account for 70% of the total counts.
+; SUMMARY-NEST-NEXT: 2 blocks (13.33%) with count >= 362805 account for 80% of the total counts.
+; SUMMARY-NEST-NEXT: 2 blocks (13.33%) with count >= 362805 account for 90% of the total counts.
+; SUMMARY-NEST-NEXT: 3 blocks (20.00%) with count >= 23327 account for 95% of the total counts.
+; SUMMARY-NEST-NEXT: 4 blocks (26.67%) with count >= 23324 account for 99% of the total counts.
+; SUMMARY-NEST-NEXT: 4 blocks (26.67%) with count >= 23324 account for 99.9% of the total counts.
+; SUMMARY-NEST-NEXT: 10 blocks (66.67%) with count >= 21 account for 99.99% of the total counts.
+; SUMMARY-NEST-NEXT: 15 blocks (100.00%) with count >= 10 account for 99.999% of the total counts.
+; SUMMARY-NEST-NEXT: 15 blocks (100.00%) with count >= 10 account for 99.9999% of the total counts.
diff --git a/llvm/test/tools/llvm-profdata/general.proftext b/llvm/test/tools/llvm-profdata/general.proftext
index 2dfb8e8b34d57e..89762f2540f6a6 100644
--- a/llvm/test/tools/llvm-profdata/general.proftext
+++ b/llvm/test/tools/llvm-profdata/general.proftext
@@ -71,18 +71,18 @@ hex_hash
 # DETAILED-SUMMARY: Total number of blocks: 10
 # DETAILED-SUMMARY: Total count: 4539628424389557499
 # DETAILED-SUMMARY: Detailed summary:
-# DETAILED-SUMMARY: 3 blocks with count >= 576460752303423488 account for 80 percentage of the total counts.
-# DETAILED-SUMMARY: 4 blocks with count >= 288230376151711744 account for 90 percentage of the total counts.
-# DETAILED-SUMMARY: 4 blocks with count >= 288230376151711744 account for 95 percentage of the total counts.
-# DETAILED-SUMMARY: 6 blocks with count >= 72057594037927936 account for 99 percentage of the total counts.
-# DETAILED-SUMMARY: 6 blocks with count >= 72057594037927936 account for 99.9 percentage of the total counts.
-# DETAILED-SUMMARY: 6 blocks with count >= 72057594037927936 account for 99.99 percentage of the total counts.
-# DETAILED-SUMMARY: 6 blocks with count >= 72057594037927936 account for 99.999 percentage of the total counts.
+# DETAILED-SUMMARY: 3 blocks (30.00%) with count >= 576460752303423488 account for 80% of the total counts.
+# DETAILED-SUMMARY: 4 blocks (40.00%) with count >= 288230376151711744 account for 90% of the total counts.
+# DETAILED-SUMMARY: 4 blocks (40.00%) with count >= 288230376151711744 account for 95% of the total counts.
+# DETAILED-SUMMARY: 6 blocks (60.00%) with count >= 72057594037927936 account for 99% of the total counts.
+# DETAILED-SUMMARY: 6 blocks (60.00%) with count >= 72057594037927936 account for 99.9% of the total counts.
+# DETAILED-SUMMARY: 6 blocks (60.00%) with count >= 72057594037927936 account for 99.99% of the total counts.
+# DETAILED-SUMMARY: 6 blocks (60.00%) with count >= 72057594037927936 account for 99.999% of the total counts.
 
 # RUN: llvm-profdata show --detailed-summary --detailed-summary-cutoffs=600000 %t.profdata | FileCheck %s -check-prefix=DETAILED-SUMMARY-2
-# DETAILED-SUMMARY-2: 2 blocks with count >= 1152921504606846976 account for 60 percentage of the total counts.
+# DETAILED-SUMMARY-2: 2 blocks (28.57%) with count >= 1152921504606846976 account for 60% of the total counts.
 #
 # RUN: llvm-profdata show --detailed-summary --detailed-summary-cutoffs=600000,900000,999999 %t.profdata | FileCheck %s -check-prefix=DETAILED-SUMMARY-3
-# DETAILED-SUMMARY-3: 2 blocks with count >= 1152921504606846976 account for 60 percentage of the total counts.
-# DETAILED-SUMMARY-3: 4 blocks with count >= 288230376151711744 account for 90 percentage of the total counts.
-# DETAILED-SUMMARY-3: 6 blocks with count >= 72057594037927936 account for 99.9999 percentage of the total counts.
+# DETAILED-SUMMARY-3: 2 blocks (28.57%) with count >= 1152921504606846976 account for 60% of the total counts.
+# DETAILED-SUMMARY-3: 4 blocks (57.14%) with count >= 288230376151711744 account for 90% of the total counts.
+# DETAILED-SUMMARY-3: 6 blocks (85.71%) with count >= 72057594037927936 account for 99.9999% of the total counts.
diff --git a/llvm/test/tools/llvm-profdata/sample-summary.test b/llvm/test/tools/llvm-profdata/sample-summary.test
index 3326c9bb29806b..310cc8bfb721cd 100644
--- a/llvm/test/tools/llvm-profdata/sample-summary.test
+++ b/llvm/test/tools/llvm-profdata/sample-summary.test
@@ -2,23 +2,23 @@
 
 ; CHECK: Total functions: 3
 ; CHECK-NEXT: Maximum function count: 1437
-; CHECK-NEXT: Maximum block count: 2080
+; CHECK-NEXT: Maximum internal block count: 0
 ; CHECK-NEXT: Total number of blocks: 11
 ; CHECK-NEXT: Total count: 12943
 ; CHECK-NEXT: Detailed summary:
-; CHECK-NEXT: 1 blocks with count >= 2080 account for 1 percentage of the total counts.
-; CHECK-NEXT: 1 blocks with count >= 2080 account for 10 percentage of the total counts.
-; CHECK-NEXT: 2 blocks with count >= 2064 account for 20 percentage of the total counts.
-; CHECK-NEXT: 2 blocks with count >= 2064 account for 30 percentage of the total counts.
-; CHECK-NEXT: 3 blocks with count >= 2000 account for 40 percentage of the total counts.
-; CHECK-NEXT: 4 blocks with count >= 1437 account for 50 percentage of the total counts.
-; CHECK-NEXT: 6 blocks with count >= 1075 account for 60 percentage of the total counts.
-; CHECK-NEXT: 6 blocks with count >= 1075 account for 70 percentage of the total counts.
-; CHECK-NEXT: 7 blocks with count >= 1000 account for 80 percentage of the total counts.
-; CHECK-NEXT: 11 blocks with count >= 534 account for 90 percentage of the total counts.
-; CHECK-NEXT: 11 blocks with count >= 534 account for 95 percentage of the total counts.
-; CHECK-NEXT: 11 blocks with count >= 534 account for 99 percentage of the total counts.
-; CHECK-NEXT: 11 blocks with count >= 534 account for 99.9 percentage of the total counts.
-; CHECK-NEXT: 11 blocks with count >= 534 account for 99.99 percentage of the total counts.
-; CHECK-NEXT: 11 blocks with count >= 534 account for 99.999 percentage of the total counts.
-; CHECK-NEXT: 11 blocks with count >= 534 account for 99.9999 percentage of the total counts.
+; CHECK-NEXT: 1 blocks (9.09%) with count >= 2080 account for 1% of the total counts.
+; CHECK-NEXT: 1 blocks (9.09%) with count >= 2080 account for 10% of the total counts.
+; CHECK-NEXT: 2 blocks (18.18%) with count >= 2064 account for 20% of the total counts.
+; CHECK-NEXT: 2 blocks (18.18%) with count >= 2064 account for 30% of the total counts.
+; CHECK-NEXT: 3 blocks (27.27%) with count >= 2000 account for 40% of the total counts.
+; CHECK-NEXT: 4 blocks (36.36%) with count >= 1437 account for 50% of the total counts.
+; CHECK-NEXT: 6 blocks (54.55%) with count >= 1075 account for 60% of the total counts.
+; CHECK-NEXT: 6 blocks (54.55%) with count >= 1075 account for 70% of the total counts.
+; CHECK-NEXT: 7 blocks (63.64%) with count >= 1000 account for 80% of the total counts.
+; CHECK-NEXT: 11 blocks (100.00%) with count >= 534 account for 90% of the total counts.
+; CHECK-NEXT: 11 blocks (100.00%) with count >= 534 account for 95% of the total counts.
+; CHECK-NEXT: 11 blocks (100.00%) with count >= 534 account for 99% of the total counts.
+; CHECK-NEXT: 11 blocks (100.00%) with count >= 534 account for 99.9% of the total counts.
+; CHECK-NEXT: 11 blocks (100.00%) with count >= 534 account for 99.99% of the total counts.
+; CHECK-NEXT: 11 blocks (100.00%) with count >= 534 account for 99.999% of the total counts.
+; CHECK-NEXT: 11 blocks (100.00%) with count >= 534 account for 99.9999% of the total counts.
diff --git a/llvm/test/tools/llvm-profdata/suppl-instr-with-sample.test b/llvm/test/tools/llvm-profdata/suppl-instr-with-sample.test
index 20d4d2198ff449..c22646e45849ea 100644
--- a/llvm/test/tools/llvm-profdata/suppl-instr-with-sample.test
+++ b/llvm/test/tools/llvm-profdata/suppl-instr-with-sample.test
@@ -98,19 +98,19 @@ MIX5-NEXT: Maximum internal block count: 2000
 MIX5-NEXT: Total number of blocks: 9
 MIX5-NEXT: Total count: 6525
 MIX5-NEXT: Detailed summary:
-MIX5-NEXT: 1 blocks with count >= 3000 account for 1 percentage of the total counts.
-MIX5-NEXT: 1 blocks with count >= 3000 account for 10 percentage of the total counts.
-MIX5-NEXT: 1 blocks with count >= 3000 account for 20 percentage of the total counts.
-MIX5-NEXT: 1 blocks with count >= 3000 account for 30 percentage of the total counts.
-MIX5-NEXT: 1 blocks with count >= 3000 account for 40 percentage of the total counts.
-MIX5-NEXT: 2 blocks with count >= 2000 account for 50 percentage of the total counts.
-MIX5-NEXT: 2 blocks with count >= 2000 account for 60 percentage of the total counts.
-MIX5-NEXT: 2 blocks with count >= 2000 account for 70 percentage of the total counts.
-MIX5-NEXT: 3 blocks with count >= 1000 account for...
[truncated]

ellishg · 2024-08-24T03:37:43Z

llvm/lib/IR/ProfileSummary.cpp

@@ -251,17 +251,18 @@ ProfileSummary *ProfileSummary::getFromMD(Metadata *MD) {
 void ProfileSummary::printSummary(raw_ostream &OS) const {
  OS << "Total functions: " << NumFunctions << "\n";
  OS << "Maximum function count: " << MaxFunctionCount << "\n";
-  OS << "Maximum block count: " << MaxCount << "\n";
+  OS << "Maximum internal block count: " << MaxInternalCount << "\n";


I also changed the behavior here to dump MaxInternalCount instead of MaxCount because I think it requires fewer test changes. But I'm also open to changing this or including both logs.

What is "internal block" and what differentiate it from just "block"?

I believe an "internal block" is a non-entry blocks while "block" includes the entry block, which is is reported in "function count".

ellishg · 2024-10-02T23:55:51Z

Friendly ping

In the detailed summary, we want to report the fraction of total blocks that account for each percentile cutoff. Also, use `ProfileSummary::printSummary()` in Instr profiles for consistancy.

ellishg · 2025-01-23T00:41:40Z

I've just rebased to main. Does anyone think this patch will cause any issues? I believe it will mostly impact reporting via -detailed-summary.

wlei-llvm

Thanks for the improvement, LGTM.

XFAIL `llvm/test/tools/llvm-profdata/general.proftext` after it was accidentally broken by #105915. I will follow up to get this fixed.

XFAIL `llvm/test/tools/llvm-profdata/general.proftext` after it was accidentally broken by llvm/llvm-project#105915. I will follow up to get this fixed.

vitalybuka · 2025-01-23T19:14:07Z

Looks like from this patch https://lab.llvm.org/buildbot/#/builders/107/builds/7329

CC @thurstond

ellishg · 2025-01-23T20:27:17Z

Looks like from this patch https://lab.llvm.org/buildbot/#/builders/107/builds/7329

CC @thurstond

Yeah, sorry about that. I XFAILed the test in #124165 and properly fixed in #124169.

vitalybuka · 2025-01-23T20:36:21Z

Looks like from this patch https://lab.llvm.org/buildbot/#/builders/107/builds/7329
CC @thurstond

Yeah, sorry about that. I XFAILed the test in #124165 and properly fixed in #124169.

Please revert the original patch if fix needs more reviews.

The detailed summary format was changed in #105915 which broke `llvm/test/tools/llvm-profdata/general.proftext` (XFAILed in #124165). Apparently the behavior of `%lu` is different between Linux and Windows, so I reverted back to using `<<` style formats.

…24169) The detailed summary format was changed in llvm/llvm-project#105915 which broke `llvm/test/tools/llvm-profdata/general.proftext` (XFAILed in llvm/llvm-project#124165). Apparently the behavior of `%lu` is different between Linux and Windows, so I reverted back to using `<<` style formats.

vitalybuka · 2025-01-24T08:08:04Z

It's still broken https://lab.llvm.org/buildbot/#/builders/107/builds/7357

antmox · 2025-01-24T17:46:21Z

Also broke clang-arm64-windows-msvc : https://lab.llvm.org/buildbot/#/builders/161/builds/4261

ellishg · 2025-01-24T17:55:26Z

Thanks @vitalybuka and @antmox. Should be fixed by #124318.

Fix a Windows compiler-rt test that #105915 broke.

…e (#124318) Fix a Windows compiler-rt test that llvm/llvm-project#105915 broke.

ellishg requested review from ZequanWu, htyu and WenleiHe August 24, 2024 02:03

llvmbot added compiler-rt PGO Profile Guided Optimizations llvm:ir labels Aug 24, 2024

ellishg commented Aug 24, 2024

View reviewed changes

[llvm-profdata] Add block percent to detailed summary

362b2cf

In the detailed summary, we want to report the fraction of total blocks that account for each percentile cutoff. Also, use `ProfileSummary::printSummary()` in Instr profiles for consistancy.

ellishg force-pushed the profdata-show branch from afef6d1 to 362b2cf Compare January 23, 2025 00:39

wlei-llvm approved these changes Jan 23, 2025

View reviewed changes

ellishg merged commit a245309 into llvm:main Jan 23, 2025
6 of 8 checks passed

ellishg deleted the profdata-show branch January 23, 2025 17:30

ellishg mentioned this pull request Jan 23, 2025

[llvm-profdata] XFAIL broken test on windows #124165

Merged

ellishg added a commit that referenced this pull request Jan 23, 2025

[llvm-profdata] XFAIL broken test on windows (#124165)

0290693

XFAIL `llvm/test/tools/llvm-profdata/general.proftext` after it was accidentally broken by #105915. I will follow up to get this fixed.

ellishg mentioned this pull request Jan 23, 2025

[llvm-profdata] Fix detailed summary format on Windows #124169

Merged

ellishg mentioned this pull request Jan 24, 2025

[compiler-rt] Fix Windows test after profile summary change #124318

Merged

ellishg added a commit that referenced this pull request Jan 24, 2025

[compiler-rt] Fix Windows test after profile summary change (#124318)

12f82fb

Fix a Windows compiler-rt test that #105915 broke.

github-actions bot pushed a commit to arm/arm-toolchain that referenced this pull request Jan 24, 2025

Automerge: [compiler-rt] Fix Windows test after profile summary chang…

9ca122e

…e (#124318) Fix a Windows compiler-rt test that llvm/llvm-project#105915 broke.

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Uh oh!

[llvm-profdata] Add block percent to detailed summary #105915

[llvm-profdata] Add block percent to detailed summary #105915

Uh oh!

ellishg commented Aug 24, 2024

Uh oh!

llvmbot commented Aug 24, 2024 •

edited

Loading

Uh oh!

ellishg Aug 24, 2024

Uh oh!

WenleiHe Aug 26, 2024

Uh oh!

ellishg Aug 26, 2024

Uh oh!

ellishg commented Oct 2, 2024

Uh oh!

ellishg commented Jan 23, 2025

Uh oh!

wlei-llvm left a comment

Uh oh!

Uh oh!

vitalybuka commented Jan 23, 2025

Uh oh!

ellishg commented Jan 23, 2025

Uh oh!

vitalybuka commented Jan 23, 2025

Uh oh!

vitalybuka commented Jan 24, 2025

Uh oh!

antmox commented Jan 24, 2025

Uh oh!

ellishg commented Jan 24, 2025

Uh oh!

Uh oh!

[llvm-profdata] Add block percent to detailed summary #105915

[llvm-profdata] Add block percent to detailed summary #105915

Uh oh!

Conversation

ellishg commented Aug 24, 2024

Uh oh!

llvmbot commented Aug 24, 2024 • edited Loading Uh oh! There was an error while loading. Please reload this page.

Uh oh!

Uh oh!

ellishg Aug 24, 2024

Choose a reason for hiding this comment

Uh oh!

WenleiHe Aug 26, 2024

Choose a reason for hiding this comment

Uh oh!

ellishg Aug 26, 2024

Choose a reason for hiding this comment

Uh oh!

ellishg commented Oct 2, 2024

Uh oh!

ellishg commented Jan 23, 2025

Uh oh!

wlei-llvm left a comment

Choose a reason for hiding this comment

Uh oh!

Uh oh!

vitalybuka commented Jan 23, 2025

Uh oh!

ellishg commented Jan 23, 2025

Uh oh!

vitalybuka commented Jan 23, 2025

Uh oh!

vitalybuka commented Jan 24, 2025

Uh oh!

antmox commented Jan 24, 2025

Uh oh!

ellishg commented Jan 24, 2025

Uh oh!

Uh oh!

llvmbot commented Aug 24, 2024 •

edited

Loading