[MCA] Add -summary-view option

[lebedev.ri]

Summary

While that is indeed a quite interesting summary stat,
there are cases where it does not really add anything
other than consuming extra lines.

Declutters the output of D48190.

Should i add some standalone tests for it?

[andreadb]

Not sure what other reviewers think about this.
However, my opinion is that the summary view should never be optional in llvm-mca.
It is only a few lines, and it gives a nice overview of the run.

[courbet]

In D48209#1133429, @andreadb wrote:

Not sure what other reviewers think about this.
However, my opinion is that the summary view should never be optional in llvm-mca.
It is only a few lines, and it gives a nice overview of the run.

I mostly agree, but it does not hurt to have the option to disable it if you don;t want it to appear. Given that the patch is very small, I would be in favor of submitting.

[lebedev.ri]

Rebased, NFC.

[andreadb]

In D48209#1133445, @courbet wrote:

In D48209#1133429, @andreadb wrote:

Not sure what other reviewers think about this.
However, my opinion is that the summary view should never be optional in llvm-mca.
It is only a few lines, and it gives a nice overview of the run.

I mostly agree, but it does not hurt to have the option to disable it if you don;t want it to appear. Given that the patch is very small, I would be in favor of submitting.

Out of curiosity: what's the motivating use case for this change? Is it just for testing purposes (i.e. D48190)?

[lebedev.ri]

In D48209#1133459, @andreadb wrote:

In D48209#1133445, @courbet wrote:

In D48209#1133429, @andreadb wrote:

Not sure what other reviewers think about this.
However, my opinion is that the summary view should never be optional in llvm-mca.
It is only a few lines, and it gives a nice overview of the run.

I mostly agree, but it does not hurt to have the option to disable it if you don;t want it to appear. Given that the patch is very small, I would be in favor of submitting.

Out of curiosity: what's the motivating use case for this change? Is it just for testing purposes (i.e. D48190)?

I personally have no need to hide that view in any *normal* case,
so at this moment - yes, just for testing purposes.
It just feels a bit inconsistent for no reason, everything else is toggleable :)

[gbedwell]

test/tools/llvm-mca/X86/option-all-views-1.s

4 ↗

(On Diff #151488)

Seems like there is no need for the ALL prefix now as there is no common output to all of them. This highlights a few issues in the update script that I'll look at fixing. I would've preferred if it had previously warned that the NOREPORT prefix was redundant and that now both the NOREPORT and ALL prefixes are redundant. Lit does spot this:

error: no check strings found with prefixes 'ALL:', 'NOREPORT:'

so I'll make sure the update script does too.

To get the test to pass, we'll need to add "-allow-empty" to that FileCheck command line, and then to make it useful, manually add:

# NOREPORT-NOT: {{.}}

to the test file. The update script will leave any prefixes where it detects a -NOT alone on the assumption that you're checking for something it doesn't understand.

[lebedev.ri]

Actually make ninja check-llvm-tools-llvm-mca pass, as suggested/noted by @gbedwell

[gbedwell]

In D48209#1133474, @lebedev.ri wrote:

Actually make ninja check-llvm-tools-llvm-mca pass, as suggested/noted by @gbedwell

Thanks. Please could you also remove the ALL prefix. I'm just testing a local patch to update_mca_test_checks.py that would make this a hard error:

$ py update_mca_test_checks.py ../test/tools/llvm-mca/X86/option-all-views-1.s
Test: ../test/tools/llvm-mca/X86/option-all-views-1.s
error: unused prefixes: ['ALL']

[lebedev.ri]

Drop -check-prefix=ALL from test/tools/llvm-mca/X86/option-all-views-1.s

[andreadb]

Given that Clement (and presumably Greg) are okay with this change, then I won't oppose it. Sorry for being pedantic.

See my comment below (as well as my comment from D48190)

Thanks
-Andrea

tools/llvm-mca/llvm-mca.cpp

112–115 ↗

(On Diff #151488)

Please mark this this option as "hidden". I don't think it should appear in the default -help output.

[lebedev.ri]

Mark -summary-view as hidden.

[gbedwell]

In D48209#1133500, @andreadb wrote:

Given that Clement (and presumably Greg) are okay with this change, then I won't oppose it. Sorry for being pedantic.

I have no really strong opinion either way, but as it doesn't seem to affect the typical workflow I don't see any real reason to oppose. It's obviously odd to have a mode where the tool outputs nothing at all, but it's not like the user can easily run into that situation without explicitly requesting it. There is at least a nice consistency in having full customization control of all of the views. I guess there could be an argument to having something like "warning: no views are enabled" but I'm wary of getting too deep into the realms of bikeshedding over a minor change like this. :)

[andreadb]

LGTM.

[gbedwell]

test/tools/llvm-mca/X86/option-all-views-1.s

118 ↗

(On Diff #151494)

pendantically, please could you just put this check above the others, as next time somebody bulk regenerates the mca tests using update_mca_test_checks, it'll want to make that change.

[lebedev.ri]

test/tools/llvm-mca/X86/option-all-views-1.s

118 ↗

(On Diff #151494)

Hopefully this is what you meant, not right after the RUN lines themselves..

[lebedev.ri]

In D48209#1133533, @andreadb wrote:

LGTM.

Thank you for the review!

llvm/trunk/test/tools/llvm-mca/X86/option-all-stats-1.s

 # NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py
 # RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=btver2 -all-stats       < %s | FileCheck %s -check-prefix=ALL -check-prefix=FULLREPORT
 # RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=btver2 -all-stats=true  < %s | FileCheck %s -check-prefix=ALL -check-prefix=FULLREPORT
 # RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=btver2 -all-stats=false < %s | FileCheck %s -check-prefix=ALL
 # RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=btver2                  < %s | FileCheck %s -check-prefix=ALL
 
 add %eax, %eax
 
-# ALL:             Iterations:        100
-# ALL-NEXT:        Instructions:      100
-# ALL-NEXT:        Total Cycles:      103
-# ALL-NEXT:        Dispatch Width:    2
-# ALL-NEXT:        IPC:               0.97
-# ALL-NEXT:        Block RThroughput: 0.5
+# FULLREPORT:      Iterations:        100
+# FULLREPORT-NEXT: Instructions:      100
+# FULLREPORT-NEXT: Total Cycles:      103
+# FULLREPORT-NEXT: Dispatch Width:    2
+# FULLREPORT-NEXT: IPC:               0.97
+# FULLREPORT-NEXT: Block RThroughput: 0.5
 
 # ALL:             Instruction Info:
 # ALL-NEXT:        [1]: #uOps
 # ALL-NEXT:        [2]: Latency
 # ALL-NEXT:        [3]: RThroughput
 # ALL-NEXT:        [4]: MayLoad
 # ALL-NEXT:        [5]: MayStore
 # ALL-NEXT:        [6]: HasSideEffects

llvm/trunk/test/tools/llvm-mca/X86/option-all-views-1.s

 # NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py
-# RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=btver2 -all-views       < %s | FileCheck %s -check-prefix=ALL -check-prefix=DEFAULTREPORT -check-prefix=FULLREPORT
-# RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=btver2 -all-views=true  < %s | FileCheck %s -check-prefix=ALL -check-prefix=DEFAULTREPORT -check-prefix=FULLREPORT
-# RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=btver2 -all-views=false < %s | FileCheck %s -check-prefix=ALL -check-prefix=NOREPORT
-# RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=btver2                  < %s | FileCheck %s -check-prefix=ALL -check-prefix=DEFAULTREPORT
+# RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=btver2 -all-views       < %s | FileCheck %s -check-prefix=DEFAULTREPORT -check-prefix=FULLREPORT
+# RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=btver2 -all-views=true  < %s | FileCheck %s -check-prefix=DEFAULTREPORT -check-prefix=FULLREPORT
+# RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=btver2 -all-views=false < %s | FileCheck %s -check-prefix=NOREPORT -allow-empty
+# RUN: llvm-mca -mtriple=x86_64-unknown-unknown -mcpu=btver2                  < %s | FileCheck %s -check-prefix=DEFAULTREPORT
 
 add %eax, %eax
 
-# ALL:                Iterations:        100
-# ALL-NEXT:           Instructions:      100
-# ALL-NEXT:           Total Cycles:      103
-# ALL-NEXT:           Dispatch Width:    2
-# ALL-NEXT:           IPC:               0.97
-# ALL-NEXT:           Block RThroughput: 0.5
+# NOREPORT-NOT: {{.}}
+
+# DEFAULTREPORT:      Iterations:        100
+# DEFAULTREPORT-NEXT: Instructions:      100
+# DEFAULTREPORT-NEXT: Total Cycles:      103
+# DEFAULTREPORT-NEXT: Dispatch Width:    2
+# DEFAULTREPORT-NEXT: IPC:               0.97
+# DEFAULTREPORT-NEXT: Block RThroughput: 0.5
 
 # DEFAULTREPORT:      Instruction Info:
 # DEFAULTREPORT-NEXT: [1]: #uOps
 # DEFAULTREPORT-NEXT: [2]: Latency
 # DEFAULTREPORT-NEXT: [3]: RThroughput
 # DEFAULTREPORT-NEXT: [4]: MayLoad
 # DEFAULTREPORT-NEXT: [5]: MayStore
 # DEFAULTREPORT-NEXT: [6]: HasSideEffects

llvm/trunk/test/tools/llvm-mca/X86/register-file-statistics.s

 # NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=btver2 -iterations=1 -all-stats=false -all-views=false -register-file-stats < %s | FileCheck --check-prefixes=ALL,BTVER2 %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=znver1 -iterations=1 -all-stats=false -all-views=false -register-file-stats < %s | FileCheck --check-prefixes=ALL,ZNVER1 %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=sandybridge -iterations=1 -all-stats=false -all-views=false -register-file-stats < %s | FileCheck --check-prefixes=ALL,SNB %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=ivybridge -iterations=1 -all-stats=false -all-views=false -register-file-stats < %s | FileCheck --check-prefixes=ALL,IVB %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=haswell -iterations=1 -all-stats=false -all-views=false -register-file-stats < %s | FileCheck --check-prefixes=ALL,HSW %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=broadwell -iterations=1 -all-stats=false -all-views=false -register-file-stats < %s | FileCheck --check-prefixes=ALL,BDW %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=knl -iterations=1 -all-stats=false -all-views=false -register-file-stats < %s | FileCheck --check-prefixes=ALL,KNL %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=skylake -iterations=1 -all-stats=false -all-views=false -register-file-stats < %s | FileCheck --check-prefixes=ALL,SKX %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=skylake-avx512 -iterations=1 -all-stats=false -all-views=false -register-file-stats < %s | FileCheck --check-prefixes=ALL,SKX-AVX512 %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=slm -iterations=1 -all-stats=false -all-views=false -register-file-stats < %s | FileCheck --check-prefixes=ALL,SLM %s
 
 xor %eax, %ebx
 
-# ALL:             Iterations:        1
-# ALL-NEXT:        Instructions:      1
-# ALL-NEXT:        Total Cycles:      4
-
-# BDW-NEXT:        Dispatch Width:    4
-# BTVER2-NEXT:     Dispatch Width:    2
-# HSW-NEXT:        Dispatch Width:    4
-# IVB-NEXT:        Dispatch Width:    4
-# KNL-NEXT:        Dispatch Width:    4
-# SKX-NEXT:        Dispatch Width:    6
-# SKX-AVX512-NEXT: Dispatch Width:    6
-# SLM-NEXT:        Dispatch Width:    2
-# SNB-NEXT:        Dispatch Width:    4
-# ZNVER1-NEXT:     Dispatch Width:    4
-
-# ALL-NEXT:        IPC:               0.25
-
-# BDW-NEXT:        Block RThroughput: 0.3
-# BTVER2-NEXT:     Block RThroughput: 0.5
-# HSW-NEXT:        Block RThroughput: 0.3
-# IVB-NEXT:        Block RThroughput: 0.3
-# KNL-NEXT:        Block RThroughput: 0.3
-# SKX-NEXT:        Block RThroughput: 0.3
-# SKX-AVX512-NEXT: Block RThroughput: 0.3
-# SLM-NEXT:        Block RThroughput: 0.5
-# SNB-NEXT:        Block RThroughput: 0.3
-# ZNVER1-NEXT:     Block RThroughput: 0.3
-
 # ALL:             Register File statistics:
 # ALL-NEXT:        Total number of mappings created:    2
 # ALL-NEXT:        Max number of mappings used:         2
 
 # BTVER2:          *  Register File #1 -- JFpuPRF:
 # BTVER2-NEXT:        Number of physical registers:     72
 # BTVER2-NEXT:        Total number of mappings created: 0
 # BTVER2-NEXT:        Max number of mappings used:      0
 
 # BTVER2:          *  Register File #2 -- JIntegerPRF:
 # BTVER2-NEXT:        Number of physical registers:     64
 # BTVER2-NEXT:        Total number of mappings created: 2
 # BTVER2-NEXT:        Max number of mappings used:      2

llvm/trunk/test/tools/llvm-mca/X86/scheduler-queue-usage.s

 # NOTE: Assertions have been autogenerated by utils/update_mca_test_checks.py
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=btver2 -iterations=1 -all-stats=false -all-views=false -scheduler-stats < %s | FileCheck --check-prefixes=ALL,BTVER2 %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=znver1 -iterations=1 -all-stats=false -all-views=false -scheduler-stats < %s | FileCheck --check-prefixes=ALL,ZNVER1 %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=sandybridge -iterations=1 -all-stats=false -all-views=false -scheduler-stats < %s | FileCheck --check-prefixes=ALL,SNB %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=ivybridge -iterations=1 -all-stats=false -all-views=false -scheduler-stats < %s | FileCheck --check-prefixes=ALL,IVB %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=haswell -iterations=1 -all-stats=false -all-views=false -scheduler-stats < %s | FileCheck --check-prefixes=ALL,HSW %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=broadwell -iterations=1 -all-stats=false -all-views=false -scheduler-stats < %s | FileCheck --check-prefixes=ALL,BDW %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=knl -iterations=1 -all-stats=false -all-views=false -scheduler-stats < %s | FileCheck --check-prefixes=ALL,KNL %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=skylake -iterations=1 -all-stats=false -all-views=false -scheduler-stats < %s | FileCheck --check-prefixes=ALL,SKX %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=skylake-avx512 -iterations=1 -all-stats=false -all-views=false -scheduler-stats < %s | FileCheck --check-prefixes=ALL,SKX-AVX512 %s
 # RUN: llvm-mca %s -mtriple=x86_64-unknown-unknown -mcpu=slm -iterations=1 -all-stats=false -all-views=false -scheduler-stats < %s | FileCheck --check-prefixes=ALL,SLM %s
 
 xor %eax, %ebx
 
-# ALL:             Iterations:        1
-# ALL-NEXT:        Instructions:      1
-# ALL-NEXT:        Total Cycles:      4
-
-# BDW-NEXT:        Dispatch Width:    4
-# BTVER2-NEXT:     Dispatch Width:    2
-# HSW-NEXT:        Dispatch Width:    4
-# IVB-NEXT:        Dispatch Width:    4
-# KNL-NEXT:        Dispatch Width:    4
-# SKX-NEXT:        Dispatch Width:    6
-# SKX-AVX512-NEXT: Dispatch Width:    6
-# SLM-NEXT:        Dispatch Width:    2
-# SNB-NEXT:        Dispatch Width:    4
-# ZNVER1-NEXT:     Dispatch Width:    4
-
-# ALL-NEXT:        IPC:               0.25
-
-# BDW-NEXT:        Block RThroughput: 0.3
-# BTVER2-NEXT:     Block RThroughput: 0.5
-# HSW-NEXT:        Block RThroughput: 0.3
-# IVB-NEXT:        Block RThroughput: 0.3
-# KNL-NEXT:        Block RThroughput: 0.3
-# SKX-NEXT:        Block RThroughput: 0.3
-# SKX-AVX512-NEXT: Block RThroughput: 0.3
-# SLM-NEXT:        Block RThroughput: 0.5
-# SNB-NEXT:        Block RThroughput: 0.3
-# ZNVER1-NEXT:     Block RThroughput: 0.3
-
 # ALL:             Schedulers - number of cycles where we saw N instructions issued:
 # ALL-NEXT:        [# issued], [# cycles]
 # ALL-NEXT:         0,          3  (75.0%)
 # ALL-NEXT:         1,          1  (25.0%)
 
 # BDW:             Scheduler's queue usage:
 # BDW-NEXT:        BWPortAny,  1/60
 

llvm/trunk/tools/llvm-mca/llvm-mca.cpp

static cl::opt<bool>
     PrintRegisterFileStats("register-file-stats",
                            cl::desc("Print register file statistics"),
                            cl::cat(ViewOptions), cl::init(false));
 
 static cl::opt<bool> PrintDispatchStats("dispatch-stats",
                                         cl::desc("Print dispatch statistics"),
                                         cl::cat(ViewOptions), cl::init(false));
 
+static cl::opt<bool>
+    PrintSummaryView("summary-view", cl::Hidden,
+                     cl::desc("Print summary view (enabled by default)"),
+                     cl::cat(ViewOptions), cl::init(true));
+
 static cl::opt<bool> PrintSchedulerStats("scheduler-stats",
                                          cl::desc("Print scheduler statistics"),
                                          cl::cat(ViewOptions), cl::init(false));
 
 static cl::opt<bool>
     PrintRetireStats("retire-stats",
                      cl::desc("Print retire control unit statistics"),
                      cl::cat(ViewOptions), cl::init(false));

 }
 
 static void processViewOptions() {
   if (!EnableAllViews.getNumOccurrences() &&
       !EnableAllStats.getNumOccurrences())
     return;
 
   if (EnableAllViews.getNumOccurrences()) {
+    processOptionImpl(PrintSummaryView, EnableAllViews);
     processOptionImpl(PrintResourcePressureView, EnableAllViews);
     processOptionImpl(PrintTimelineView, EnableAllViews);
     processOptionImpl(PrintInstructionInfoView, EnableAllViews);
   }
 
   const cl::opt<bool> &Default =
       EnableAllViews.getPosition() < EnableAllStats.getPosition()
           ? EnableAllStats
           : EnableAllViews;
+  processOptionImpl(PrintSummaryView, Default);
   processOptionImpl(PrintRegisterFileStats, Default);
   processOptionImpl(PrintDispatchStats, Default);
   processOptionImpl(PrintSchedulerStats, Default);
   processOptionImpl(PrintRetireStats, Default);
 }
 
 int main(int argc, char **argv) {
   InitLLVM X(argc, argv);
for (const std::unique_ptr<mca::CodeRegion> &Region : Regions) {
     // by registering all of the Stage instances.
     // But for now, it's just this single puppy.
     std::unique_ptr<mca::FetchStage> Fetch =
         llvm::make_unique<mca::FetchStage>(IB, S);
     mca::Backend B(*STI, *MRI, std::move(Fetch), Width, RegisterFileSize,
                    LoadQueueSize, StoreQueueSize, AssumeNoAlias);
     mca::BackendPrinter Printer(B);
 
+    if (PrintSummaryView)
       Printer.addView(llvm::make_unique<mca::SummaryView>(SM, S, Width));
+
     if (PrintInstructionInfoView)
       Printer.addView(
           llvm::make_unique<mca::InstructionInfoView>(*STI, *MCII, S, *IP));
 
     if (PrintDispatchStats)
       Printer.addView(llvm::make_unique<mca::DispatchStatistics>());
 
     if (PrintSchedulerStats)