Xinliang David Li
2014-Apr-18 19:45 UTC
[LLVMdev] multithreaded performance disaster with -fprofile-instr-generate (contention on profile counters)
On Fri, Apr 18, 2014 at 12:13 AM, Dmitry Vyukov <dvyukov at google.com> wrote:> Hi, > > This is long thread, so I will combine several comments into single email. > > > >> - 8-bit per-thread counters, dumping into central counters on overflow. > >The overflow will happen very quickly with 8bit counter. > > Yes, but it reduces contention by 256x (a thread must execute at least 256 > loop iterations between increments). In practice, if you reduce contention > below some threshold, it does not represent a problem anymore. > > > > >> - per-thread counters. Solves the problem at huge cost in RAM per-thread > >It is not practical. Especially for TLS counters -- it creates huge > pressure on stack memory. > > Do we have any numbers about number of counters? If there are 100K 1-byte > counters, I would consider it as practical. > >A medium sized app I looked at has about 10M counters (arcs only). It is also not uncommon to see such apps running with hundreds of threads.> > > > > > In Google GCC, we implemented another technique which proves to be very > effective -- it is called FDO sampling. > > Basically counters will be updated every N samples. > > How does it work? >Similar to how occurrences based PMU sampling work. Setting sampling period to 100 can reduce the instrumentation overhead by close to 100x without introducing much precision loss.> > > > >> It seems to me like we’re going to have a hard time getting good > multithreaded performance without significant impact on the single-threaded > behavior. > > I don't really agree. > > > >We are talking about developers here. Nobody would know the exact thread > counts, but developers know the ballpark number > > I strongly believe that we must relief developers from this choice during > build time, and do our best to auto-tune (if the final scheme requires > tuning). >That really depends. If the tuning space is small, it won't be a problem for the developer/builder.> First, such questions puts unnecessary burden on developers. We don't ask > what register allocation algorithm to use for each function, right? >Crazy developers can actually do that via internal options, but this is totally different case. People just needs one flag to turn on/off sharding. When sharding is on, compiler can pick the best 'N' according to some heuristics at compile time.> Second, there are significant chances we will get a wrong answer, because > e.g. developer's view of how threaded is the app can differ from reality or > from our classification. > Third, the app can be build by a build engineer; or the procedure can be > applied to a base with 10'000 apps; or threaded-ness can change; or the app > can work in different modes; or the app can have different phases. > >We have forgotten to mention the benefit of implementation simplicity. If the static/simple solution solves the problem for most of the use cases, designing fancy dynamic solution sounds like over-engineering to me. It (overhead reduction technique) may also get in the way of easier functional enhancement in the future. David -------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20140418/6861153d/attachment.html>
Kostya Serebryany
2014-Apr-23 14:31 UTC
[LLVMdev] multithreaded performance disaster with -fprofile-instr-generate (contention on profile counters)
I've run one proprietary benchmark that reflects a large portion of the google's server side code. -fprofile-instr-generate leads to 14x slowdown due to counter contention. That's serious. Admittedly, there is a single hot function that accounts for half of that slowdown, but even if I rebuild that function w/o -fprofile-instr-generate, the slowdown remains above 5x. This is not a toy code that I've written to prove my point -- this is real code one may want to profile with -fprofile-instr-generate. We need another approach for threaded code. There is another ungood feature of the current instrumentation. Consider this function: std::vector<int> v(1000); void foo() { v[0] = 42; } Here we have a single basic block and a call, but since the coverage is emitted by the FE before inlining (and is also emitted for std::vector methods) we get this assembler at -O2: 0000000000400b90 <_Z3foov>: 400b90: 48 ff 05 11 25 20 00 incq 0x202511(%rip) # 6030a8 <__llvm_profile_counters__Z3foov> 400b97: 48 ff 05 42 25 20 00 incq 0x202542(%rip) # 6030e0 <__llvm_profile_counters__ZNSt6vectorIiSaIiEEixEm> 400b9e: 48 8b 05 4b 26 20 00 mov 0x20264b(%rip),%rax # 6031f0 <v> 400ba5: c7 00 2a 00 00 00 movl $0x2a,(%rax) 400bab: c3 retq Suddenly, an innocent function that uses std::vector becomes a terrible point of contention. Full test case below, -fprofile-instr-generate leads to 10x slowdown. ======================== Now, here is a more detailed proposal of logarithmic self-cooling counter mentioned before. Please comment. The counter is a number of the form (2^k-1). It starts with 0. After the first update it is 1. After *approximately* 1 more update it becomes 3 After *approximately* 2 more updates it becomes 7 After *approximately* 4 more updates it becomes 15 ... After *approximately* 2^k more updates it becomes 2^(k+2)-1 The code would look like this: if ((fast_thread_local_rand() & counter) == 0) counter = 2 * counter + 1; Possible implementation for fast_thread_local_rand: long fast_thread_local_rand() { static __thread long r; return r++; } Although I would try to find something cheaper that this. (Ideas?) The counter is not precise (well, the current racy counters are not precise either). But statistically it should be no more than 2x away from the real counter. Will this accuracy be enough for the major use cases? Moreover, this approach allows to implement the counter increment using a callback: if ((fast_thread_local_rand() & counter) == 0) __cov_increment(&counter); which in turn will let us use the same hack as in AsanCoverage: use the PC to map the counter to the source code. (= no need to create separate section in the objects). Thoughts? --kcc % clang++ -O2 -lpthread coverage_mt_vec.cc && time ./a.out TIME: real: 0.219; user: 0.430; system: 0.000 % clang++ -O2 -lpthread -fprofile-instr-generate coverage_mt_vec.cc && time ./a.out TIME: real: 3.743; user: 7.280; system: 0.000 % cat coverage_mt_vec.cc #include <pthread.h> #include <vector> std::vector<int> v(1000); __attribute__((noinline)) void foo() { v[0] = 42; } void *Thread1(void *) { for (int i = 0; i < 100000000; i++) foo(); return 0; } __attribute__((noinline)) void bar() { v[999] = 66; } void *Thread2(void *) { for (int i = 0; i < 100000000; i++) bar(); return 0; } int main() { static const int kNumThreads = 16; pthread_t t[kNumThreads]; pthread_create(&t[0], 0, Thread1, 0); pthread_create(&t[1], 0, Thread2, 0); pthread_join(t[0], 0); pthread_join(t[1], 0); return 0; } On Fri, Apr 18, 2014 at 11:45 PM, Xinliang David Li <xinliangli at gmail.com>wrote:> > > > On Fri, Apr 18, 2014 at 12:13 AM, Dmitry Vyukov <dvyukov at google.com>wrote: > >> Hi, >> >> This is long thread, so I will combine several comments into single email. >> >> >> >> - 8-bit per-thread counters, dumping into central counters on overflow. >> >The overflow will happen very quickly with 8bit counter. >> >> Yes, but it reduces contention by 256x (a thread must execute at least >> 256 loop iterations between increments). In practice, if you reduce >> contention below some threshold, it does not represent a problem anymore. >> >> >> >> >> - per-thread counters. Solves the problem at huge cost in RAM >> per-thread >> >It is not practical. Especially for TLS counters -- it creates huge >> pressure on stack memory. >> >> Do we have any numbers about number of counters? If there are 100K 1-byte >> counters, I would consider it as practical. >> >> > A medium sized app I looked at has about 10M counters (arcs only). It is > also not uncommon to see such apps running with hundreds of threads. > > > >> >> >> >> >> > In Google GCC, we implemented another technique which proves to be very >> effective -- it is called FDO sampling. >> > Basically counters will be updated every N samples. >> >> How does it work? >> > > Similar to how occurrences based PMU sampling work. Setting sampling > period to 100 can reduce the instrumentation overhead by close to 100x > without introducing much precision loss. > > >> >> >> >> >> It seems to me like we’re going to have a hard time getting good >> multithreaded performance without significant impact on the single-threaded >> behavior. >> > I don't really agree. >> >> >> >We are talking about developers here. Nobody would know the exact thread >> counts, but developers know the ballpark number >> >> I strongly believe that we must relief developers from this choice during >> build time, and do our best to auto-tune (if the final scheme requires >> tuning). >> > > > That really depends. If the tuning space is small, it won't be a problem > for the developer/builder. > > > >> First, such questions puts unnecessary burden on developers. We don't >> ask what register allocation algorithm to use for each function, right? >> > > Crazy developers can actually do that via internal options, but this is > totally different case. People just needs one flag to turn on/off > sharding. When sharding is on, compiler can pick the best 'N' according to > some heuristics at compile time. > > > >> Second, there are significant chances we will get a wrong answer, because >> e.g. developer's view of how threaded is the app can differ from reality or >> from our classification. >> Third, the app can be build by a build engineer; or the procedure can be >> applied to a base with 10'000 apps; or threaded-ness can change; or the app >> can work in different modes; or the app can have different phases. >> >> > We have forgotten to mention the benefit of implementation simplicity. If > the static/simple solution solves the problem for most of the use cases, > designing fancy dynamic solution sounds like over-engineering to me. It > (overhead reduction technique) may also get in the way of easier functional > enhancement in the future. > > David > > > _______________________________________________ > LLVM Developers mailing list > LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev > >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20140423/2ded3fc2/attachment.html>
Xinliang David Li
2014-Apr-23 16:13 UTC
[LLVMdev] multithreaded performance disaster with -fprofile-instr-generate (contention on profile counters)
This is very neat. The concern is that for long running process, the precision loss can be huge. Run to run counter variance can also be very large (without atomic update).> > Moreover, this approach allows to implement the counter increment using a > callback: > if ((fast_thread_local_rand() & counter) == 0) > __cov_increment(&counter); > which in turn will let us use the same hack as in AsanCoverage: use the PC > to map the counter to the source code. > (= no need to create separate section in the objects). >Similar callback mechanism (GCC based) is used in Google for testing coverage. David> > Thoughts? > > --kcc > > > % clang++ -O2 -lpthread coverage_mt_vec.cc && time ./a.out > TIME: real: 0.219; user: 0.430; system: 0.000 > % clang++ -O2 -lpthread -fprofile-instr-generate coverage_mt_vec.cc && > time ./a.out > TIME: real: 3.743; user: 7.280; system: 0.000 > > % cat coverage_mt_vec.cc > #include <pthread.h> > #include <vector> > > std::vector<int> v(1000); > > __attribute__((noinline)) void foo() { v[0] = 42; } > > void *Thread1(void *) { > for (int i = 0; i < 100000000; i++) > foo(); > return 0; > } > > __attribute__((noinline)) void bar() { v[999] = 66; } > > void *Thread2(void *) { > for (int i = 0; i < 100000000; i++) > bar(); > return 0; > } > > int main() { > static const int kNumThreads = 16; > pthread_t t[kNumThreads]; > pthread_create(&t[0], 0, Thread1, 0); > pthread_create(&t[1], 0, Thread2, 0); > pthread_join(t[0], 0); > pthread_join(t[1], 0); > return 0; > } > > > > > On Fri, Apr 18, 2014 at 11:45 PM, Xinliang David Li <xinliangli at gmail.com>wrote: > >> >> >> >> On Fri, Apr 18, 2014 at 12:13 AM, Dmitry Vyukov <dvyukov at google.com>wrote: >> >>> Hi, >>> >>> This is long thread, so I will combine several comments into single >>> email. >>> >>> >>> >> - 8-bit per-thread counters, dumping into central counters on >>> overflow. >>> >The overflow will happen very quickly with 8bit counter. >>> >>> Yes, but it reduces contention by 256x (a thread must execute at least >>> 256 loop iterations between increments). In practice, if you reduce >>> contention below some threshold, it does not represent a problem anymore. >>> >>> >>> >>> >> - per-thread counters. Solves the problem at huge cost in RAM >>> per-thread >>> >It is not practical. Especially for TLS counters -- it creates huge >>> pressure on stack memory. >>> >>> Do we have any numbers about number of counters? If there are 100K >>> 1-byte counters, I would consider it as practical. >>> >>> >> A medium sized app I looked at has about 10M counters (arcs only). It is >> also not uncommon to see such apps running with hundreds of threads. >> >> >> >>> >>> >>> >>> >>> > In Google GCC, we implemented another technique which proves to be >>> very effective -- it is called FDO sampling. >>> > Basically counters will be updated every N samples. >>> >>> How does it work? >>> >> >> Similar to how occurrences based PMU sampling work. Setting sampling >> period to 100 can reduce the instrumentation overhead by close to 100x >> without introducing much precision loss. >> >> >>> >>> >>> >>> >> It seems to me like we’re going to have a hard time getting good >>> multithreaded performance without significant impact on the single-threaded >>> behavior. >>> > I don't really agree. >>> >>> >>> >We are talking about developers here. Nobody would know the exact >>> thread counts, but developers know the ballpark number >>> >>> I strongly believe that we must relief developers from this choice >>> during build time, and do our best to auto-tune (if the final scheme >>> requires tuning). >>> >> >> >> That really depends. If the tuning space is small, it won't be a >> problem for the developer/builder. >> >> >> >>> First, such questions puts unnecessary burden on developers. We don't >>> ask what register allocation algorithm to use for each function, right? >>> >> >> Crazy developers can actually do that via internal options, but this is >> totally different case. People just needs one flag to turn on/off >> sharding. When sharding is on, compiler can pick the best 'N' according to >> some heuristics at compile time. >> >> >> >>> Second, there are significant chances we will get a wrong answer, >>> because e.g. developer's view of how threaded is the app can differ from >>> reality or from our classification. >>> Third, the app can be build by a build engineer; or the procedure can be >>> applied to a base with 10'000 apps; or threaded-ness can change; or the app >>> can work in different modes; or the app can have different phases. >>> >>> >> We have forgotten to mention the benefit of implementation simplicity. >> If the static/simple solution solves the problem for most of the use >> cases, designing fancy dynamic solution sounds like over-engineering to me. >> It (overhead reduction technique) may also get in the way of easier >> functional enhancement in the future. >> >> David >> >> >> _______________________________________________ >> LLVM Developers mailing list >> LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu >> http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev >> >> >-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20140423/5f2d7493/attachment.html>
Bob Wilson
2014-Apr-23 18:48 UTC
[LLVMdev] multithreaded performance disaster with -fprofile-instr-generate (contention on profile counters)
On Apr 23, 2014, at 7:31 AM, Kostya Serebryany <kcc at google.com> wrote:> I've run one proprietary benchmark that reflects a large portion of the google's server side code. > -fprofile-instr-generate leads to 14x slowdown due to counter contention. That's serious. > Admittedly, there is a single hot function that accounts for half of that slowdown, > but even if I rebuild that function w/o -fprofile-instr-generate, the slowdown remains above 5x. > This is not a toy code that I've written to prove my point -- this is real code one may want to profile with -fprofile-instr-generate. > We need another approach for threaded code. > > There is another ungood feature of the current instrumentation. Consider this function: > std::vector<int> v(1000); > void foo() { v[0] = 42; } > > Here we have a single basic block and a call, but since the coverage is emitted by the > FE before inlining (and is also emitted for std::vector methods) we get this assembler at -O2: > 0000000000400b90 <_Z3foov>: > 400b90: 48 ff 05 11 25 20 00 incq 0x202511(%rip) # 6030a8 <__llvm_profile_counters__Z3foov> > 400b97: 48 ff 05 42 25 20 00 incq 0x202542(%rip) # 6030e0 <__llvm_profile_counters__ZNSt6vectorIiSaIiEEixEm> > 400b9e: 48 8b 05 4b 26 20 00 mov 0x20264b(%rip),%rax # 6031f0 <v> > 400ba5: c7 00 2a 00 00 00 movl $0x2a,(%rax) > 400bab: c3 retq > > Suddenly, an innocent function that uses std::vector becomes a terrible point of contention. > Full test case below, -fprofile-instr-generate leads to 10x slowdown. > > ========================> > Now, here is a more detailed proposal of logarithmic self-cooling counter mentioned before. Please comment. > The counter is a number of the form (2^k-1). > It starts with 0. > After the first update it is 1. > After *approximately* 1 more update it becomes 3 > After *approximately* 2 more updates it becomes 7 > After *approximately* 4 more updates it becomes 15 > ... > After *approximately* 2^k more updates it becomes 2^(k+2)-1 > > The code would look like this: > if ((fast_thread_local_rand() & counter) == 0) > counter = 2 * counter + 1; > > Possible implementation for fast_thread_local_rand: > long fast_thread_local_rand() { > static __thread long r; > return r++; > } > Although I would try to find something cheaper that this. (Ideas?) > > > The counter is not precise (well, the current racy counters are not precise either). > But statistically it should be no more than 2x away from the real counter. > Will this accuracy be enough for the major use cases? > > Moreover, this approach allows to implement the counter increment using a callback: > if ((fast_thread_local_rand() & counter) == 0) __cov_increment(&counter); > which in turn will let us use the same hack as in AsanCoverage: use the PC to map the counter to the source code. > (= no need to create separate section in the objects). > > Thoughts? > > —kccI can see that the behavior of our current instrumentation is going to be a problem for the kinds of applications that you’re looking at. If you can find a way to get the overhead down without losing accuracy and without hurting the performance for applications without significant contention, then we can just adopt that. If you can’t do it without tradeoffs, then we should have a separate option to let those who care switch between different kinds of instrumentation. Using the PC to map to the source code is simply not going to work with -fprofile-instr-generate. The mapping from counter values to the user-visible execution counts is complex and relies on detailed knowledge of the clang ASTs.> > > % clang++ -O2 -lpthread coverage_mt_vec.cc && time ./a.out > TIME: real: 0.219; user: 0.430; system: 0.000 > % clang++ -O2 -lpthread -fprofile-instr-generate coverage_mt_vec.cc && time ./a.out > TIME: real: 3.743; user: 7.280; system: 0.000 > > % cat coverage_mt_vec.cc > #include <pthread.h> > #include <vector> > > std::vector<int> v(1000); > > __attribute__((noinline)) void foo() { v[0] = 42; } > > void *Thread1(void *) { > for (int i = 0; i < 100000000; i++) > foo(); > return 0; > } > > __attribute__((noinline)) void bar() { v[999] = 66; } > > void *Thread2(void *) { > for (int i = 0; i < 100000000; i++) > bar(); > return 0; > } > > int main() { > static const int kNumThreads = 16; > pthread_t t[kNumThreads]; > pthread_create(&t[0], 0, Thread1, 0); > pthread_create(&t[1], 0, Thread2, 0); > pthread_join(t[0], 0); > pthread_join(t[1], 0); > return 0; > } > > > > > On Fri, Apr 18, 2014 at 11:45 PM, Xinliang David Li <xinliangli at gmail.com> wrote: > > > > On Fri, Apr 18, 2014 at 12:13 AM, Dmitry Vyukov <dvyukov at google.com> wrote: > Hi, > > This is long thread, so I will combine several comments into single email. > > > >> - 8-bit per-thread counters, dumping into central counters on overflow. > >The overflow will happen very quickly with 8bit counter. > > Yes, but it reduces contention by 256x (a thread must execute at least 256 loop iterations between increments). In practice, if you reduce contention below some threshold, it does not represent a problem anymore. > > > > >> - per-thread counters. Solves the problem at huge cost in RAM per-thread > >It is not practical. Especially for TLS counters -- it creates huge pressure on stack memory. > > Do we have any numbers about number of counters? If there are 100K 1-byte counters, I would consider it as practical. > > > A medium sized app I looked at has about 10M counters (arcs only). It is also not uncommon to see such apps running with hundreds of threads. > > > > > > > > In Google GCC, we implemented another technique which proves to be very effective -- it is called FDO sampling. > > Basically counters will be updated every N samples. > > How does it work? > > Similar to how occurrences based PMU sampling work. Setting sampling period to 100 can reduce the instrumentation overhead by close to 100x without introducing much precision loss. > > > > > >> It seems to me like we’re going to have a hard time getting good multithreaded performance without significant impact on the single-threaded behavior. > > I don't really agree. > > > >We are talking about developers here. Nobody would know the exact thread counts, but developers know the ballpark number > > I strongly believe that we must relief developers from this choice during build time, and do our best to auto-tune (if the final scheme requires tuning). > > > That really depends. If the tuning space is small, it won't be a problem for the developer/builder. > > > First, such questions puts unnecessary burden on developers. We don't ask what register allocation algorithm to use for each function, right? > > Crazy developers can actually do that via internal options, but this is totally different case. People just needs one flag to turn on/off sharding. When sharding is on, compiler can pick the best 'N' according to some heuristics at compile time. > > > Second, there are significant chances we will get a wrong answer, because e.g. developer's view of how threaded is the app can differ from reality or from our classification. > Third, the app can be build by a build engineer; or the procedure can be applied to a base with 10'000 apps; or threaded-ness can change; or the app can work in different modes; or the app can have different phases. > > > We have forgotten to mention the benefit of implementation simplicity. If the static/simple solution solves the problem for most of the use cases, designing fancy dynamic solution sounds like over-engineering to me. It (overhead reduction technique) may also get in the way of easier functional enhancement in the future. > > David > > > _______________________________________________ > LLVM Developers mailing list > LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev > > > _______________________________________________ > LLVM Developers mailing list > LLVMdev at cs.uiuc.edu http://llvm.cs.uiuc.edu > http://lists.cs.uiuc.edu/mailman/listinfo/llvmdev-------------- next part -------------- An HTML attachment was scrubbed... URL: <http://lists.llvm.org/pipermail/llvm-dev/attachments/20140423/77ed1cf7/attachment.html>
Duncan P. N. Exon Smith
2014-Apr-25 17:11 UTC
[LLVMdev] multithreaded performance disaster with -fprofile-instr-generate (contention on profile counters)
On 2014-Apr-23, at 7:31, Kostya Serebryany <kcc at google.com> wrote:> I've run one proprietary benchmark that reflects a large portion of the google's server side code. > -fprofile-instr-generate leads to 14x slowdown due to counter contention. That's serious. > Admittedly, there is a single hot function that accounts for half of that slowdown, > but even if I rebuild that function w/o -fprofile-instr-generate, the slowdown remains above 5x. > This is not a toy code that I've written to prove my point -- this is real code one may want to profile with -fprofile-instr-generate. > We need another approach for threaded code. > > There is another ungood feature of the current instrumentation. Consider this function: > std::vector<int> v(1000); > void foo() { v[0] = 42; } > > Here we have a single basic block and a call, but since the coverage is emitted by the > FE before inlining (and is also emitted for std::vector methods) we get this assembler at -O2: > 0000000000400b90 <_Z3foov>: > 400b90: 48 ff 05 11 25 20 00 incq 0x202511(%rip) # 6030a8 <__llvm_profile_counters__Z3foov> > 400b97: 48 ff 05 42 25 20 00 incq 0x202542(%rip) # 6030e0 <__llvm_profile_counters__ZNSt6vectorIiSaIiEEixEm> > 400b9e: 48 8b 05 4b 26 20 00 mov 0x20264b(%rip),%rax # 6031f0 <v> > 400ba5: c7 00 2a 00 00 00 movl $0x2a,(%rax) > 400bab: c3 retq > > Suddenly, an innocent function that uses std::vector becomes a terrible point of contention.I know you're just using std::vector<> as an example, but I think there should be an option to avoid instrumenting the STL. STL implementations have typically been hand-tuned already. I think the benefit of PGO is extremely small there (although I have no numbers to back this up.)> Full test case below, -fprofile-instr-generate leads to 10x slowdown. > > ========================> > Now, here is a more detailed proposal of logarithmic self-cooling counter mentioned before. Please comment. > The counter is a number of the form (2^k-1). > It starts with 0. > After the first update it is 1. > After *approximately* 1 more update it becomes 3 > After *approximately* 2 more updates it becomes 7 > After *approximately* 4 more updates it becomes 15 > ... > After *approximately* 2^k more updates it becomes 2^(k+2)-1 > > The code would look like this: > if ((fast_thread_local_rand() & counter) == 0) > counter = 2 * counter + 1; > > Possible implementation for fast_thread_local_rand: > long fast_thread_local_rand() { > static __thread long r; > return r++; > } > Although I would try to find something cheaper that this. (Ideas?)Very cool.> The counter is not precise (well, the current racy counters are not precise either). > But statistically it should be no more than 2x away from the real counter. > Will this accuracy be enough for the major use cases?I really like this idea. I think the loss of accuracy should be opt-in (-fprofile-instr-generate=fuzzy), but I think this should be available. I think this could be cleanly integrated into -fprofile-instr-generate without changing -fprofile-instr-use, compiler-rt, or llvm-profdata. Having this option looks like a clear win to me.> Moreover, this approach allows to implement the counter increment using a callback: > if ((fast_thread_local_rand() & counter) == 0) __cov_increment(&counter); > which in turn will let us use the same hack as in AsanCoverage: use the PC to map the counter to the source code. > (= no need to create separate section in the objects).As Bob said already, the PC hack doesn't work for the frontend-based instrumentation. I'm skeptical even about relying on debug info for retrieving function names. But optimizing the binary size is a separate topic anyway.
Seemingly Similar Threads
- [LLVMdev] multithreaded performance disaster with -fprofile-instr-generate (contention on profile counters)
- [LLVMdev] multithreaded performance disaster with -fprofile-instr-generate (contention on profile counters)
- [LLVMdev] multithreaded performance disaster with -fprofile-instr-generate (contention on profile counters)
- [LLVMdev] multithreaded performance disaster with -fprofile-instr-generate (contention on profile counters)
- [LLVMdev] multithreaded performance disaster with -fprofile-instr-generate (contention on profile counters)