Use Perfect Forwarding in all functions that use `apply` family of functors #3221

SteveBronder · 2025-07-11T20:42:30Z

Summary

This fixes #3208 by using perfect forwarding for all functions that use our underlying apply family of functions for calling functions on containers and containers and containers. The issue was that the Holder class, when used in the apply functors, did not have enough type information to know which arguments it should take ownership of.

Consider the following function, where all types are passed in via constant reference.

template <typename T1, typename T2, require_any_container_t<T1, T2>* = nullptr>
inline auto gamma_p(const T1& a, const T2& b) {
  return apply_scalar_binary(
      [](const auto& c, const auto& d) { return gamma_p(c, d); }, a, b);
}

Calling this function with an Eigen expression that has a temporary in it would not give apply_scalar_binary and the Holder inside of apply_scalar_binary enough information to know that the Holder class should own any of the input arguments. As an example we can look at a simplified version of the code used in poisson_lccdf.hpp.

auto log_Pi = log(gamma_p(n_val + 1.0, lambda_val)));
double log_sum = sum(log_Pi);

gamma_p uses apply_scalar_binary and log uses apply_scalar_unary. We need to make sure the inputs and results of the gamma_p function do not fall out of scope by the time we go through log and then assign to log_Pi. Before this PR it would be possible for the expression n_val + 1.0 to fall out of scope as well as the result of gamma_p to go out of scope from log after log_Pi is assigned.

To combat this we now use perfect forwarding for all of the functions that use our internal apply family of functors. This should allow the Holder used internally by the apply functors to know which types need to be owned by it to make sure things do not fall out of scope.

Tests

There is no new tests for this. Since it is an isue on gcc I do wonder how we should test this in our CI/CD?

Side Effects

I'd like to think of some test we can write so that, in the future, developers do not accidentally write functions that use the apply family of functors that do not use perfect forwarding.

Release notes

Adds perfect forwarding to all functions that use the apply family of functors.

Checklist

Copyright holder: Steve Bronder

The copyright holder is typically you or your assignee, such as a university or company. By submitting this pull request, the copyright holder is agreeing to the license the submitted work under the following licenses:
- Code: BSD 3-clause (https://opensource.org/licenses/BSD-3-Clause)
- Documentation: CC-BY 4.0 (https://creativecommons.org/licenses/by/4.0/)
the basic tests are passing
- unit tests pass (to run, use: ./runTests.py test/unit)
- header checks pass, (make test-headers)
- dependencies checks pass, (make test-math-dependencies)
- docs build, (make doxygen)
- code passes the built in C++ standards checks (make cpplint)
the code is written in idiomatic C++ and changes are documented in the doxygen
the new changes are tested

WardBrian · 2025-07-14T16:12:46Z

@SteveBronder anything I can do to help this over the line?

SteveBronder · 2025-07-14T19:29:42Z

Right now it is just making sure the tests pass. I think I got it so we should be good!

WardBrian

Gonna leave the proper review to @andrjohns, just two things that stood out

WardBrian · 2025-07-14T16:48:40Z

stan/math/prim/core/complex_base.hpp

@@ -17,6 +17,7 @@ namespace math {
 template <typename ValueType>
 class complex_base {
 public:
+  auto __rep() const { return *this; }


I can't figure out from searching what this would do

Our complex<var> and complex<fvar<T>> impls still call std library functions sometimes. In those functions, at least for gcc, sometimes they request this __rep() function. The __rep() in their code just returns a copy of this like the one here.

But let me double check if this is still needed

Jenkinsfile

… and O2

WardBrian · 2025-07-15T13:50:28Z

Math pipeline looks good, upstream has some (I believe legitimate) failures. All seem to blame trigamma in some way or another


--- Compiling C++ code ---
clang++-6.0 -std=c++17 -D_REENTRANT -Wno-sign-compare -Wno-ignored-attributes      -I stan/lib/stan_math/lib/tbb_2020.3/include    -O0 -I src -I stan/src -I stan/lib/rapidjson_1.1.0/ -I lib/CLI11-1.9.1/ -I stan/lib/stan_math/ -I stan/lib/stan_math/lib/eigen_3.4.0 -I stan/lib/stan_math/lib/boost_1.87.0 -I stan/lib/stan_math/lib/sundials_6.1.1/include -I stan/lib/stan_math/lib/sundials_6.1.1/src/sundials    -DBOOST_DISABLE_ASSERTS          -c -include-pch stan/src/stan/model/model_header.hpp.gch/model_header_4_2.hpp.gch -x c++ -o ../stanc3/test/integration/good/stanc_helper.o ../stanc3/test/integration/good/stanc_helper.hpp
In file included from ../stanc3/test/integration/good/int_overloads.hpp:1:
In file included from stan/src/stan/model/model_header.hpp:6:
In file included from /home/jenkins/workspace/Stan_Stan_downstream_tests/performance-tests-cmdstan/cmdstan/stan/lib/stan_math/stan/math.hpp:19:
In file included from /home/jenkins/workspace/Stan_Stan_downstream_tests/performance-tests-cmdstan/cmdstan/stan/lib/stan_math/stan/math/rev.hpp:11:
In file included from /home/jenkins/workspace/Stan_Stan_downstream_tests/performance-tests-cmdstan/cmdstan/stan/lib/stan_math/stan/math/rev/constraint.hpp:24:
In file included from /home/jenkins/workspace/Stan_Stan_downstream_tests/performance-tests-cmdstan/cmdstan/stan/lib/stan_math/stan/math/prim/constraint.hpp:4:
In file included from /home/jenkins/workspace/Stan_Stan_downstream_tests/performance-tests-cmdstan/cmdstan/stan/lib/stan_math/stan/math/prim/fun.hpp:335:
/home/jenkins/workspace/Stan_Stan_downstream_tests/performance-tests-cmdstan/cmdstan/stan/lib/stan_math/stan/math/prim/fun/trigamma.hpp:142:12: error: function 'trigamma<stan::math::var_value<double, void> &, nullptr>' with deduced return type cannot be used before it is defined
    return trigamma(std::forward<T>(x));
           ^
/home/jenkins/workspace/Stan_Stan_downstream_tests/performance-tests-cmdstan/cmdstan/stan/lib/stan_math/stan/math/rev/functor/apply_scalar_unary.hpp:34:15: note: in instantiation of function template specialization 'stan::math::trigamma_fun::fun<stan::math::var_value<double, void> &>' requested here
    return F::fun(std::forward<T2>(x));
              ^
/home/jenkins/workspace/Stan_Stan_downstream_tests/performance-tests-cmdstan/cmdstan/stan/lib/stan_math/stan/math/prim/fun/trigamma.hpp:159:47: note: in instantiation of function template specialization 'stan::math::apply_scalar_unary<stan::math::trigamma_fun, stan::math::var_value<double, void> &, void>::apply<stan::math::var_value<double, void> &>' requested here
  return apply_scalar_unary<trigamma_fun, T>::apply(std::forward<T>(x));
                                              ^
../stanc3/test/integration/good/int_overloads.hpp:521:44: note: in instantiation of function template specialization 'stan::math::trigamma<stan::math::var_value<double, void> &, nullptr>' requested here
      transformed_param_real = stan::math::trigamma(p_real);
                                           ^
../stanc3/test/integration/good/int_overloads.hpp:811:12: note: in instantiation of function template specialization 'int_overloads_model_namespace::int_overloads_model::log_prob_impl<false, false, Eigen::Matrix<stan::math::var_value<double, void>, -1, 1, 0, -1, 1>, Eigen::Matrix<int, -1, 1, 0, -1, 1>, nullptr, nullptr, nullptr>' requested here
    return log_prob_impl<propto__, jacobian__>(params_r, params_i, pstream);
           ^
/home/jenkins/workspace/Stan_Stan_downstream_tests/performance-tests-cmdstan/cmdstan/stan/src/stan/model/model_base_crtp.hpp:98:50: note: in instantiation of function template specialization 'int_overloads_model_namespace::int_overloads_model::log_prob<false, false, stan::math::var_value<double, void> >' requested here
    return static_cast<const M*>(this)->template log_prob<false, false>(theta,
                                                 ^
../stanc3/test/integration/good/int_overloads.hpp:129:3: note: in instantiation of member function 'stan::model::model_base_crtp<int_overloads_model_namespace::int_overloads_model>::log_prob' requested here
  ~int_overloads_model() {}
  ^
/home/jenkins/workspace/Stan_Stan_downstream_tests/performance-tests-cmdstan/cmdstan/stan/lib/stan_math/stan/math/prim/fun/trigamma.hpp:158:13: note: 'trigamma<stan::math::var_value<double, void> &, nullptr>' declared here
inline auto trigamma(T&& x) {
            ^
1 error generated.
make: [make/program:77: ../stanc3/test/integration/good/int_overloads.o] Error 1 (ignored)

SteveBronder · 2025-07-15T15:18:59Z

Yes I missed one requires in trigamma

andrjohns · 2025-07-18T00:19:30Z

@andrjohns are you still up for reviewing?

Yep still all good! @SteveBronder should I do a first pass or do you have more changes?

SteveBronder · 2025-07-18T17:50:14Z

I have just a little fix I need to make for the constraints but am afk ATM. Everything else is ready for review!

andrjohns · 2025-07-21T11:57:41Z

stan/math/prim/constraint/positive_constrain.hpp

+inline auto positive_constrain(T&& x, S& lp) {
+  auto&& x_ref = to_ref(std::forward<T>(x));
+  lp += sum(x_ref);
+  return exp(std::forward<decltype(x_ref)>(x));


Suggested change

return exp(std::forward<decltype(x_ref)>(x));

return exp(std::forward<decltype(x_ref)>(x_ref));

andrjohns

A bunch of minor requests/queries, the main thing I'm not too sure about is the new specification for the apply_scalar_unary functions (with the is_arithmetic_v<> conditional) - what's necessitating that?

andrjohns · 2025-07-21T12:08:34Z

stan/math/prim/fun/abs.hpp

+    if constexpr (std::is_arithmetic_v<std::decay_t<T>>) {
+      return std::abs(x);
+    } else {


Is this necessary or just a testing thing? The earlier scalar overload should capture the arithmetic case right?

Yes I think I can remove these

andrjohns · 2025-07-21T12:14:51Z

stan/math/prim/fun/fabs.hpp

  return apply_vector_unary<Container>::apply(
-      x, [](const auto& v) { return v.array().abs(); });
+      std::forward<Container>(x), [](auto&& v) { return v.array().abs(); });


The PR is a bit inconsistent on whether the apply_vector_unary functor should be (auto&& v) or (const auto& v), can you standardise on one?

Just fixed all these up

andrjohns · 2025-07-21T12:17:37Z

stan/math/prim/fun/inc_beta.hpp

-                              a, b, c);
+inline auto inc_beta(T1&& a, T2&& b, T3&& c) {
+  return apply_scalar_ternary(
+      [](auto&& d, auto&& e, auto&& f) { return inc_beta(d, e, f); },


Should the functor also be forwarding the d, e, f arguments to inc_beta as well?

Yes good catch

andrjohns · 2025-07-21T12:19:37Z

stan/math/prim/fun/inv_inc_beta.hpp

-        return inv_inc_beta(d, e, f);
-      },
-      a, b, c);
+      [](auto&& d, auto&& e, auto&& f) { return inv_inc_beta(d, e, f); },


Same query around forwarding the inner arguments here

andrjohns · 2025-07-21T12:23:09Z

stan/math/prim/fun/inv_logit.hpp

-    if (a < LOG_EPSILON) {
-      return exp_a;
-    }


This shouldn't be dropped imo, change is unrelated to the PR (and saves some operations)

andrjohns · 2025-07-21T12:55:28Z

stan/math/prim/functor/apply_vector_unary.hpp

@@ -191,7 +205,7 @@ struct apply_vector_unary<
        = plain_type_t<decltype(apply_vector_unary<T_vt>::apply(x[0], f))>;
    std::vector<T_return> result(x.size());
    std::transform(x.begin(), x.end(), result.begin(), [&f](auto&& xx) {
-      return apply_vector_unary<T_vt>::apply_no_holder(xx, f);
+      return apply_vector_unary<T_vt>::apply(xx, f);


Suggested change

return apply_vector_unary<T_vt>::apply(xx, f);

return apply_vector_unary<T_vt>::apply(std::forward<decltype(xx)>(xx), f);

std transform does not know whether x is a temporary so it will never do forwarding here.

andrjohns · 2025-07-21T12:58:57Z

stan/math/prim/functor/apply_vector_unary.hpp

-  static inline auto apply_no_holder(const T& x, const F& f) {
-    return f(x).matrix().derived();
+  static inline auto apply_no_holder(T2& x, F&& f) {
+    return std::forward<F>(f)(std::forward<T2>(x));


This is no longer returning the plain type now that the .derived() call is removed, should this at least be evaluating first since make_holder isn't used?

Is there anywhere these functions are being used now?

andrjohns · 2025-07-21T13:02:57Z

stan/math/prim/meta/holder.hpp

-    typename F, typename... Args,
-    require_not_plain_type_t<std::invoke_result_t<F, Args&&...>>* = nullptr>
+template <typename F, typename... Args,
+          require_not_plain_type_t<std::invoke_result_t<F, Args&&...>>*>


Are these still well-formed without the nullptr default?

These functions are forward declared at the top of the file. There we set the default of nullptr

andrjohns · 2025-07-21T13:06:57Z

stan/math/rev/fun/norm1.hpp

-    arena_v.adj().array() += res.adj() * sign(arena_v.val().array());
+template <typename Container,
+          require_eigen_vector_vt<is_var, Container>* = nullptr>
+inline var norm1(const Container& x) {


Both the norm1 and norm2 here have been updated but are still passing by const ref

andrjohns · 2025-07-21T13:09:38Z

test/unit/math/mix/fun/ub_constrain_test.cpp

@@ -119,9 +119,9 @@ TEST(mathMixMatFun, ub_stdvec_constrain_inf) {
 // matrix[], real
 TEST(mathMixMatFun, ub_stdvec_mat_mat_constrain) {
  Eigen::MatrixXd A_inner(2, 3);
-  A_inner << 5.0, 2.0, 4.0, -2.0, 0.0, 0.005;
+  A_inner << 5.0, 2.0, 4.0, -2.0, 1.0, 2.0;


Do these values need to be changed for the test to pass?

On my desktop machine I was getting weird floating point errors when the values were small. But on my mac I did not get any fp errors. But these tests were failing on my desktop computer before this PR.

stan-buildbot · 2025-07-29T22:04:45Z

Name	Old Result	New Result	Ratio	Performance change( 1 - new / old )
arma/arma.stan	0.36	0.34	1.07	6.44% faster
low_dim_corr_gauss/low_dim_corr_gauss.stan	0.01	0.01	1.22	17.82% faster
gp_regr/gen_gp_data.stan	0.03	0.03	0.8	-25.35% slower
gp_regr/gp_regr.stan	0.11	0.1	1.05	4.48% faster
sir/sir.stan	80.37	71.41	1.13	11.14% faster
irt_2pl/irt_2pl.stan	4.69	4.29	1.09	8.46% faster
eight_schools/eight_schools.stan	0.08	0.06	1.27	21.57% faster
pkpd/sim_one_comp_mm_elim_abs.stan	0.3	0.31	0.97	-2.77% slower
pkpd/one_comp_mm_elim_abs.stan	22.51	20.03	1.12	10.99% faster
garch/garch.stan	0.47	0.46	1.03	2.9% faster
low_dim_gauss_mix/low_dim_gauss_mix.stan	3.02	2.73	1.11	9.59% faster
arK/arK.stan	2.02	1.89	1.07	6.33% faster
gp_pois_regr/gp_pois_regr.stan	3.03	2.86	1.06	5.67% faster
low_dim_gauss_mix_collapse/low_dim_gauss_mix_collapse.stan	9.77	8.83	1.11	9.64% faster
performance.compilation	230.94	215.99	1.07	6.48% faster
Mean result: 1.077276098210825

Jenkins Console Log
Blue Ocean
Commit hash: 107cb07da5f07ab5bb08c566864260b2ab41c5d9

Machine information

No LSB modules are available. Distributor ID: Ubuntu Description: Ubuntu 20.04.3 LTS Release: 20.04 Codename: focal

CPU:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
Address sizes: 46 bits physical, 48 bits virtual
CPU(s): 80
On-line CPU(s) list: 0-79
Thread(s) per core: 2
Core(s) per socket: 20
Socket(s): 2
NUMA node(s): 2
Vendor ID: GenuineIntel
CPU family: 6
Model: 85
Model name: Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz
Stepping: 4
CPU MHz: 2400.000
CPU max MHz: 3700.0000
CPU min MHz: 1000.0000
BogoMIPS: 4800.00
Virtualization: VT-x
L1d cache: 1.3 MiB
L1i cache: 1.3 MiB
L2 cache: 40 MiB
L3 cache: 55 MiB
NUMA node0 CPU(s): 0,2,4,6,8,10,12,14,16,18,20,22,24,26,28,30,32,34,36,38,40,42,44,46,48,50,52,54,56,58,60,62,64,66,68,70,72,74,76,78
NUMA node1 CPU(s): 1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,33,35,37,39,41,43,45,47,49,51,53,55,57,59,61,63,65,67,69,71,73,75,77,79
Vulnerability Gather data sampling: Mitigation; Microcode
Vulnerability Itlb multihit: KVM: Mitigation: VMX disabled
Vulnerability L1tf: Mitigation; PTE Inversion; VMX conditional cache flushes, SMT vulnerable
Vulnerability Mds: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Meltdown: Mitigation; PTI
Vulnerability Mmio stale data: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Mitigation; IBRS
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; IBRS; IBPB conditional; STIBP conditional; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Mitigation; Clear CPU buffers; SMT vulnerable
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cdp_l3 invpcid_single pti intel_ppin ssbd mba ibrs ibpb stibp tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm cqm mpx rdt_a avx512f avx512dq rdseed adx smap clflushopt clwb intel_pt avx512cd avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req pku ospke md_clear flush_l1d arch_capabilities

G++:
g++ (Ubuntu 9.4.0-1ubuntu1~20.04) 9.4.0
Copyright (C) 2019 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

Clang:
clang version 10.0.0-4ubuntu1
Target: x86_64-pc-linux-gnu
Thread model: posix
InstalledDir: /usr/bin

WardBrian · 2025-07-30T01:16:48Z

test/prob/beta_neg_binomial/beta_neg_binomial_cdf_00000_generated_vv_test segfaulted on the latest run it looks like

…less they are an rvalue

…y are an rvalue

…ypes

stan-buildbot · 2025-08-01T20:21:09Z

Name	Old Result	New Result	Ratio	Performance change( 1 - new / old )
arma/arma.stan	0.31	0.3	1.03	3.31% faster
low_dim_corr_gauss/low_dim_corr_gauss.stan	0.01	0.01	1.1	9.1% faster
gp_regr/gen_gp_data.stan	0.02	0.02	1.05	4.77% faster
gp_regr/gp_regr.stan	0.09	0.09	1.04	3.39% faster
sir/sir.stan	68.28	68.45	1.0	-0.25% slower
irt_2pl/irt_2pl.stan	4.0	4.03	0.99	-0.57% slower
eight_schools/eight_schools.stan	0.06	0.05	1.05	4.5% faster
pkpd/sim_one_comp_mm_elim_abs.stan	0.24	0.24	1.0	0.31% faster
pkpd/one_comp_mm_elim_abs.stan	19.33	18.78	1.03	2.82% faster
garch/garch.stan	0.43	0.41	1.06	5.54% faster
low_dim_gauss_mix/low_dim_gauss_mix.stan	2.82	2.57	1.1	8.89% faster
arK/arK.stan	1.86	1.71	1.09	8.02% faster
gp_pois_regr/gp_pois_regr.stan	2.85	2.81	1.02	1.56% faster
low_dim_gauss_mix_collapse/low_dim_gauss_mix_collapse.stan	8.52	8.38	1.02	1.63% faster
performance.compilation	174.93	178.41	0.98	-1.99% slower
Mean result: 1.0364710339049357

Jenkins Console Log
Blue Ocean
Commit hash: e775f450bb86eacc0a0443dbaf4a36bc3407e56f

Machine information

No LSB modules are available. Distributor ID: Ubuntu Description: Ubuntu 20.04.3 LTS Release: 20.04 Codename: focal

CPU:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
Address sizes: 46 bits physical, 48 bits virtual
CPU(s): 80
On-line CPU(s) list: 0-79
Thread(s) per core: 2
Core(s) per socket: 20
Socket(s): 2
NUMA node(s): 2
Vendor ID: GenuineIntel
CPU family: 6
Model: 85
Model name: Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz
Stepping: 4
CPU MHz: 2400.000
CPU max MHz: 3700.0000
CPU min MHz: 1000.0000
BogoMIPS: 4800.00
Virtualization: VT-x
L1d cache: 1.3 MiB
L1i cache: 1.3 MiB
L2 cache: 40 MiB
L3 cache: 55 MiB
NUMA node0 CPU(s): 0,2,4,6,8,10,12,14,16,18,20,22,24,26,28,30,32,34,36,38,40,42,44,46,48,50,52,54,56,58,60,62,64,66,68,70,72,74,76,78
NUMA node1 CPU(s): 1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,33,35,37,39,41,43,45,47,49,51,53,55,57,59,61,63,65,67,69,71,73,75,77,79
Vulnerability Gather data sampling: Mitigation; Microcode
Vulnerability Itlb multihit: KVM: Mitigation: VMX disabled
Vulnerability L1tf: Mitigation; PTE Inversion; VMX conditional cache flushes, SMT vulnerable
Vulnerability Mds: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Meltdown: Mitigation; PTI
Vulnerability Mmio stale data: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Mitigation; IBRS
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; IBRS; IBPB conditional; STIBP conditional; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Mitigation; Clear CPU buffers; SMT vulnerable
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cdp_l3 invpcid_single pti intel_ppin ssbd mba ibrs ibpb stibp tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm cqm mpx rdt_a avx512f avx512dq rdseed adx smap clflushopt clwb intel_pt avx512cd avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req pku ospke md_clear flush_l1d arch_capabilities

G++:
g++ (Ubuntu 9.4.0-1ubuntu1~20.04) 9.4.0
Copyright (C) 2019 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

Clang:
clang version 10.0.0-4ubuntu1
Target: x86_64-pc-linux-gnu
Thread model: posix
InstalledDir: /usr/bin

WardBrian · 2025-08-05T19:41:29Z

I'm running the distribution tests on this branch with all the tests enabled + -O3 + ASan here. It is almost done, and so far looks good

stan-buildbot · 2025-08-06T06:14:38Z

Name	Old Result	New Result	Ratio	Performance change( 1 - new / old )
arma/arma.stan	0.31	0.32	0.98	-2.35% slower
low_dim_corr_gauss/low_dim_corr_gauss.stan	0.01	0.01	1.1	8.92% faster
gp_regr/gen_gp_data.stan	0.02	0.02	1.05	4.47% faster
gp_regr/gp_regr.stan	0.09	0.09	1.06	5.73% faster
sir/sir.stan	67.69	68.55	0.99	-1.27% slower
irt_2pl/irt_2pl.stan	4.0	3.99	1.0	0.33% faster
eight_schools/eight_schools.stan	0.05	0.05	1.02	1.88% faster
pkpd/sim_one_comp_mm_elim_abs.stan	0.24	0.25	0.97	-3.09% slower
pkpd/one_comp_mm_elim_abs.stan	19.06	18.68	1.02	1.99% faster
garch/garch.stan	0.43	0.4	1.07	6.25% faster
low_dim_gauss_mix/low_dim_gauss_mix.stan	2.67	2.56	1.04	4.13% faster
arK/arK.stan	1.75	1.72	1.02	1.85% faster
gp_pois_regr/gp_pois_regr.stan	2.8	2.71	1.03	3.15% faster
low_dim_gauss_mix_collapse/low_dim_gauss_mix_collapse.stan	8.51	8.39	1.01	1.36% faster
performance.compilation	177.97	178.78	1.0	-0.45% slower
Mean result: 1.0235379783466787

Jenkins Console Log
Blue Ocean
Commit hash: e775f450bb86eacc0a0443dbaf4a36bc3407e56f

Machine information

No LSB modules are available. Distributor ID: Ubuntu Description: Ubuntu 20.04.3 LTS Release: 20.04 Codename: focal

CPU:
Architecture: x86_64
CPU op-mode(s): 32-bit, 64-bit
Byte Order: Little Endian
Address sizes: 46 bits physical, 48 bits virtual
CPU(s): 80
On-line CPU(s) list: 0-79
Thread(s) per core: 2
Core(s) per socket: 20
Socket(s): 2
NUMA node(s): 2
Vendor ID: GenuineIntel
CPU family: 6
Model: 85
Model name: Intel(R) Xeon(R) Gold 6148 CPU @ 2.40GHz
Stepping: 4
CPU MHz: 2400.000
CPU max MHz: 3700.0000
CPU min MHz: 1000.0000
BogoMIPS: 4800.00
Virtualization: VT-x
L1d cache: 1.3 MiB
L1i cache: 1.3 MiB
L2 cache: 40 MiB
L3 cache: 55 MiB
NUMA node0 CPU(s): 0,2,4,6,8,10,12,14,16,18,20,22,24,26,28,30,32,34,36,38,40,42,44,46,48,50,52,54,56,58,60,62,64,66,68,70,72,74,76,78
NUMA node1 CPU(s): 1,3,5,7,9,11,13,15,17,19,21,23,25,27,29,31,33,35,37,39,41,43,45,47,49,51,53,55,57,59,61,63,65,67,69,71,73,75,77,79
Vulnerability Gather data sampling: Mitigation; Microcode
Vulnerability Itlb multihit: KVM: Mitigation: VMX disabled
Vulnerability L1tf: Mitigation; PTE Inversion; VMX conditional cache flushes, SMT vulnerable
Vulnerability Mds: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Meltdown: Mitigation; PTI
Vulnerability Mmio stale data: Mitigation; Clear CPU buffers; SMT vulnerable
Vulnerability Reg file data sampling: Not affected
Vulnerability Retbleed: Mitigation; IBRS
Vulnerability Spec rstack overflow: Not affected
Vulnerability Spec store bypass: Mitigation; Speculative Store Bypass disabled via prctl
Vulnerability Spectre v1: Mitigation; usercopy/swapgs barriers and __user pointer sanitization
Vulnerability Spectre v2: Mitigation; IBRS; IBPB conditional; STIBP conditional; RSB filling; PBRSB-eIBRS Not affected; BHI Not affected
Vulnerability Srbds: Not affected
Vulnerability Tsx async abort: Mitigation; Clear CPU buffers; SMT vulnerable
Flags: fpu vme de pse tsc msr pae mce cx8 apic sep mtrr pge mca cmov pat pse36 clflush dts acpi mmx fxsr sse sse2 ss ht tm pbe syscall nx pdpe1gb rdtscp lm constant_tsc art arch_perfmon pebs bts rep_good nopl xtopology nonstop_tsc cpuid aperfmperf pni pclmulqdq dtes64 monitor ds_cpl vmx smx est tm2 ssse3 sdbg fma cx16 xtpr pdcm pcid dca sse4_1 sse4_2 x2apic movbe popcnt tsc_deadline_timer aes xsave avx f16c rdrand lahf_lm abm 3dnowprefetch cpuid_fault epb cat_l3 cdp_l3 invpcid_single pti intel_ppin ssbd mba ibrs ibpb stibp tpr_shadow vnmi flexpriority ept vpid ept_ad fsgsbase tsc_adjust bmi1 hle avx2 smep bmi2 erms invpcid rtm cqm mpx rdt_a avx512f avx512dq rdseed adx smap clflushopt clwb intel_pt avx512cd avx512bw avx512vl xsaveopt xsavec xgetbv1 xsaves cqm_llc cqm_occup_llc cqm_mbm_total cqm_mbm_local dtherm ida arat pln pts hwp hwp_act_window hwp_epp hwp_pkg_req pku ospke md_clear flush_l1d arch_capabilities

G++:
g++ (Ubuntu 9.4.0-1ubuntu1~20.04) 9.4.0
Copyright (C) 2019 Free Software Foundation, Inc.
This is free software; see the source for copying conditions. There is NO
warranty; not even for MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.

Clang:
clang version 10.0.0-4ubuntu1
Target: x86_64-pc-linux-gnu
Thread model: posix
InstalledDir: /usr/bin

WardBrian · 2025-08-06T18:19:35Z

Looks like that full run passed 👍🏻

@andrjohns are you able to review again? Thanks!

SteveBronder and others added 18 commits July 9, 2025 17:58

First pass and passes mix fun test

df9a9b2

apply_scalar_binary. loglogistic now passes

b6e0a45

fix apply vector

13c4fd0

update ternary

dd3458c

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

45128f6

fix docs

aa5eb5e

fix docs

6ff0b0e

update test_ad_test to not use apply::return_t

4edea9f

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

696d2b2

fix template for sub_col and sub_row

2c8c9c5

update log1m_exp template constraints

2c1a328

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

912855d

fix templates for var and var matrix funs

dce5a63

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

fd7ad01

fix exp2

021e150

fix exp2

c677470

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

a77c2b6

remover binary operators for +/- etc from holder

f385fd2

WardBrian mentioned this pull request Jul 14, 2025

Release CmdStan 2.37 stan-dev/cmdstan#1324

Open

26 tasks

SteveBronder and others added 3 commits July 14, 2025 12:42

make operators for holder into free functions

1d5a64c

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

f5f0ead

have expr test on jenkins use gcc with sanitizer

e884013

WardBrian reviewed Jul 14, 2025

View reviewed changes

revert before merge: have expr test on jenkins use gcc with sanitizer…

57cd3bf

… and O2

SteveBronder added 2 commits July 15, 2025 11:09

fix templates for trigamma

b7336d2

revert changes to jenkinsfile

ac2aa50

stan-buildbot and others added 3 commits July 16, 2025 15:16

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

4517133

fix inv_square function deduction

6550d1e

fix typo in ordered_constrain

bd698a3

andrjohns reviewed Jul 21, 2025

View reviewed changes

andrjohns requested changes Jul 21, 2025

View reviewed changes

SteveBronder and others added 8 commits July 28, 2025 13:36

fixup from andrews review

4eae692

Merge commit '0c13850f5b8d853491d8bdfc85465ca7e500d432' into HEAD

b07f4d6

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

0c7c507

remove if constexpr in apply functors

61d8192

update sum_to_zero_constrain requires in rev

76229f0

do not use x after x_ref is made in ordered_contrain

926864a

make a const data() member function in holder

aba568d

make a const data() member function in holder

51ef9f8

SteveBronder and others added 8 commits July 31, 2025 11:52

fix as_column_vector_or_scalar so that it copies std::vector types un…

02d57f8

…less they are an rvalue

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

c083810

fix as_array_or_scalar so that it copies std::vector types unless the…

27dd922

…y are an rvalue

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

ba4252c

update dirichlet_multinomial

d51ce3f

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

cef520c

allow as_array_or_scalar to return a holder of a map for std vector t…

ce7af32

…ypes

[Jenkins] auto-formatting by clang-format version 10.0.0-4ubuntu1

1913a45

WardBrian requested a review from andrjohns August 4, 2025 13:57

	return exp(std::forward<decltype(x_ref)>(x));
	return exp(std::forward<decltype(x_ref)>(x_ref));

	return apply_vector_unary<T_vt>::apply(xx, f);
	return apply_vector_unary<T_vt>::apply(std::forward<decltype(xx)>(xx), f);

Uh oh!

Use Perfect Forwarding in all functions that use apply family of functors #3221

Are you sure you want to change the base?

Use Perfect Forwarding in all functions that use apply family of functors #3221

Uh oh!

Conversation

SteveBronder commented Jul 11, 2025

Summary

Tests

Side Effects

Release notes

Checklist

Uh oh!

WardBrian commented Jul 14, 2025

Uh oh!

SteveBronder commented Jul 14, 2025

Uh oh!

WardBrian left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Uh oh!

WardBrian commented Jul 15, 2025

Uh oh!

SteveBronder commented Jul 15, 2025

Uh oh!

andrjohns commented Jul 18, 2025

Uh oh!

SteveBronder commented Jul 18, 2025

Uh oh!

Choose a reason for hiding this comment

Uh oh!

andrjohns left a comment

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

Choose a reason for hiding this comment

Uh oh!

stan-buildbot commented Jul 29, 2025

Uh oh!

Use Perfect Forwarding in all functions that use `apply` family of functors #3221

Use Perfect Forwarding in all functions that use `apply` family of functors #3221