Improve finite_diff_auto with Eigen expressions #3096

andrjohns · 2024-07-19T15:19:14Z

Summary

finite_diff_gradient_auto currently requires copying the input parameters and iteratively replacing values in the vector, requiring serial evaluation of the loop.

Rather than replacing values, we can instead use an Eigen expression to 'mask' a value at a specified index - avoiding the initial copy

Tests

N/A - current tests should still pass

Side Effects

N/A

Release notes

Improved efficiency of finite_diff_gradient_auto

Checklist

Copyright holder: Andrew Johnson

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

stan-buildbot · 2024-07-19T21:05:01Z

Name	Old Result	New Result	Ratio	Performance change( 1 - new / old )
arma/arma.stan	0.36	0.34	1.05	5.02% faster
low_dim_corr_gauss/low_dim_corr_gauss.stan	0.01	0.01	0.92	-8.69% slower
gp_regr/gen_gp_data.stan	0.02	0.02	0.98	-2.11% slower
gp_regr/gp_regr.stan	0.1	0.1	0.99	-1.34% slower
sir/sir.stan	70.22	71.47	0.98	-1.78% slower
irt_2pl/irt_2pl.stan	4.24	4.38	0.97	-3.2% slower
eight_schools/eight_schools.stan	0.06	0.06	1.0	-0.24% slower
pkpd/sim_one_comp_mm_elim_abs.stan	0.26	0.25	1.01	1.2% faster
pkpd/one_comp_mm_elim_abs.stan	19.43	19.04	1.02	2.02% faster
garch/garch.stan	0.44	0.41	1.08	7.17% faster
low_dim_gauss_mix/low_dim_gauss_mix.stan	2.74	2.63	1.04	4.13% faster
arK/arK.stan	1.82	1.75	1.04	3.83% faster
gp_pois_regr/gp_pois_regr.stan	2.87	2.73	1.05	4.67% faster
low_dim_gauss_mix_collapse/low_dim_gauss_mix_collapse.stan	8.88	8.38	1.06	5.66% faster
performance.compilation	195.09	189.29	1.03	2.98% faster
Mean result: 1.0147169123958284

Jenkins Console Log
Blue Ocean
Commit hash: c83a397be67c59ba135871db25c53717e78d796e

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: 3349.494
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 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
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

SteveBronder

Quick Q about move semantics but rest looks good!

SteveBronder · 2024-07-22T16:26:56Z

stan/math/prim/functor/finite_diff_gradient_auto.hpp

+      decltype(auto) x_temp
+          = EigT::NullaryExpr(x.size(), [&x, &i, &h, &j](Eigen::Index k) {
+              return k == i ? x[i] + h * h_scale[j] : x[k];
+            });
+      delta_f += f(std::forward<decltype(x_temp)>(x_temp)) * mults[j];


Why is this decltype(auto) instead of auto?

Also if it is being made here then it can be done using std::move

Given that it was taking a universal/forwarding reference, I wasn't sure if decltype(auto) was needed on the return to maintain any reference/const/etc status - but happy to drop if unnecessary!

stan-buildbot · 2024-07-23T10:26:32Z

Name	Old Result	New Result	Ratio	Performance change( 1 - new / old )
arma/arma.stan	0.33	0.33	0.99	-1.29% slower
low_dim_corr_gauss/low_dim_corr_gauss.stan	0.01	0.01	1.0	-0.21% slower
gp_regr/gen_gp_data.stan	0.02	0.03	0.78	-28.93% slower
gp_regr/gp_regr.stan	0.1	0.1	1.01	0.67% faster
sir/sir.stan	68.8	70.27	0.98	-2.15% slower
irt_2pl/irt_2pl.stan	4.16	3.93	1.06	5.53% faster
eight_schools/eight_schools.stan	0.06	0.06	1.08	7.67% faster
pkpd/sim_one_comp_mm_elim_abs.stan	0.25	0.23	1.06	5.76% faster
pkpd/one_comp_mm_elim_abs.stan	19.56	18.31	1.07	6.41% faster
garch/garch.stan	0.43	0.4	1.07	6.46% faster
low_dim_gauss_mix/low_dim_gauss_mix.stan	2.72	2.61	1.04	4.01% faster
arK/arK.stan	1.82	1.71	1.06	5.88% faster
gp_pois_regr/gp_pois_regr.stan	2.8	2.68	1.04	4.04% faster
low_dim_gauss_mix_collapse/low_dim_gauss_mix_collapse.stan	8.82	8.41	1.05	4.59% faster
performance.compilation	183.06	185.62	0.99	-1.4% slower
Mean result: 1.0178236872330537

Jenkins Console Log
Blue Ocean
Commit hash: c83a397be67c59ba135871db25c53717e78d796e

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 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
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

andrjohns and others added 2 commits July 19, 2024 18:09

Improve finite_diff_auto with Eigen expressions

78eabf6

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

318f28f

SteveBronder requested changes Jul 22, 2024

View reviewed changes

Drop decltype and forward

6789b9d

SteveBronder approved these changes Jul 26, 2024

View reviewed changes

andrjohns merged commit f120c5e into develop Jul 27, 2024
8 checks passed

andrjohns deleted the fdiff-copy branch July 27, 2024 12:23

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Improve finite_diff_auto with Eigen expressions #3096

Improve finite_diff_auto with Eigen expressions #3096

andrjohns commented Jul 19, 2024

stan-buildbot commented Jul 19, 2024

SteveBronder left a comment

SteveBronder Jul 22, 2024

andrjohns Jul 23, 2024 •

edited

Loading

stan-buildbot commented Jul 23, 2024

Improve finite_diff_auto with Eigen expressions #3096

Improve finite_diff_auto with Eigen expressions #3096

Conversation

andrjohns commented Jul 19, 2024

Summary

Tests

Side Effects

Release notes

Checklist

stan-buildbot commented Jul 19, 2024

SteveBronder left a comment

Choose a reason for hiding this comment

SteveBronder Jul 22, 2024

Choose a reason for hiding this comment

andrjohns Jul 23, 2024 • edited Loading

Choose a reason for hiding this comment

stan-buildbot commented Jul 23, 2024

andrjohns Jul 23, 2024 •

edited

Loading