Change default sig_figs value to 8 by WardBrian · Pull Request #1308

Change default sig_figs value to 8 by WardBrian · Pull Request #1308 · stan-dev/cmdstan

Submisison Checklist

Run tests: ./runCmdStanTests.py src/test
Declare copyright holder and open-source license: see below

Summary:

Closes #1307

Intended Effect:

Increase the number of outputs saved by default

How to Verify:

Run a model without changing the sig_figs argument

Side Effects:

Additional storage may be required

Documentation:

TBD

Copyright and Licensing

Please list the copyright holder for the work you are submitting (this will be you or your assignee, such as a university or company):

Simons Foundation

By submitting this pull request, the copyright holder is agreeing to 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/)

Name	Old Result	New Result	Ratio	Performance change( 1 - new / old )
arma/arma.stan	0.34	0.33	1.04	3.66% faster
low_dim_corr_gauss/low_dim_corr_gauss.stan	0.01	0.01	0.98	-2.28% slower
gp_regr/gen_gp_data.stan	0.03	0.03	0.98	-1.67% slower
gp_regr/gp_regr.stan	0.1	0.1	1.01	1.31% faster
sir/sir.stan	72.62	71.38	1.02	1.72% faster
irt_2pl/irt_2pl.stan	4.41	4.19	1.05	4.86% faster
eight_schools/eight_schools.stan	0.06	0.06	1.03	2.52% faster
pkpd/sim_one_comp_mm_elim_abs.stan	0.26	0.26	1.0	0.09% faster
pkpd/one_comp_mm_elim_abs.stan	20.22	19.9	1.02	1.58% faster
garch/garch.stan	0.46	0.44	1.05	4.91% faster
low_dim_gauss_mix/low_dim_gauss_mix.stan	2.81	2.86	0.98	-1.63% slower
arK/arK.stan	1.82	1.88	0.97	-3.57% slower
gp_pois_regr/gp_pois_regr.stan	2.93	2.99	0.98	-1.82% slower
low_dim_gauss_mix_collapse/low_dim_gauss_mix_collapse.stan	9.15	9.12	1.0	0.41% faster
performance.compilation	193.06	190.72	1.01	1.21% faster
Mean result: 1.0082349896964151

Jenkins Console Log
Blue Ocean
Commit hash: 318925c5ff1fb92e891ac6e599f194f520a8d465

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

also need to change the CmdStan docs

also need to change the CmdStan docs

Yes, will change several places if this gets merged

Can you try with the example in stan-dev/cmdstanr#420 (it's an issue in CmdStanR repo, but can be tested with any approach calling standalone generated quantities)?

@avehtari for the specific example you link, it does appear to be resolved by this PR (passed over 100 times in a row)

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I think there's a slipped default, so I took the liberty of suggesting a wording change while you're at it.

		_default = "-1";
		_default_value = -1;
		= "0 <= integer <= 18, or -1 to use the system library's default "
		"number of significant figures (usually 6).";

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

shouldn't this 6 be an 8?

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

No, -1 is preserved to do the same thing as before, which is to just not specify any target precision when the stream is created, leaving it with whatever your OS defaults to. This is why it’s the “system library’s default”, not ours — 6 comes from glibc and friends.

If we want, we can make -1 synonymous with our default (8) instead

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

(Though I should note, if you want to use Stan’s default, a reasonable option is just not pass this argument at all…)

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I don't think this is a huge backward compatiblity concern and it would certainly be simple to have just one default. In which case, I don't think we'd need a behavior for -1, which would simplify the doc.

Having said that, I think it's OK for "unspecified" to be the Stan default and -1 the system default as long as it's a bit more clearly stated in the doc that the system default is the underlying system default, not the Stan default.

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I would prefer to just not allow -1 at all, but removing a valid value would definitely break at least somebody’s code

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

Good point---I missed that was the old default!

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I've updated the wording. Here's what the full help text reads as now:

    sig_figs=<int>
      The number of significant figures used for the output CSV files.
      Valid values: 0 <= integer <= 18, or -1 to use the current operating system's default number of significant figures (This will vary, but on many systems it is 6).
      Defaults to 8

Name	Old Result	New Result	Ratio	Performance change( 1 - new / old )
arma/arma.stan	0.34	0.39	0.87	-15.49% slower
low_dim_corr_gauss/low_dim_corr_gauss.stan	0.01	0.01	0.96	-4.64% slower
gp_regr/gen_gp_data.stan	0.02	0.02	0.98	-2.33% slower
gp_regr/gp_regr.stan	0.1	0.09	1.01	1.36% faster
sir/sir.stan	70.61	70.61	1.0	0.0% slower
irt_2pl/irt_2pl.stan	4.13	3.88	1.06	6.07% faster
eight_schools/eight_schools.stan	0.07	0.05	1.25	19.69% faster
pkpd/sim_one_comp_mm_elim_abs.stan	0.26	0.24	1.07	6.59% faster
pkpd/one_comp_mm_elim_abs.stan	19.8	18.86	1.05	4.74% faster
garch/garch.stan	0.43	0.4	1.07	6.31% faster
low_dim_gauss_mix/low_dim_gauss_mix.stan	2.76	2.58	1.07	6.31% faster
arK/arK.stan	1.83	1.7	1.07	6.73% faster
gp_pois_regr/gp_pois_regr.stan	2.89	2.7	1.07	6.41% faster
low_dim_gauss_mix_collapse/low_dim_gauss_mix_collapse.stan	8.86	8.35	1.06	5.82% faster
performance.compilation	194.07	182.6	1.06	5.91% faster
Mean result: 1.042847637560979

Jenkins Console Log
Blue Ocean
Commit hash: b3ebfc9cf0b6967e9be77a03696d221959159748

Machine information

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

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

Choose a reason for hiding this comment

The reason will be displayed to describe this comment to others. Learn more.

I still find two things about this confusing:

the use of "integer"---is that in other ones of these?
there's no mention that the platform default is not the Stan default

But if you want to go with it like this, I'm just going to approve because it's not worth a lot of back and forth.

the use of "integer"---is that in other ones of these?

It doesn't look like it, I've updated it to be closer to the others.

    sig_figs=<int>
      The number of significant figures used for the output CSV files. 
      Stan recommends at least 8 (the default) if you will be using the output as part 
      of a future input (e.g., for standalone generated quantities or as an initialization). 
      If -1 is supplied, Stan will use the current operating system's default precision 
      (This typically be smaller than Stan's default; on many systems it is 6 digits).
      Valid values: 0 <= sig_figs <= 18 || sig_figs == -1
      Defaults to 8

Name	Old Result	New Result	Ratio	Performance change( 1 - new / old )
arma/arma.stan	0.31	0.34	0.9	-10.54% slower
low_dim_corr_gauss/low_dim_corr_gauss.stan	0.01	0.01	1.09	8.16% faster
gp_regr/gen_gp_data.stan	0.02	0.02	1.0	0.37% faster
gp_regr/gp_regr.stan	0.09	0.09	1.0	0.39% faster
sir/sir.stan	67.15	69.38	0.97	-3.32% slower
irt_2pl/irt_2pl.stan	3.89	3.9	1.0	-0.27% slower
eight_schools/eight_schools.stan	0.05	0.05	1.02	1.88% faster
pkpd/sim_one_comp_mm_elim_abs.stan	0.24	0.24	0.99	-0.83% slower
pkpd/one_comp_mm_elim_abs.stan	18.69	18.74	1.0	-0.25% slower
garch/garch.stan	0.41	0.4	1.02	1.54% faster
low_dim_gauss_mix/low_dim_gauss_mix.stan	2.63	2.59	1.02	1.48% faster
arK/arK.stan	1.73	1.72	1.0	0.44% faster
gp_pois_regr/gp_pois_regr.stan	2.72	2.71	1.01	0.69% faster
low_dim_gauss_mix_collapse/low_dim_gauss_mix_collapse.stan	8.5	8.37	1.02	1.5% faster
performance.compilation	181.2	178.4	1.02	1.54% faster
Mean result: 1.0031761861827349

Jenkins Console Log
Blue Ocean
Commit hash: b3ebfc9cf0b6967e9be77a03696d221959159748

Machine information

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

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