[R-pkg-devel] Examples with CPU time is greater than elapsed time.
Hi all, I tried to submit an update to the xgboost package but didn't pass the pre-tests with the following note (solved the other one, but this one is a bit confusing): ``` Flavor: r-devel-linux-x86_64-debian-gcc Check: examples, Result: NOTE Examples with CPU time > 2.5 times elapsed time user system elapsed ratio cb.gblinear.history 1.454 0.0170.49 3.002 ``` I can't reproduce the note on win-builder: https://win-builder.r-project.org/ as it's running on Windows but the note appears on debian tests. I'm not able to reproduce it on my local machine either with Ubuntu 22.04. I'm wondering what the note is trying to tell me and how can I resolve it with confidence. The full log is here: https://win-builder.r-project.org/incoming_pretest/xgboost_1.7.1.1_20221104_194252/Debian/00check.log . Many thanks! Jiaming __ R-package-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-package-devel
Re: [R-pkg-devel] Examples with CPU time is greater than elapsed time.
Thank you for the detailed explanation, that's really helpful! On 11/6/22 02:22, Henrik Bengtsson wrote: I think it's because it suggests that the package uses more than 250% of CPU load on average, which suggests it runs in parallel with more than two parallel workers, which is the upper limit in the CRAN Policies (https://cran.r-project.org/web/packages/policies.html); "If running a package uses multiple threads/cores it must never use more than two simultaneously: the check farm is a shared resource and will typically be running many checks simultaneously. " From the R Internals (https://cran.r-project.org/doc/manuals/r-release/R-ints.html): "_R_CHECK_EXAMPLE_TIMING_CPU_TO_ELAPSED_THRESHOLD_: For checks with timings enabled, report examples where the ratio of CPU time to elapsed time exceeds this threshold (and the CPU time is at least one second). This can help detect the simultaneous use of multiple CPU cores. Default: NA." It looks like CRAN incoming sets _R_CHECK_EXAMPLE_TIMING_CPU_TO_ELAPSED_THRESHOLD_=2.5. If multi-threading is involved, I guess you need to make sure to limit it to ~2 parallel threads. /Henrik On Sat, Nov 5, 2022 at 11:10 AM Jiaming Yuan wrote: Hi all, I tried to submit an update to the xgboost package but didn't pass the pre-tests with the following note (solved the other one, but this one is a bit confusing): ``` Flavor: r-devel-linux-x86_64-debian-gcc Check: examples, Result: NOTE Examples with CPU time > 2.5 times elapsed time user system elapsed ratio cb.gblinear.history 1.454 0.0170.49 3.002 ``` I can't reproduce the note on win-builder: https://win-builder.r-project.org/ as it's running on Windows but the note appears on debian tests. I'm not able to reproduce it on my local machine either with Ubuntu 22.04. I'm wondering what the note is trying to tell me and how can I resolve it with confidence. The full log is here: https://win-builder.r-project.org/incoming_pretest/xgboost_1.7.1.1_20221104_194252/Debian/00check.log . Many thanks! Jiaming __ R-package-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-package-devel __ R-package-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-package-devel
Re: [R-pkg-devel] Examples with CPU time is greater than elapsed time.
Thank you for the suggestions! I think I will hard-code the number of threads to 2 in that specific example. On 11/6/22 02:30, Ivan Krylov wrote: On Sat, 5 Nov 2022 22:41:45 +0800 Jiaming Yuan wrote: Examples with CPU time > 2.5 times elapsed time I'm wondering what the note is trying to tell me and how can I resolve it with confidence. Henrik Bengtsson already gave a good explanation of the problem. Not sure what exactly is the right thing to do, but one way to get rid of this warning could be to pass the argument nthread = min(parallel::detectCores(), 2) to the xgb.cv invocation (if you already use the parallel package anywhere) or maybe to hard-code the use of only one thread in the example. __ R-package-devel@r-project.org mailing list https://stat.ethz.ch/mailman/listinfo/r-package-devel
[R-pkg-devel] [SPAM | WERBUNG] Re: CRAN Submission xgboost 1.7.11.1
Thank you for sharing, Ivan! This is really helpful.
From: Ivan Krylov
Sent: Saturday, May 10, 2025 5:41:51 AM
To: jiaming yuan
Cc: r-package-devel@r-project.org
Subject: Re: [R-pkg-devel] CRAN Submission xgboost 1.7.11.1
� Fri, 9 May 2025 04:22:52 +
jiaming yuan �:
> We are trying to resolve the test failures of XGBoost but so far no
> one has managed to reproduce them locally.
> https://github.com/dmlc/xgboost/issues/11431
One test failure really stands out:
> Error ('test_feature_weights.R:17:5'): training with feature
> weights works
> Error in `setinfo.xgb.DMatrix(dmat, names(p), p[[1]])`:
> cannot get data pointer of 'double' objects
This is particularly strange because the internal CHKVEC() function is
supposed to check for REALSXP vectors and pass them through. Such
"impossible" errors could be signs of a protection error. (Well, any
memory corruption, but arbitrary memory corruption is harder to
diagnose, so let's hope it's something to do with the garbage
collector.) Let's run the test suite check using a build of R-devel
compiled with .../configure --enable-strict-barrier.
testthat::test_check("xgboost")
# <...>
# Fatal error: Wrong thread calling 'RunFinalizers'
# Fatal error: Wrong thread calling 'RunFinalizers'
# Fatal error: Wrong thread calling 'RunFinalizers'
# Fatal error: Wrong thread calling 'RunFinalizers'
# Fatal error: Wrong thread calling 'RunFinalizers'
# Segmentation fault
Hmm.
Setting a breakpoint in Rstd_Suicide reveals the mistake. It's not a
protection error; it's R API being called from an OpenMP thread:
#5 ALTVEC_DATAPTR (x=0x2d825f58) at ../../../R/src/main/altrep.c:376
#6 0x7fbbd1e5ed98 in operator() (__closure=,
i=) at xgboost_R.cc:258
#7 dmlc::OMPException::Run, int> (this=0x7fffbfaef760, f=...)
at ./dmlc-core/include/dmlc/common.h:67
#8
_ZN7xgboost6common11ParallelForIiZ18XGDMatrixSetInfo_REUlmE0_EEvT_iNS0_5SchedET0_._omp_fn.3(void)
()
at ./dmlc-core/include/../../src/common/threading_utils.h:208
#9 0x7fbbd4552a0e in gomp_thread_start (xdata=)
at ../../../libgomp/team.c:129
(gdb) frame 5
#5 ALTVEC_DATAPTR (x=0x2d825f58) at ../../../R/src/main/altrep.c:376
376 return ALTVEC_DATAPTR_EX(x, TRUE);
(gdb) call R_inspect(x)
@2d825f58 14 REALSXP g0c0 [REF(65535)] 1 : 9 (expanded)
The code would be better served by calling REAL_RO / INTEGER_RO once
and then using the returned pointer inside the OpenMP threads.
--
Best regards,
Ivan
[[alternative HTML version deleted]]
__
R-package-devel@r-project.org mailing list
https://stat.ethz.ch/mailman/listinfo/r-package-devel
Re: [R-pkg-devel] CRAN Submission xgboost 1.7.11.1
Hi Uwe,
Please help take a look into reproducing the failure if possible. We
really appreciate your help. Thanks to Ivan, we managed to reproduce the
memory protection error. But we haven't reproduced other test failures.
Cheers!
Jiaming
On 5/12/25 17:23, Uwe Ligges wrote:
We realized this has already been answered by Ivan Krylov on the
r-packages-devel list.
Best,
Uwe Ligges
On 12.05.2025 11:07, Uwe Ligges wrote:
I'll try to reproduce this locally.
On 09.05.2025 06:22, jiaming yuan wrote:
Hi CRAN,
May I ask if someone has tried to reproduce the openblas test
environment from CRAN? We are trying to resolve the test failures of
XGBoost but so far no one has managed to reproduce them locally.
https://github.com/dmlc/xgboost/issues/11431
Would be great if you can share some guidance on how to reproduce
that exact environment.
Cheers
Jiaming
____
From: jiaming yuan
Sent: Friday, May 9, 2025 12:04:45 PM
To: Uwe Ligges ; CRAN submissi...@r-project.org>
Cc: CRAN Package Submission Form
Subject: Re: CRAN Submission xgboost 1.7.11.1
Hi,
Others kindly provided help to reproduce the failure but so far no
one has managed to do so.
Please see https://github.com/dmlc/xgboost/
issues/11431#issuecomment-2864947065 and related discussions the
thread. Would be great if you can share something more precise.
Cheers
Jiaming
From: Uwe Ligges
Sent: Wednesday, May 7, 2025 9:48:56 PM
To: jiaming yuan ; CRAN project.org>
Cc: CRAN Package Submission Form
Subject: Re: CRAN Submission xgboost 1.7.11.1
Note this is relevant, as most Linux clusters will have admis who will
link scientifiv software aagainst OpenBLAS for faster matriox
operations.
Best,
Uwe Ligges
On 07.05.2025 15:48, Uwe Ligges wrote:
I think any Liniux system with R linked against the system's default
OpenBLAS installation will show this issue.
I'd try it with a standard Ubuntu or Debian with OpenBLAS installed
and
link agasinst it.
Best,
Uwe Ligges
On 03.05.2025 08:29, jiaming yuan wrote:
Thank you for reaching out. We can't really dive into it unless
there's an easier way to reproduce the environment (like a container
or using some deterministic package managers). It's very unlikely
that
we can try to build that environment on our own then try to fix all
errors and verify all fixes.
From: Uwe Ligges
Sent: Friday, May 2, 2025 8:00:17 PM
To: Jiaming Yuan ; CRAN
Cc: CRAN Package Submission Form
Subject: Re: CRAN Submission xgboost 1.7.11.1
Thanks, we see you removed lots of tests. Is this really sensible and
are you sure that users with OpenBLAS (as most Linux users and
cluster
admins will use) will get correct results? Sensibly relaxing
numerical
assumptions may be a better way to tweak the tests.?
Best,
Uwe Ligges
On 01.05.2025 12:58, CRAN Package Submission Form via
CRAN-submissions
wrote:
[This was generated from CRAN.R-project.org/submit.html]
The following package was uploaded to CRAN:
===
Package Information:
Package: xgboost
Version: 1.7.11.1
Title: Extreme Gradient Boosting
Author(s): Tianqi Chen [aut], Tong He [aut], Michael Benesty [aut],
Vadim
Khotilovich [aut], Yuan Tang [aut]
(<https://orcid.org/-0001-5243-233X>), Hyunsu Cho [aut],
Kailong Chen [aut], Rory Mitchell [aut], Ignacio Cano [aut],
Tianyi Zhou [aut], Mu Li [aut], Junyuan Xie [aut], Min Lin
[aut], Yifeng Geng [aut], Yutian Li [aut], Jiaming Yuan [aut,
cre], XGBoost contributors [cph] (base XGBoost implementation)
Maintainer: Jiaming Yuan
Depends: R (>= 3.3.0)
Suggests: knitr, rmarkdown, ggplot2 (>= 1.0.1), DiagrammeR (>=
0.9.0),
Ckmeans.1d.dp (>= 3.3.1), vcd (>= 1.3), cplm, e1071, caret,
testthat, lintr, igraph (>= 1.0.1), float, crayon, titanic
Description: Extreme Gradient Boosting, which is an efficient
implementation of the gradient boosting framework from Chen &
Guestrin (2016) . This package
is its R interface. The package includes efficient linear
model solver and tree learning algorithms. The package can
automatically do parallel computation on a single machine
which could be more than 10 times faster than existing
gradient boosting packages. It supports various objective
functions, including regression, classification and ranking.
The package is made to be extensible, so that users are also
allowed to define their own objectives easily.
License: Apache License (== 2.0) | file LICENSE
Imports: Matrix (>= 1.1-0), methods, data.table (>= 1.9.6), jsonlite
(>= 1.0),
The maintainer confirms that he or she
has read and agrees to the CRAN policies.
=
Original content of DESCRIPTION file:
Package: xgboost
Type: Package
Title: Extreme Gradient Boosting
Version: 1.7.11.1
Date: 2025-05-01
Authors@R: c(
person("Tianqi", "Chen", role =
Re: [R-pkg-devel] CRAN Submission xgboost 1.7.11.1
Hi CRAN, May I ask if someone has tried to reproduce the openblas test environment from CRAN? We are trying to resolve the test failures of XGBoost but so far no one has managed to reproduce them locally. https://github.com/dmlc/xgboost/issues/11431 Would be great if you can share some guidance on how to reproduce that exact environment. Cheers Jiaming From: jiaming yuan Sent: Friday, May 9, 2025 12:04:45 PM To: Uwe Ligges ; CRAN Cc: CRAN Package Submission Form Subject: Re: CRAN Submission xgboost 1.7.11.1 Hi, Others kindly provided help to reproduce the failure but so far no one has managed to do so. Please see https://github.com/dmlc/xgboost/issues/11431#issuecomment-2864947065 and related discussions the thread. Would be great if you can share something more precise. Cheers Jiaming From: Uwe Ligges Sent: Wednesday, May 7, 2025 9:48:56 PM To: jiaming yuan ; CRAN Cc: CRAN Package Submission Form Subject: Re: CRAN Submission xgboost 1.7.11.1 Note this is relevant, as most Linux clusters will have admis who will link scientifiv software aagainst OpenBLAS for faster matriox operations. Best, Uwe Ligges On 07.05.2025 15:48, Uwe Ligges wrote: > I think any Liniux system with R linked against the system's default > OpenBLAS installation will show this issue. > I'd try it with a standard Ubuntu or Debian with OpenBLAS installed and > link agasinst it. > > Best, > Uwe Ligges > > > > On 03.05.2025 08:29, jiaming yuan wrote: >> Thank you for reaching out. We can't really dive into it unless >> there's an easier way to reproduce the environment (like a container >> or using some deterministic package managers). It's very unlikely that >> we can try to build that environment on our own then try to fix all >> errors and verify all fixes. >> >> >> >> ____ >> From: Uwe Ligges >> Sent: Friday, May 2, 2025 8:00:17 PM >> To: Jiaming Yuan ; CRAN > project.org> >> Cc: CRAN Package Submission Form >> Subject: Re: CRAN Submission xgboost 1.7.11.1 >> >> Thanks, we see you removed lots of tests. Is this really sensible and >> are you sure that users with OpenBLAS (as most Linux users and cluster >> admins will use) will get correct results? Sensibly relaxing numerical >> assumptions may be a better way to tweak the tests.? >> >> Best, >> Uwe Ligges >> >> >> On 01.05.2025 12:58, CRAN Package Submission Form via CRAN-submissions >> wrote: >>> [This was generated from CRAN.R-project.org/submit.html] >>> >>> The following package was uploaded to CRAN: >>> === >>> >>> Package Information: >>> Package: xgboost >>> Version: 1.7.11.1 >>> Title: Extreme Gradient Boosting >>> Author(s): Tianqi Chen [aut], Tong He [aut], Michael Benesty [aut], >>> Vadim >>> Khotilovich [aut], Yuan Tang [aut] >>> (<https://orcid.org/-0001-5243-233X>), Hyunsu Cho [aut], >>> Kailong Chen [aut], Rory Mitchell [aut], Ignacio Cano [aut], >>> Tianyi Zhou [aut], Mu Li [aut], Junyuan Xie [aut], Min Lin >>> [aut], Yifeng Geng [aut], Yutian Li [aut], Jiaming Yuan [aut, >>> cre], XGBoost contributors [cph] (base XGBoost implementation) >>> Maintainer: Jiaming Yuan >>> Depends: R (>= 3.3.0) >>> Suggests: knitr, rmarkdown, ggplot2 (>= 1.0.1), DiagrammeR (>= 0.9.0), >>> Ckmeans.1d.dp (>= 3.3.1), vcd (>= 1.3), cplm, e1071, caret, >>> testthat, lintr, igraph (>= 1.0.1), float, crayon, titanic >>> Description: Extreme Gradient Boosting, which is an efficient >>> implementation of the gradient boosting framework from Chen & >>> Guestrin (2016) . This package >>> is its R interface. The package includes efficient linear >>> model solver and tree learning algorithms. The package can >>> automatically do parallel computation on a single machine >>> which could be more than 10 times faster than existing >>> gradient boosting packages. It supports various objective >>> functions, including regression, classification and ranking. >>> The package is made to be extensible, so that users are also >>> allowed to define their own objectives easily. >>> License: Apache License (== 2.0) | file LICENSE >>> Imports: Matrix (>= 1.1-0), methods, data.table (>= 1.9.6), jsonlite >>> (>= 1.0), >>> >>> >>> The maintainer confirms that he or she >>> has read and agrees to the CRAN policies. >>> >&g
