precice icon indicating copy to clipboard operation
precice copied to clipboard
verified publisher icon precice

QR Factorization test in acceleration fails for Release builds with Intel Compiler

Open fsimonis opened this issue 2 years ago • 6 comments

Describe your setup

Operating system (e.g. Linux distribution and version): Ubuntu 22.04 docker: precice/ci-intel:latest preCICE Version: 37a8111670971986fca96604cf93b60e7a8b8668

Add the versions/commit hashes of further adapter and/or bindings.

Describe the problem

The testQRFactorization test fails when compiling with the latest Intel Compiler icpx, in release with -ffp-model=precise.

mpirun -n 4 ./testprecice -t AccelerationTests/testQRFactorization

error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.5!=0.2
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.333333!=0.2
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.25!=0.2
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.333333!=0.166667
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.25!=0.166667
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.2!=0.166667
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.25!=0.142857
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.2!=0.142857
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.166667!=0.142857
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.2!=0.125
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.166667!=0.125
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.142857!=0.125
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.166667!=0.111111
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.142857!=0.111111
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.125!=0.111111
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.142857!=0.1
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.125!=0.1
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.111111!=0.1
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.125!=0.0909091
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.111111!=0.0909091
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.1!=0.0909091
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.111111!=0.0833333
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.1!=0.0833333
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.0909091!=0.0833333
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.2!=0.166667
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.166667!=0.142857
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.142857!=0.125
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.125!=0.111111
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.111111!=0.1
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.1!=0.0909091
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.0909091!=0.0833333
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/root/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.0833333!=0.0769231
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.

*** 32 failures are detected in the test module "preCICE Tests"

Step To Reproduce

  1. Install the latest intel compiler or use the precice/ci-intel:latest image.
  2. cmake -DCMAKE_BUILD_TYPE=Release -DCMAKE_CXX_FLAGS=-ffp-model=precise . and build.
  3. mpirun -n 4 ./testprecice -t AccelerationTests/testQRFactorization

Expected behaviour Passing test

Additional context

See #1826

fsimonis avatar Oct 25 '23 18:10 fsimonis

~~After #1854, this issue doesn't seem to persist.~~ Aaaand it's back #1826

The current state, however, doesn't require -ffp-model=precise, but instead adds -fhonor-infinities -fhonor-nans when compiling with IntelLLVM.

fsimonis avatar Oct 26 '23 09:10 fsimonis

Try this again when we have a release candidate for 3.0.0

BenjaminRodenberg avatar Nov 20 '23 14:11 BenjaminRodenberg

Testing this again now. The remaining PRs of 3.0.0 shouldn't have an impact on this.

fsimonis avatar Dec 01 '23 12:12 fsimonis

I can reproduce this with a release build with both -ffp-model=precice and -fhonor-infinities -fhonor-nans

/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.5!=0.2
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.333333!=0.2
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.25!=0.2
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.333333!=0.166667
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.25!=0.166667
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.2!=0.166667
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.25!=0.142857
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.2!=0.142857
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.166667!=0.142857
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.2!=0.125
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.166667!=0.125
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.142857!=0.125
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.166667!=0.111111
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.142857!=0.111111
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.125!=0.111111
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.142857!=0.1
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.125!=0.1
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.111111!=0.1
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.125!=0.0909091
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.111111!=0.0909091
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.1!=0.0909091
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.111111!=0.0833333
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.1!=0.0833333
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.0909091!=0.0833333
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.2!=0.166667
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.166667!=0.142857
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.142857!=0.125
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.125!=0.111111
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.111111!=0.1
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.1!=0.0909091
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.0909091!=0.0833333
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.
/home/precice/precice/src/acceleration/test/QRFactorizationTest.cpp(26): error: in "AccelerationTests/testQRFactorization": check testing::equals(A_prime(i, j), A(i, j)) has failed. Not equal: 0.0833333!=0.0769231
Failure occurred in a following context:
    Test context represents "Serial" and runs on rank 0 out of 1.

*** 32 failures are detected in the test module "preCICE Tests"

fsimonis avatar Dec 01 '23 13:12 fsimonis

Also, I cannot reproduce this with clang++ in Release and -ffast-math.

fsimonis avatar Dec 05 '23 12:12 fsimonis

Given how specific this problem is, it's worth dealing with once a becomes a problem and not now for 3.0. Moving to 3.x.x

fsimonis avatar Dec 06 '23 12:12 fsimonis

We integrated the new intel compilers into the CI #1826. While working on this, the issue has not popped up again.

I cannot reproduce this anymore. And users deliberately compiling preCICE with fast-math know that this can backfire. We tell intel to use the precise float model as it breaks the time handling.

Closing

fsimonis avatar Aug 16 '24 08:08 fsimonis