This is a set of tools for testing the applicability of Fortran compilers in my project that develops optimization solvers.
- 鲲鹏计算平台 (Kunpeng platform with Ubuntu 18.04)
- Ubuntu 20.04, Intel(R) Core(TM) i7-10610U
- Ubuntu 20.04, Intel(R) Core(TM) i7-4790 CPU
- ❎ 毕昇编译器 (Bisheng Compiler, based on the Classic flang) 1.3.3.b023 (fail:
Array
,Alloc
,Implied do
) - ☑️ Absoft
af95
in Absoft 64-bit Pro Fortran 22.0.2 with patch 3 - ❎ AOCC
flang
4.0.0 (fail:solve
,Array
,Alloc
,Implied do
) - ❎ Classic
flang
15.0.3 (fail:test_value
,solve
,Count
,Alloc
) - ❎ G95
g95
0.94 (insufficient support for F03 constructs) - ❎ GNU
gfortran
13.1.0 (fail:test_eqv
) - ☑️ Intel
ifort
2021.9.0 - ☑️ Intel
ifx
2023.1.0 - ❎ Lahey
lf95
L8.10b (insufficient support for F03 constructs) - ❎ NAG
nagfor
Release 7.1 (Hanzomon) Build 7143 (fail:test_sym
,test_isorth
,test_mult
,test_abs
) - ❎ NVIDIA
nvfortran
23.5 (fail:Alloc
,Count
,Implied do
) - ❎ Oracle
sunf95
12.6 (fail:test_vec
)
The compilers can be tested using the Makefile
by the commands below. Of course, you need to have the tested compiler installed on your computer,
and you may need to edit the Makefile
to fit your platform.
make atest # Test af95
make dtest # Test AOCC flang
make ftest # Test classic flang
make 9test # Test g95
make gtest # Test gfortran
make itest # Test ifort
make xtest # Test ifx
make ltest # Test lf95
make ntest # Test nagfor
make vtest # Test nvfortran
make stest # Test sunf95
-
Fortran Discourse: Availability and applicability of Fortran compilers for a project
-
Flang GitHub Issue: flang-new 19.0.0 bug:
nan > 0
,nan > 1
-
Flang GitHub Issue: flang-new 19.0.0 bug:
sum([Inf, 0.0]) = NaN
, whileInf + 0.0 = Inf
-
Flang GitHub Issue: False positive: out-bound subscripts
-
NVIDIA Developer Forums: Bug in nvfortran 22.3: false positive of out-bound subscripts
-
NVIDIA Developer Forums: Bug of nvfortran 22.2-0: array subscript triplet handled wrongly
-
Flang GitHub Issue: Array constructors fail
-
NVIDIA Developer Forums: Array constructor fails in
nvfortran 22.3-0
-
Intel Community: Bug: ifort and ifx crash when compiling a piece of (invalid) code involving ieee_arithmetic
-
Fortran Discourse:
Implied do
and array constructor: Strange behavior offlang
and related compilers (see also Flang issue #1200, NVIDIA Developer Forums, and AMD Community) -
Intel Community: Bug in
ifort
andifx
from oneAPI 2021.2.0 & 2022.0.1: false positive of unused variable -
Intel Community: Bug? Strange error raised by
ifort -check shape
-
Intel Communities: A bug of ifort (IFORT) 2021.7.1 20221019
-
Flang GitHub Issue: Flang Binary Release 2019-03-29:
NaN > 0
-
Intel Community: ifort raises a SIGSEGV when evaluating "NaN >= 0"
-
Fortran Discourse: Is comparison with
NaN
considered an erroneous arithmetic operation? -
Intel Community: Ifort (IFORT) 2021.8 evaluates
1.0+37 / 1.0E+38
to 0 -
Fortran Discourse: Ifort (IFORT) 2021.8.0:
1.0E+37 / 1.0E+38 = 0
-
Intel Community: Fail to install oneAPI 2021 on Windows Server 2022
-
Fortran Discourse: Warning: ‘.str’ may be used uninitialized
Feel free to contact me if you would like to suggest a compiler or have questions about the test. See my homepage for contact information.