Mixed precision: fms_mod #1147

J-Lentz · 2023-03-07T22:23:57Z

Description
monotonic_array in fms_mod has been converted to mixed precision, and a unit test has been written for it.

How Has This Been Tested?
Unit test compiles, runs, and passes on the AMD box with gcc and Intel.

Checklist:

My code follows the style guidelines of this project
I have performed a self-review of my own code
I have commented my code, particularly in hard-to-understand areas
I have made corresponding changes to the documentation
My changes generate no new warnings
Any dependent changes have been merged and published in downstream modules
New check tests, if applicable, are included
make distcheck passes

Implement mixed mode for the `monotonic_array` function in `fms_mod`.

Implement a unit test for the `monotonic_array` function in `fms_mod`.

Merge the `string` branch, which contains an extended version of `string` and the new `stringify` interface from `fms_string_utils_mod`.

Remove the PRETTY() macro from the `fms_mod` test, as `string` has been updated to strip leading and trailing whitespace from its output.

To be consistent with mixed-mode code guidelines, define the kind macros in `test_fms_r4.fh` and `test_fms_r8.fh` rather than in `test_fms.F90`.

Replace the `C(x)` macro with a local variable that aliases the `FMS_MOD_TEST_KIND_` macro. Also, add underscores to the end of all macro names to be consistent with mixed-mode code guidelines.

Add more test cases to the `monotonic_array` unit test.

fms/include/fms.inc

thomas-robinson · 2023-03-10T18:43:36Z

fms/include/fms.inc

 integer, intent(out), optional :: direction !< If the input array is:
                                            !! >> monotonic (small to large) then direction = +1.
                                            !! >> monotonic (large to small) then direction = -1.
                                            !! >> not monotonic then direction = 0.
-logical :: monotonic_array !< If the input array of real values either increases or decreases monotonically
-                           !! then TRUE is returned, otherwise FALSE is returned.
+logical :: MONOTONIC_ARRAY_ !< If the input array of real values either increases or decreases monotonically


I realize you didn't write this code @J-Lentz , but this seems very odd to me. I would just declare this function as logical, or have a return value and declare that return value as logical. This is somehow both and neither at the same time.

Return variable has been replaced with ret in c245ad6.

thomas-robinson · 2023-03-10T18:48:04Z

string_utils/fms_string_utils.F90

+  module procedure string_from_logical
+  module procedure string_from_integer
+  module procedure string_from_r4, string_from_r8
+end interface


Will the conflict with 2023.01-beta3?

Updated fms_string_utils_mod merged from main in 42c2d1c.

mlee03 · 2023-03-15T11:26:15Z

CMakeLists.txt

                                                 mpp/include
-                                                 constants)
+                                                 constants
+                                                 test_fms/fms/include)


not sure if test_fms is needed for CMake yet. @rem1776, is this correct?

Yeah that's correct we don't have cmake set up to run the tests so we don't need to compile them

Removed from CMakeLists.txt in 1a1f887.

mlee03 · 2023-03-15T11:26:28Z

CMakeLists.txt

-    $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/mpp/include>)
+    $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/string_utils/include>
+    $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/mpp/include>
+    $<BUILD_INTERFACE:${CMAKE_CURRENT_SOURCE_DIR}/test_fms/fms/include>)


Removed in 1a1f887.

mlee03 · 2023-03-15T11:27:37Z

fms/Makefile.am

@@ -32,6 +32,9 @@ noinst_LTLIBRARIES = libfms.la
 # Each convenience library depends on its source.
 libfms_la_SOURCES = \
  fms.F90 \
+  include/fms.inc \


are these also needed for fms_mod.$(FC_MODEXT)?

mlee03 · 2023-03-15T11:34:31Z

fms/include/fms.inc

-real,    intent(in)            :: array(:) !< An array of real values. If the size(array) < 2 this function
-                                           !! assumes the array is not monotonic, no fatal error will occur.
+
+function MONOTONIC_ARRAY_(array, direction) result(ret)


I guess this routine doesn't take into account elements having the same value, e.g. -2, -1, 0, 0, 1, 2

Need to get my math concepts straight. Monotonic increasing/decreasing = strictly increasing/decreasing and not remaining constant. All's good.

I believe monotonic functions, as commonly defined, can have plateaus of constant value. So this function seems to use a peculiar definition of "monotonic".

There is a difference between a monotonic and a strictly monotonic function. If this function only enforces (X₁ < X₂), then it is testing for strict increasing monotonicity. If it is (X₁ > X₂), then it is enforcing strict decreasing monotonicity.

Perhaps it's worth commenting in the code that monotonic refers to strictly monotonic.

mlee03 · 2023-03-23T12:27:48Z

fms/include/fms_r4.fh

#def #undef #def #undef include format

mlee03 · 2023-03-23T12:28:00Z

fms/include/fms_r8.fh

mlee03 · 2023-03-23T12:29:31Z

test_fms/fms/include/test_fms_r8.fh

Indicate in the Doxygen comments that `monotonic_array` checks for strict monotonicity.

…re, and fms_mod (#1239) * feat: mixed precision axis_utils2 (#1104) * feat: mixed precision fms_mod (#1147) * feat: horiz interp mixed precision (#1067) * mixed precision sat_vapor_pressure (#1095) * feat: add mixed precision axis_utils unit tests (#1172) * fix: move type definitions to before first usage to fix nvhpc bug (#1187) * fix: change allocatable type for intel errors (#1221) Co-authored-by: Caitlyn McAllister <65364559+mcallic2@users.noreply.github.com> Co-authored-by: Jesse Lentz <42011922+J-Lentz@users.noreply.github.com> Co-authored-by: MiKyung Lee <58964324+mlee03@users.noreply.github.com>

…re, and fms_mod (#1239) (#1258) * feat: mixed precision axis_utils2 (#1104) * feat: mixed precision fms_mod (#1147) * feat: horiz interp mixed precision (#1067) * mixed precision sat_vapor_pressure (#1095) * feat: add mixed precision axis_utils unit tests (#1172) * fix: move type definitions to before first usage to fix nvhpc bug (#1187) * fix: change allocatable type for intel errors (#1221) Co-authored-by: Caitlyn McAllister <65364559+mcallic2@users.noreply.github.com> Co-authored-by: Jesse Lentz <42011922+J-Lentz@users.noreply.github.com> Co-authored-by: MiKyung Lee <58964324+mlee03@users.noreply.github.com>

Jesse Lentz added 9 commits March 1, 2023 14:29

fms_mod: Implement mixed mode for monotonic_array

f5a7eec

Implement mixed mode for the `monotonic_array` function in `fms_mod`.

Implement monotonic_array unit test

9129377

Implement a unit test for the `monotonic_array` function in `fms_mod`.

Merge string branch

a703614

Merge the `string` branch, which contains an extended version of `string` and the new `stringify` interface from `fms_string_utils_mod`.

Remove PRETTY() macro

fbcb5e1

Remove the PRETTY() macro from the `fms_mod` test, as `string` has been updated to strip leading and trailing whitespace from its output.

Define kind macros in *.fh include files

327d1a5

To be consistent with mixed-mode code guidelines, define the kind macros in `test_fms_r4.fh` and `test_fms_r8.fh` rather than in `test_fms.F90`.

Remove C(x) macro

e9b122d

Replace the `C(x)` macro with a local variable that aliases the `FMS_MOD_TEST_KIND_` macro. Also, add underscores to the end of all macro names to be consistent with mixed-mode code guidelines.

Add additional test cases

e436c85

Add more test cases to the `monotonic_array` unit test.

Move fms_mod macro definitions into *.fh files

a6de483

Break lines >110 characters

20ee3aa

J-Lentz requested review from thomas-robinson, rem1776 and bensonr as code owners March 7, 2023 22:23

J-Lentz requested review from mcallic2 and mlee03 March 7, 2023 22:24

mlee03 reviewed Mar 8, 2023

View reviewed changes

fms/include/fms.inc Show resolved Hide resolved

thomas-robinson reviewed Mar 10, 2023

View reviewed changes

Jesse Lentz added 2 commits March 10, 2023 16:50

Merge updated fms_string_utils_mod from main

a71c20e

Use result(ret) for monotonic_array return value

c245ad6

mlee03 reviewed Mar 15, 2023

View reviewed changes

Merge branch 'mixedmode' into fms_mod

1a1f887

mlee03 reviewed Mar 23, 2023

View reviewed changes

Jesse Lentz added 4 commits March 23, 2023 14:28

Declare include files as dependencies for fms_mod.mod.

850d957

Improve monotonic_array documentation

0b397a6

Indicate in the Doxygen comments that `monotonic_array` checks for strict monotonicity.

Add extra #undefs

2f60d9b

More extra #undefs

2dee075

thomas-robinson approved these changes Mar 27, 2023

View reviewed changes

mlee03 approved these changes Mar 27, 2023

View reviewed changes

thomas-robinson mentioned this pull request Mar 27, 2023

Mixed precision axis_utils2 #1104

Merged

8 tasks

mcallic2 approved these changes Mar 27, 2023

View reviewed changes

rem1776 merged commit 6b3855a into NOAA-GFDL:mixedmode Mar 29, 2023

rem1776 pushed a commit to rem1776/FMS that referenced this pull request May 25, 2023

feat: mixed precision fms_mod (NOAA-GFDL#1147)

51e7541

J-Lentz deleted the fms_mod branch July 19, 2023 18:26

Provide feedback

Saved searches

Use saved searches to filter your results more quickly

Mixed precision: fms_mod #1147

Mixed precision: fms_mod #1147

J-Lentz commented Mar 7, 2023 •

edited

Loading

thomas-robinson Mar 10, 2023

J-Lentz Mar 10, 2023

thomas-robinson Mar 10, 2023

J-Lentz Mar 10, 2023

mlee03 Mar 15, 2023

rem1776 Mar 15, 2023

J-Lentz Mar 23, 2023

mlee03 Mar 15, 2023

J-Lentz Mar 23, 2023

mlee03 Mar 15, 2023

J-Lentz Mar 29, 2023

mlee03 Mar 15, 2023

mlee03 Mar 15, 2023

J-Lentz Mar 15, 2023

bensonr Mar 15, 2023

mlee03 Mar 15, 2023

J-Lentz Mar 23, 2023

mlee03 Mar 23, 2023

J-Lentz Mar 23, 2023

mlee03 Mar 23, 2023

J-Lentz Mar 23, 2023

mlee03 Mar 23, 2023

J-Lentz Mar 23, 2023

Mixed precision: fms_mod #1147

Mixed precision: fms_mod #1147

Conversation

J-Lentz commented Mar 7, 2023 • edited Loading

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

Choose a reason for hiding this comment

J-Lentz commented Mar 7, 2023 •

edited

Loading