From 28c8e9224f887a56936cf5f18423b426704334c2 Mon Sep 17 00:00:00 2001
From: Developer-Ecosystem-Engineering
 <65677710+Developer-Ecosystem-Engineering@users.noreply.github.com>
Date: Tue, 23 Aug 2022 12:53:48 -0700
Subject: ENH: Remove AVX related functions from non x86 based builds

Apple silicon builds of NumPy have extra functions in them for AVX2/AVX512.  The changes here remove those implementations if we're not building for x86.

Apple silicon:
- original size: 3946035 bytes
- new size: 3657731 bytes
- savings: 288304 bytes (7.31%)

Changes pass all tests on M1 native, M1 Rosetta, and iMacPro (AVX512). We've verified performance is the same before/after for Rosetta and iMacPro. We've also verified that binaries are exactly the same size and have the same number of symbols in them.
---
 numpy/core/setup.py | 21 +++++++++++++++++----
 1 file changed, 17 insertions(+), 4 deletions(-)

(limited to 'numpy/core/setup.py')

diff --git a/numpy/core/setup.py b/numpy/core/setup.py
index 65aacfdad..483e62a34 100644
--- a/numpy/core/setup.py
+++ b/numpy/core/setup.py
@@ -232,10 +232,23 @@ def check_math_capabilities(config, ext, moredefs, mathlibs):
                     ext.extra_compile_args.extend(
                             ['-ffixed-xmm%s' % n for n in range(16, 32)])
 
-    for dec, fn, code, header in OPTIONAL_FUNCTION_ATTRIBUTES_WITH_INTRINSICS:
-        if config.check_gcc_function_attribute_with_intrinsics(dec, fn, code,
-                                                               header):
-            moredefs.append((fname2def(fn), 1))
+    platform = sysconfig.get_platform()
+    if("x86_64" in platform):
+        for dec, fn in OPTIONAL_FUNCTION_ATTRIBUTES_AVX:
+            if config.check_gcc_function_attribute(dec, fn):
+                moredefs.append((fname2def(fn), 1))
+                if fn == 'attribute_target_avx512f':
+                    # GH-14787: Work around GCC<8.4 bug when compiling with AVX512
+                    # support on Windows-based platforms
+                    if (sys.platform in ('win32', 'cygwin') and
+                            config.check_compiler_gcc() and
+                            not config.check_gcc_version_at_least(8, 4)):
+                        ext.extra_compile_args.extend(
+                                ['-ffixed-xmm%s' % n for n in range(16, 32)])
+        for dec, fn, code, header in OPTIONAL_FUNCTION_ATTRIBUTES_WITH_INTRINSICS_AVX:
+            if config.check_gcc_function_attribute_with_intrinsics(dec, fn, code,
+                                                                   header):
+                moredefs.append((fname2def(fn), 1))
 
     for fn in OPTIONAL_VARIABLE_ATTRIBUTES:
         if config.check_gcc_variable_attribute(fn):
-- 
cgit v1.2.1