commit 94672fb2af6535adc6ea7fe8b8498580ad8cf3f4
author Georgios Pinitas <georgios.pinitas@arm.com> Wed Jan 22 18:36:27 2020 +0000
committer Georgios Pinitas <georgios.pinitas@arm.com> Fri Jan 31 11:36:14 2020 +0000
tree 189c68e7ff2c29d7800e09239da6d9a6364d5b05
parent 6a342648ae50beb8457871862f14fc9baef6b74f

COMPMID-3003: Integrate assembly kernels utilizing MMLA instruction.

MMLA is a matrix-multiply instruction introduced on armv8.6-A

Signed-off-by: Georgios Pinitas <georgios.pinitas@arm.com>
Change-Id: I572a54981d48f5a1e0e9e51102cb7ae28ad87806
Reviewed-on: https://review.mlplatform.org/c/ml/ComputeLibrary/+/2663
Tested-by: Arm Jenkins <bsgcomp@arm.com>
Reviewed-by: Michalis Spyrou <michalis.spyrou@arm.com>
Comments-Addressed: Arm Jenkins <bsgcomp@arm.com>

SConstruct

diff --git a/SConstruct b/SConstruct
index 3236e7c..0b49184 100644
--- a/SConstruct
+++ b/SConstruct
@@ -41,7 +41,8 @@
     BoolVariable("asserts", "Enable asserts (this flag is forced to 1 for debug=1)", False),
     BoolVariable("logging", "Logging (this flag is forced to 1 for debug=1)", False),
     EnumVariable("arch", "Target Architecture", "armv7a",
-                  allowed_values=("armv7a", "arm64-v8a", "arm64-v8.2-a", "arm64-v8.2-a-sve", "x86_32", "x86_64", "armv8a", "armv8.2-a", "armv8.2-a-sve", "x86")),
+                  allowed_values=("armv7a", "arm64-v8a", "arm64-v8.2-a", "arm64-v8.2-a-sve", "x86_32", "x86_64",
+                                  "armv8a", "armv8.2-a", "armv8.2-a-sve", "armv8.6-a", "x86")),
     EnumVariable("estate", "Execution State", "auto", allowed_values=("auto", "32", "64")),
     EnumVariable("os", "Target OS", "linux", allowed_values=("linux", "android", "bare_metal")),
     EnumVariable("build", "Build type", "cross_compile", allowed_values=("native", "cross_compile", "embed_only")),
@@ -194,17 +195,17 @@
         env.Append(CXXFLAGS = ['-mfloat-abi=softfp'])
     else:
         env.Append(CXXFLAGS = ['-mfloat-abi=hard'])
-elif 'v8a' in env['arch']:
-    env.Append(CXXFLAGS = ['-march=armv8-a'])
-    if env['estate'] == '32':
-        env.Append(CXXFLAGS = ['-mfpu=neon-fp-armv8'])
-elif 'v8.2-a' in env['arch']:
-    if env['estate'] == '32':
-        env.Append(CXXFLAGS = ['-mfpu=neon-fp-armv8'])
+elif 'v8' in env['arch']:
     if 'sve' in env['arch']:
         env.Append(CXXFLAGS = ['-march=armv8.2-a+sve+fp16+dotprod'])
-    else:
+    elif 'v8.2-a' in env['arch']:
         env.Append(CXXFLAGS = ['-march=armv8.2-a+fp16']) # explicitly enable fp16 extension otherwise __ARM_FEATURE_FP16_VECTOR_ARITHMETIC is undefined
+    else:
+        env.Append(CXXFLAGS = ['-march=armv8-a'])
+
+    if 'v8.6-a' in env['arch']:
+        env.Append(CXXFLAGS = ['-DV8P6'])
+
 elif 'x86' in env['arch']:
     if env['estate'] == '32':
         env.Append(CCFLAGS = ['-m32'])