about summary refs log tree commit diff
path: root/sysdeps/x86_64
diff options
context:
space:
mode:
authorAndrew Senkevich <andrew.senkevich@intel.com>2015-06-08 14:07:59 +0300
committerAndrew Senkevich <andrew.senkevich@intel.com>2015-06-08 14:07:59 +0300
commit5fe2a126d115e6b3220886f416b6b544252f8589 (patch)
tree2d0b7b7fb57135bc5885bcc0a771c9e99a52c734 /sysdeps/x86_64
parent3ac3ff325d4042c8a8e20821195938601994d8f8 (diff)
downloadglibc-5fe2a126d115e6b3220886f416b6b544252f8589.tar.gz
glibc-5fe2a126d115e6b3220886f416b6b544252f8589.tar.xz
glibc-5fe2a126d115e6b3220886f416b6b544252f8589.zip
This patch adds detection of availability for AVX512F and AVX512DQ ISAs.
    * sysdeps/x86_64/multiarch/init-arch.h (bit_AVX512F_Usable,
    bit_AVX512DQ_Usable, bit_Opmask_state, bit_ZMM0_15_state,
    bit_ZMM16_31_state): New macro.
    * sysdeps/x86_64/multiarch/init-arch.c (__init_cpu_features):
    Check and set bit_AVX512F_Usable, bit_AVX512DQ_Usable.
Diffstat (limited to 'sysdeps/x86_64')
-rw-r--r--sysdeps/x86_64/multiarch/init-arch.c17
-rw-r--r--sysdeps/x86_64/multiarch/init-arch.h17
2 files changed, 34 insertions, 0 deletions
diff --git a/sysdeps/x86_64/multiarch/init-arch.c b/sysdeps/x86_64/multiarch/init-arch.c
index 7dec21884d..aaad5fa841 100644
--- a/sysdeps/x86_64/multiarch/init-arch.c
+++ b/sysdeps/x86_64/multiarch/init-arch.c
@@ -179,6 +179,23 @@ __init_cpu_features (void)
 	  if (CPUID_AVX2)
 	    __cpu_features.feature[index_AVX2_Usable]
 	      |= bit_AVX2_Usable | bit_AVX_Fast_Unaligned_Load;
+	  /* Check if OPMASK state, upper 256-bit of ZMM0-ZMM15 and
+	     ZMM16-ZMM31 state are enabled.  */
+	  if ((xcrlow & (bit_Opmask_state | bit_ZMM0_15_state
+			 | bit_ZMM16_31_state)) ==
+	      (bit_Opmask_state | bit_ZMM0_15_state | bit_ZMM16_31_state))
+	    {
+	      /* Determine if AVX512F is usable.  */
+	      if (CPUID_AVX512F)
+		{
+		  __cpu_features.feature[index_AVX512F_Usable]
+		    |= bit_AVX512F_Usable;
+		  /* Determine if AVX512DQ is usable.  */
+		  if (CPUID_AVX512DQ)
+		    __cpu_features.feature[index_AVX512DQ_Usable]
+		      |= bit_AVX512DQ_Usable;
+		}
+	    }
 	  /* Determine if FMA is usable.  */
 	  if (CPUID_FMA)
 	    __cpu_features.feature[index_FMA_Usable] |= bit_FMA_Usable;
diff --git a/sysdeps/x86_64/multiarch/init-arch.h b/sysdeps/x86_64/multiarch/init-arch.h
index e6b5ba5530..cfc6e7049e 100644
--- a/sysdeps/x86_64/multiarch/init-arch.h
+++ b/sysdeps/x86_64/multiarch/init-arch.h
@@ -26,6 +26,8 @@
 #define bit_Slow_SSE4_2			(1 << 9)
 #define bit_AVX2_Usable			(1 << 10)
 #define bit_AVX_Fast_Unaligned_Load	(1 << 11)
+#define bit_AVX512F_Usable		(1 << 12)
+#define bit_AVX512DQ_Usable		(1 << 13)
 
 /* CPUID Feature flags.  */
 
@@ -43,10 +45,15 @@
 /* COMMON_CPUID_INDEX_7.  */
 #define bit_RTM		(1 << 11)
 #define bit_AVX2	(1 << 5)
+#define bit_AVX512F	(1 << 16)
+#define bit_AVX512DQ	(1 << 17)
 
 /* XCR0 Feature flags.  */
 #define bit_XMM_state  (1 << 1)
 #define bit_YMM_state  (2 << 1)
+#define bit_Opmask_state	(1 << 5)
+#define bit_ZMM0_15_state	(1 << 6)
+#define bit_ZMM16_31_state	(1 << 7)
 
 /* The integer bit array index for the first set of internal feature bits.  */
 # define FEATURE_INDEX_1 0
@@ -76,6 +83,8 @@
 # define index_Slow_SSE4_2		FEATURE_INDEX_1*FEATURE_SIZE
 # define index_AVX2_Usable		FEATURE_INDEX_1*FEATURE_SIZE
 # define index_AVX_Fast_Unaligned_Load	FEATURE_INDEX_1*FEATURE_SIZE
+# define index_AVX512F_Usable		FEATURE_INDEX_1*FEATURE_SIZE
+# define index_AVX512DQ_Usable		FEATURE_INDEX_1*FEATURE_SIZE
 
 #else	/* __ASSEMBLER__ */
 
@@ -152,6 +161,10 @@ extern const struct cpu_features *__get_cpu_features (void)
   HAS_CPUID_FLAG (COMMON_CPUID_INDEX_7, ebx, bit_RTM)
 # define CPUID_AVX2 \
   HAS_CPUID_FLAG (COMMON_CPUID_INDEX_7, ebx, bit_AVX2)
+# define CPUID_AVX512F \
+  HAS_CPUID_FLAG (COMMON_CPUID_INDEX_7, ebx, bit_AVX512F)
+# define CPUID_AVX512DQ \
+  HAS_CPUID_FLAG (COMMON_CPUID_INDEX_7, ebx, bit_AVX512DQ)
 
 /* HAS_* evaluates to true if we may use the feature at runtime.  */
 # define HAS_SSE2	HAS_CPU_FEATURE (COMMON_CPUID_INDEX_1, edx, bit_SSE2)
@@ -172,6 +185,8 @@ extern const struct cpu_features *__get_cpu_features (void)
 # define index_Slow_SSE4_2		FEATURE_INDEX_1
 # define index_AVX2_Usable		FEATURE_INDEX_1
 # define index_AVX_Fast_Unaligned_Load	FEATURE_INDEX_1
+# define index_AVX512F_Usable		FEATURE_INDEX_1
+# define index_AVX512DQ_Usable		FEATURE_INDEX_1
 
 # define HAS_ARCH_FEATURE(name) \
   ((__get_cpu_features ()->feature[index_##name] & (bit_##name)) != 0)
@@ -182,6 +197,8 @@ extern const struct cpu_features *__get_cpu_features (void)
 # define HAS_FAST_UNALIGNED_LOAD	HAS_ARCH_FEATURE (Fast_Unaligned_Load)
 # define HAS_AVX			HAS_ARCH_FEATURE (AVX_Usable)
 # define HAS_AVX2			HAS_ARCH_FEATURE (AVX2_Usable)
+# define HAS_AVX512F			HAS_ARCH_FEATURE (AVX512F_Usable)
+# define HAS_AVX512DQ			HAS_ARCH_FEATURE (AVX512DQ_Usable)
 # define HAS_FMA			HAS_ARCH_FEATURE (FMA_Usable)
 # define HAS_FMA4			HAS_ARCH_FEATURE (FMA4_Usable)
 # define HAS_AVX_FAST_UNALIGNED_LOAD	HAS_ARCH_FEATURE (AVX_Fast_Unaligned_Load)