about summary refs log tree commit diff
path: root/sysdeps/unix/sysv/linux
diff options
context:
space:
mode:
authorNaohiro Tamura <naohirot@jp.fujitsu.com>2021-05-27 07:42:35 +0000
committerSzabolcs Nagy <szabolcs.nagy@arm.com>2021-05-27 09:47:53 +0100
commitfa527f345cbbe852ec085932fbea979956c195b5 (patch)
tree6d84a3c41c4ed9ac4ffce5c33e7448dd70075e87 /sysdeps/unix/sysv/linux
parentf12ec02f5389a443d892241c486d87b3c5940ff6 (diff)
downloadglibc-fa527f345cbbe852ec085932fbea979956c195b5.tar.gz
glibc-fa527f345cbbe852ec085932fbea979956c195b5.tar.xz
glibc-fa527f345cbbe852ec085932fbea979956c195b5.zip
aarch64: Added optimized memcpy and memmove for A64FX
This patch optimizes the performance of memcpy/memmove for A64FX [1]
which implements ARMv8-A SVE and has L1 64KB cache per core and L2 8MB
cache per NUMA node.

The performance optimization makes use of Scalable Vector Register
with several techniques such as loop unrolling, memory access
alignment, cache zero fill, and software pipelining.

SVE assembler code for memcpy/memmove is implemented as Vector Length
Agnostic code so theoretically it can be run on any SOC which supports
ARMv8-A SVE standard.

We confirmed that all testcases have been passed by running 'make
check' and 'make xcheck' not only on A64FX but also on ThunderX2.

And also we confirmed that the SVE 512 bit vector register performance
is roughly 4 times better than Advanced SIMD 128 bit register and 8
times better than scalar 64 bit register by running 'make bench'.

[1] https://github.com/fujitsu/A64FX

Reviewed-by: Wilco Dijkstra <Wilco.Dijkstra@arm.com>
Reviewed-by: Szabolcs Nagy <Szabolcs.Nagy@arm.com>
Diffstat (limited to 'sysdeps/unix/sysv/linux')
-rw-r--r--sysdeps/unix/sysv/linux/aarch64/cpu-features.c4
-rw-r--r--sysdeps/unix/sysv/linux/aarch64/cpu-features.h4
2 files changed, 8 insertions, 0 deletions
diff --git a/sysdeps/unix/sysv/linux/aarch64/cpu-features.c b/sysdeps/unix/sysv/linux/aarch64/cpu-features.c
index db6aa3516c..6206a2f618 100644
--- a/sysdeps/unix/sysv/linux/aarch64/cpu-features.c
+++ b/sysdeps/unix/sysv/linux/aarch64/cpu-features.c
@@ -46,6 +46,7 @@ static struct cpu_list cpu_list[] = {
       {"ares",		 0x411FD0C0},
       {"emag",		 0x503F0001},
       {"kunpeng920", 	 0x481FD010},
+      {"a64fx",		 0x460F0010},
       {"generic", 	 0x0}
 };
 
@@ -116,4 +117,7 @@ init_cpu_features (struct cpu_features *cpu_features)
 	     (PR_TAGGED_ADDR_ENABLE | PR_MTE_TCF_ASYNC | MTE_ALLOWED_TAGS),
 	     0, 0, 0);
 #endif
+
+  /* Check if SVE is supported.  */
+  cpu_features->sve = GLRO (dl_hwcap) & HWCAP_SVE;
 }
diff --git a/sysdeps/unix/sysv/linux/aarch64/cpu-features.h b/sysdeps/unix/sysv/linux/aarch64/cpu-features.h
index 3b9bfed134..2b322e5414 100644
--- a/sysdeps/unix/sysv/linux/aarch64/cpu-features.h
+++ b/sysdeps/unix/sysv/linux/aarch64/cpu-features.h
@@ -65,6 +65,9 @@
 #define IS_KUNPENG920(midr) (MIDR_IMPLEMENTOR(midr) == 'H'			   \
                         && MIDR_PARTNUM(midr) == 0xd01)
 
+#define IS_A64FX(midr) (MIDR_IMPLEMENTOR(midr) == 'F'			      \
+			&& MIDR_PARTNUM(midr) == 0x001)
+
 struct cpu_features
 {
   uint64_t midr_el1;
@@ -72,6 +75,7 @@ struct cpu_features
   bool bti;
   /* Currently, the GLIBC memory tagging tunable only defines 8 bits.  */
   uint8_t mte_state;
+  bool sve;
 };
 
 #endif /* _CPU_FEATURES_AARCH64_H  */