This is the mail archive of the
libc-alpha@sourceware.org
mailing list for the glibc project.
Re: [PATCHv2] powerpc: POWER8 memcpy optimization for cached memory
- From: Rajalakshmi Srinivasaraghavan <raji at linux dot vnet dot ibm dot com>
- To: libc-alpha at sourceware dot org
- Date: Sun, 10 Dec 2017 12:38:24 +0530
- Subject: Re: [PATCHv2] powerpc: POWER8 memcpy optimization for cached memory
- Authentication-results: sourceware.org; auth=none
- References: <87vaik8uxy.fsf@linux.vnet.ibm.com> <20171208194020.5005-1-tuliom@linux.vnet.ibm.com>
On 12/09/2017 01:10 AM, Tulio Magno Quites Machado Filho wrote:
From: Adhemerval Zanella<azanella@linux.vnet.ibm.com>
I made the changes I requested, updated copyright entries, added a
manual entry and fixed a build issue on powerpc64.
--- 8< ---
On POWER8, unaligned memory accesses to cached memory has little impact
on performance as opposed to its ancestors.
It is disabled by default and will only be available when the tunable
glibc.tune.cached_memopt is set to 1.
__memcpy_power8_cached __memcpy_power7
============================================================
max-size=4096: 33325.70 ( 12.65%) 38153.00
max-size=8192: 32878.20 ( 11.17%) 37012.30
max-size=16384: 33782.20 ( 11.61%) 38219.20
max-size=32768: 33296.20 ( 11.30%) 37538.30
max-size=65536: 33765.60 ( 10.53%) 37738.40
2017-12-08 Adhemerval Zanella<azanella@linux.vnet.ibm.com>
Tulio Magno Quites Machado Filho<tuliom@linux.vnet.ibm.com>
* manual/tunables.texi (Hardware Capability Tunables): Document
glibc.tune.cached_memopt.
* sysdeps/powerpc/cpu-features.c: New file.
* sysdeps/powerpc/cpu-features.h: New file.
* sysdeps/powerpc/dl-procinfo.c [!IS_IN(ldconfig)]: Add
_dl_powerpc_cpu_features.
* sysdeps/powerpc/dl-tunables.list: New file.
* sysdeps/powerpc/ldsodefs.h: Include cpu-features.h.
* sysdeps/powerpc/powerpc32/power4/multiarch/init-arch.h: .
Comment missing.
* sysdeps/powerpc/powerpc64/dl-machine.h (INIT_ARCH): Initialize
use_aligned_memopt.
Should this be moved to init-arch.h? (also use_cached_memopt)
* sysdeps/powerpc/powerpc64/multiarch/Makefile (sysdep_routines):
Add memcpy-power8-cached.
* sysdeps/powerpc/powerpc64/multiarch/ifunc-impl-list.c: Add
__memcpy_power8_cached.
* sysdeps/powerpc/powerpc64/multiarch/memcpy.c: Likewise.
* sysdeps/powerpc/powerpc64/multiarch/memcpy-power8-cached.S:
New file.
---
diff --git a/sysdeps/powerpc/powerpc64/multiarch/memcpy-power8-cached.S b/sysdeps/powerpc/powerpc64/multiarch/memcpy-power8-cached.S
new file mode 100644
index 0000000..e5b6f25
--- /dev/null
+++ b/sysdeps/powerpc/powerpc64/multiarch/memcpy-power8-cached.S
@@ -0,0 +1,179 @@
+ stxvd2x v0,r0,r3
+L(dst_is_align_16):
+ cmpldi cr7,r5,127
+ ble cr7,L(tail_copy)
+ addi r8,r5,-128
+ mr r9,r12
+ rldicr r8,r8,0,56
+ li r11,16
+ srdi r10,r8,7
+ addi r0,r8,128
+ addi r10,r10,1
Can we directly do
rldicr r0, r5, 0, 56
srdi r10,r5,7
instead of this sequence?
79 addi r8,r5,-128
81 rldicr r8,r8,0,56
83 srdi r10,r8,7
84 addi r0,r8,128
85 addi r10,r10,1
+ li r6,32
+ mtctr r10
+ li r7,48
+
+ /* Main loop, copy 128 bytes each time. */
LGTM.
Reviewed-by: Rajalakshmi Srinivasaraghavan <raji@linux.vnet.ibm.com>
--
Thanks
Rajalakshmi S