erasure_code: add AVX2 5vect mad with GFNI implementation

Signed-off-by: Marcel Cornu <marcel.d.cornu@intel.com>
This commit is contained in:
Marcel Cornu 2023-12-13 14:51:10 +00:00 committed by Pablo de Lara
parent 47ed2847af
commit a53a20ea2a
4 changed files with 277 additions and 5 deletions

View File

@ -89,6 +89,7 @@ objs = \
bin\gf_2vect_mad_avx2_gfni.obj \ bin\gf_2vect_mad_avx2_gfni.obj \
bin\gf_3vect_mad_avx2_gfni.obj \ bin\gf_3vect_mad_avx2_gfni.obj \
bin\gf_4vect_mad_avx2_gfni.obj \ bin\gf_4vect_mad_avx2_gfni.obj \
bin\gf_5vect_mad_avx2_gfni.obj \
bin\gf_vect_dot_prod_avx512.obj \ bin\gf_vect_dot_prod_avx512.obj \
bin\gf_2vect_dot_prod_avx512.obj \ bin\gf_2vect_dot_prod_avx512.obj \
bin\gf_3vect_dot_prod_avx512.obj \ bin\gf_3vect_dot_prod_avx512.obj \

View File

@ -82,6 +82,7 @@ lsrc_x86_64 += \
erasure_code/gf_2vect_mad_avx2_gfni.asm \ erasure_code/gf_2vect_mad_avx2_gfni.asm \
erasure_code/gf_3vect_mad_avx2_gfni.asm \ erasure_code/gf_3vect_mad_avx2_gfni.asm \
erasure_code/gf_4vect_mad_avx2_gfni.asm \ erasure_code/gf_4vect_mad_avx2_gfni.asm \
erasure_code/gf_5vect_mad_avx2_gfni.asm \
erasure_code/gf_vect_dot_prod_avx512.asm \ erasure_code/gf_vect_dot_prod_avx512.asm \
erasure_code/gf_2vect_dot_prod_avx512.asm \ erasure_code/gf_2vect_dot_prod_avx512.asm \
erasure_code/gf_3vect_dot_prod_avx512.asm \ erasure_code/gf_3vect_dot_prod_avx512.asm \

View File

@ -281,6 +281,8 @@ extern void gf_3vect_mad_avx2_gfni(int len, int vec, int vec_i, unsigned char *g
unsigned char *src, unsigned char **dest); unsigned char *src, unsigned char **dest);
extern void gf_4vect_mad_avx2_gfni(int len, int vec, int vec_i, unsigned char *gftbls, extern void gf_4vect_mad_avx2_gfni(int len, int vec, int vec_i, unsigned char *gftbls,
unsigned char *src, unsigned char **dest); unsigned char *src, unsigned char **dest);
extern void gf_5vect_mad_avx2_gfni(int len, int vec, int vec_i, unsigned char *gftbls,
unsigned char *src, unsigned char **dest);
void ec_init_tables_gfni(int k, int rows, unsigned char *a, unsigned char *g_tbls) void ec_init_tables_gfni(int k, int rows, unsigned char *a, unsigned char *g_tbls)
{ {
@ -384,13 +386,16 @@ void ec_encode_data_update_avx2_gfni(int len, int k, int rows, int vec_i,
unsigned char *g_tbls, unsigned char *data, unsigned char *g_tbls, unsigned char *data,
unsigned char **coding) unsigned char **coding)
{ {
while (rows >= 4) { while (rows >= 5) {
gf_4vect_mad_avx2_gfni(len, k, vec_i, g_tbls, data, coding); gf_5vect_mad_avx2_gfni(len, k, vec_i, g_tbls, data, coding);
g_tbls += 4 * k * 8; g_tbls += 5 * k * 8;
coding += 4; coding += 5;
rows -= 4; rows -= 5;
} }
switch (rows) { switch (rows) {
case 4:
gf_4vect_mad_avx2_gfni(len, k, vec_i, g_tbls, data, coding);
break;
case 3: case 3:
gf_3vect_mad_avx2_gfni(len, k, vec_i, g_tbls, data, coding); gf_3vect_mad_avx2_gfni(len, k, vec_i, g_tbls, data, coding);
break; break;

View File

@ -0,0 +1,265 @@
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
; Copyright(c) 2023 Intel Corporation All rights reserved.
;
; Redistribution and use in source and binary forms, with or without
; modification, are permitted provided that the following conditions
; are met:
; * Redistributions of source code must retain the above copyright
; notice, this list of conditions and the following disclaimer.
; * Redistributions in binary form must reproduce the above copyright
; notice, this list of conditions and the following disclaimer in
; the documentation and/or other materials provided with the
; distribution.
; * Neither the name of Intel Corporation nor the names of its
; contributors may be used to endorse or promote products derived
; from this software without specific prior written permission.
;
; THIS SOFTWARE IS PROVIDED BY THE COPYRIGHT HOLDERS AND CONTRIBUTORS
; "AS IS" AND ANY EXPRESS OR IMPLIED WARRANTIES, INCLUDING, BUT NOT
; LIMITED TO, THE IMPLIED WARRANTIES OF MERCHANTABILITY AND FITNESS FOR
; A PARTICULAR PURPOSE ARE DISCLAIMED. IN NO EVENT SHALL THE COPYRIGHT
; OWNER OR CONTRIBUTORS BE LIABLE FOR ANY DIRECT, INDIRECT, INCIDENTAL,
; SPECIAL, EXEMPLARY, OR CONSEQUENTIAL DAMAGES (INCLUDING, BUT NOT
; LIMITED TO, PROCUREMENT OF SUBSTITUTE GOODS OR SERVICES; LOSS OF USE,
; DATA, OR PROFITS; OR BUSINESS INTERRUPTION) HOWEVER CAUSED AND ON ANY
; THEORY OF LIABILITY, WHETHER IN CONTRACT, STRICT LIABILITY, OR TORT
; (INCLUDING NEGLIGENCE OR OTHERWISE) ARISING IN ANY WAY OUT OF THE USE
; OF THIS SOFTWARE, EVEN IF ADVISED OF THE POSSIBILITY OF SUCH DAMAGE.
;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;;
;;;
;;; gf_5vect_mad_avx2_gfni(len, vec, vec_i, mul_array, src, dest);
;;;
%include "reg_sizes.asm"
%include "gf_vect_gfni.inc"
%include "memcpy.asm"
%if AS_FEATURE_LEVEL >= 10
%ifidn __OUTPUT_FORMAT__, elf64
%define arg0 rdi
%define arg1 rsi
%define arg2 rdx
%define arg3 rcx
%define arg4 r8
%define arg5 r9
%define tmp r11
%define tmp2 r10
%define tmp3 r12
%define tmp4 r13
%define func(x) x: endbranch
%define stack_size 2*8
%macro FUNC_SAVE 0
sub rsp, stack_size
mov [rsp + 0*8], r12
mov [rsp + 1*8], r13
%endmacro
%macro FUNC_RESTORE 0
mov r12, [rsp + 0*8]
mov r13, [rsp + 1*8]
add rsp, stack_size
%endmacro
%endif
%ifidn __OUTPUT_FORMAT__, win64
%define arg0 rcx
%define arg1 rdx
%define arg2 r8
%define arg3 r9
%define arg4 r12 ; must be saved, loaded and restored
%define arg5 r13 ; must be saved and restored
%define tmp r11
%define tmp2 r10
%define tmp3 r14
%define tmp4 r15
%define stack_size 16*10 + 5*8
%define arg(x) [rsp + stack_size + 8 + 8*x]
%define func(x) proc_frame x
%macro FUNC_SAVE 0
sub rsp, stack_size
vmovdqa [rsp + 0*16], xmm6
vmovdqa [rsp + 1*16], xmm7
vmovdqa [rsp + 2*16], xmm8
vmovdqa [rsp + 3*16], xmm9
vmovdqa [rsp + 4*16], xmm10
vmovdqa [rsp + 5*16], xmm11
vmovdqa [rsp + 6*16], xmm12
vmovdqa [rsp + 7*16], xmm13
vmovdqa [rsp + 8*16], xmm14
vmovdqa [rsp + 9*16], xmm15
mov [rsp + 10*16 + 0*8], r12
mov [rsp + 10*16 + 1*8], r13
mov [rsp + 10*16 + 2*8], r14
mov [rsp + 10*16 + 3*8], r15
end_prolog
mov arg4, arg(4)
mov arg5, arg(5)
%endmacro
%macro FUNC_RESTORE 0
vmovdqa xmm6, [rsp + 0*16]
vmovdqa xmm7, [rsp + 1*16]
vmovdqa xmm8, [rsp + 2*16]
vmovdqa xmm9, [rsp + 3*16]
vmovdqa xmm10, [rsp + 4*16]
vmovdqa xmm11, [rsp + 5*16]
vmovdqa xmm12, [rsp + 6*16]
vmovdqa xmm13, [rsp + 7*16]
vmovdqa xmm14, [rsp + 8*16]
vmovdqa xmm15, [rsp + 9*16]
mov r12, [rsp + 10*16 + 0*8]
mov r13, [rsp + 10*16 + 1*8]
mov r14, [rsp + 10*16 + 2*8]
mov r15, [rsp + 10*16 + 3*8]
add rsp, stack_size
%endmacro
%endif
%define len arg0
%define vec arg1
%define vec_i arg2
%define mul_array arg3
%define src arg4
%define dest1 arg5
%define pos rax
%define dest2 mul_array
%define dest3 vec_i
%define dest4 tmp3
%define dest5 tmp4
%ifndef EC_ALIGNED_ADDR
;;; Use Un-aligned load/store
%define XLDR vmovdqu
%define XSTR vmovdqu
%else
;;; Use Non-temporal load/stor
%ifdef NO_NT_LDST
%define XLDR vmovdqa
%define XSTR vmovdqa
%else
%define XLDR vmovntdqa
%define XSTR vmovntdq
%endif
%endif
default rel
[bits 64]
section .text
%define x0 ymm0
%define xd1 ymm1
%define xd2 ymm2
%define xd3 ymm3
%define xd4 ymm4
%define xd5 ymm5
%define xgft1 ymm6
%define xgft2 ymm7
%define xgft3 ymm8
%define xgft4 ymm9
%define xgft5 ymm10
%define xret1 ymm11
%define xret2 ymm12
%define xret3 ymm13
%define xret4 ymm14
%define xret5 ymm15
;;
;; Encodes 32 bytes of a single source into 5x 32 bytes (parity disks)
;;
%macro ENCODE_32B_5 0
;; get next source vector
XLDR x0, [src + pos]
;; get next dest vectors
XLDR xd1, [dest1 + pos]
XLDR xd2, [dest2 + pos]
XLDR xd3, [dest3 + pos]
XLDR xd4, [dest4 + pos]
XLDR xd5, [dest5 + pos]
GF_MUL_XOR VEX, x0, xgft1, xret1, xd1, xgft2, xret2, xd2, \
xgft3, xret3, xd3, xgft4, xret4, xd4, xgft5, xret5, xd5
XSTR [dest1 + pos], xd1
XSTR [dest2 + pos], xd2
XSTR [dest3 + pos], xd3
XSTR [dest4 + pos], xd4
XSTR [dest5 + pos], xd5
%endmacro
;;
;; Encodes less than 32 bytes of a single source into 5x parity disks
;;
%macro ENCODE_LT_32B_5 1
%define %%LEN %1
;; get next source vector
simd_load_avx2 x0, src + pos, %%LEN, tmp, tmp2
;; get next dest vectors
simd_load_avx2 xd1, dest1 + pos, %%LEN, tmp, tmp2
simd_load_avx2 xd2, dest2 + pos, %%LEN, tmp, tmp2
simd_load_avx2 xd3, dest3 + pos, %%LEN, tmp, tmp2
simd_load_avx2 xd4, dest4 + pos, %%LEN, tmp, tmp2
simd_load_avx2 xd5, dest5 + pos, %%LEN, tmp, tmp2
GF_MUL_XOR VEX, x0, xgft1, xret1, xd1, xgft2, xret2, xd2, \
xgft3, xret3, xd3, xgft4, xret4, xd4, xgft5, xret5, xd5
lea dest1, [dest1 + pos]
simd_store_avx2 dest1, xd1, %%LEN, tmp, tmp2
lea dest2, [dest2 + pos]
simd_store_avx2 dest2, xd2, %%LEN, tmp, tmp2
lea dest3, [dest3 + pos]
simd_store_avx2 dest3, xd3, %%LEN, tmp, tmp2
lea dest4, [dest4 + pos]
simd_store_avx2 dest4, xd4, %%LEN, tmp, tmp2
lea dest5, [dest5 + pos]
simd_store_avx2 dest5, xd5, %%LEN, tmp, tmp2
%endmacro
align 16
mk_global gf_5vect_mad_avx2_gfni, function
func(gf_5vect_mad_avx2_gfni)
FUNC_SAVE
xor pos, pos
shl vec_i, 3 ;Multiply by 8
shl vec, 3 ;Multiply by 8
lea tmp, [mul_array + vec_i]
lea tmp2, [vec*3]
vbroadcastsd xgft1, [tmp]
vbroadcastsd xgft2, [tmp + vec]
vbroadcastsd xgft3, [tmp + vec*2]
vbroadcastsd xgft4, [tmp + tmp2]
vbroadcastsd xgft5, [tmp + vec*4]
mov dest2, [dest1 + 1*8] ; reuse mul_array
mov dest3, [dest1 + 2*8] ; reuse vec_i
mov dest4, [dest1 + 3*8]
mov dest5, [dest1 + 4*8]
mov dest1, [dest1]
cmp len, 32
jb .len_lt_32
.loop32:
ENCODE_32B_5 ;; loop on 32 bytes at a time
add pos, 32
sub len, 32
cmp len, 32
jge .loop32
.len_lt_32:
cmp len, 0
jle .exit
ENCODE_LT_32B_5 len ;; encode final bytes
.exit:
vzeroupper
FUNC_RESTORE
ret
endproc_frame
%endif ; if AS_FEATURE_LEVEL >= 10