ABD raidz avx512f support

Implement shift based multiplication for 512f. Higher IPC over lookup based
methods yields up to 40% better performance on the current hardware.

Results on Xeon Phi(TM) CPU 7210:
implementation   gen_p           gen_pq          gen_pqr         rec_p           rec_q           rec_r           rec_pq          rec_pr          rec_qr          rec_pqr
original         142232671       24411492        12948205        283053705       22348167        4215911         9171609         2265548         2378370         1648495
scalar           295711162       49851491        33253815        293198109       88179448        61866752        27941684        25764416        17384442        12138153
sse2             410055998       199642658       117973654       406240463       152688682       121092250       84968180        79291076        47473657        20779719
ssse3            411641595       199669571       117937647       406211024       137638508       117050346       81263322        76120405        46281559        32696722
avx2             616485806       311515332       188595628       605455115       260602390       230554476       148198817       138800254       92273356        62937819
avx512f          832191523       408509425       253599522       810094481       404325734       317590971       218235687       197204920       133101937       94001219
fastest          avx512f         avx512f         avx512f         avx512f         avx512f         avx512f         avx512f         avx512f         avx512f         avx512f

Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
This commit is contained in:
Gvozden Neskovic 2016-11-20 06:01:31 +01:00 committed by Brian Behlendorf
parent cbf484f8ad
commit 65d71d4212
7 changed files with 252 additions and 399 deletions

View File

@ -235,19 +235,6 @@ init_rand(void *data, size_t size, void *private)
return (0);
}
static int
corrupt_rand(void *data, size_t size, void *private)
{
int i;
int *dst = (int *) data;
for (i = 0; i < size / sizeof (int); i++)
dst[i] = rand();
return (0);
}
static void
corrupt_colums(raidz_map_t *rm, const int *tgts, const int cnt)
{
@ -256,8 +243,7 @@ corrupt_colums(raidz_map_t *rm, const int *tgts, const int cnt)
for (i = 0; i < cnt; i++) {
col = &rm->rm_col[tgts[i]];
abd_iterate_func(col->rc_abd, 0, col->rc_size, corrupt_rand,
NULL);
abd_iterate_func(col->rc_abd, 0, col->rc_size, init_rand, NULL);
}
}

View File

@ -58,7 +58,7 @@ const raidz_impl_ops_t *raidz_all_maths[] = {
&vdev_raidz_avx2_impl,
#endif
#if defined(__x86_64) && defined(HAVE_AVX512F) /* only x86_64 for now */
// &vdev_raidz_avx512f_impl,
&vdev_raidz_avx512f_impl,
#endif
#if defined(__x86_64) && defined(HAVE_AVX512BW) /* only x86_64 for now */
// &vdev_raidz_avx512bw_impl,

View File

@ -65,19 +65,6 @@ typedef struct v {
uint8_t b[ELEM_SIZE] __attribute__((aligned(ELEM_SIZE)));
} v_t;
#define PREFETCHNTA(ptr, offset) \
{ \
__asm( \
"prefetchnta " #offset "(%[MEM])\n" \
: : [MEM] "r" (ptr)); \
}
#define PREFETCH(ptr, offset) \
{ \
__asm( \
"prefetcht0 " #offset "(%[MEM])\n" \
: : [MEM] "r" (ptr)); \
}
#define XOR_ACC(src, r...) \
{ \
@ -121,25 +108,7 @@ typedef struct v {
} \
}
#define ZERO(r...) \
{ \
switch (REG_CNT(r)) { \
case 4: \
__asm( \
"vpxor %" VR0(r) ", %" VR0(r)", %" VR0(r) "\n" \
"vpxor %" VR1(r) ", %" VR1(r)", %" VR1(r) "\n" \
"vpxor %" VR2(r) ", %" VR2(r)", %" VR2(r) "\n" \
"vpxor %" VR3(r) ", %" VR3(r)", %" VR3(r)); \
break; \
case 2: \
__asm( \
"vpxor %" VR0(r) ", %" VR0(r)", %" VR0(r) "\n" \
"vpxor %" VR1(r) ", %" VR1(r)", %" VR1(r)); \
break; \
default: \
ASM_BUG(); \
} \
}
#define ZERO(r...) XOR(r, r)
#define COPY(r...) \
{ \

View File

@ -20,11 +20,12 @@
*/
/*
* Copyright (C) 2016 Romain Dolbeau. All rights reserved.
* Copyright (C) 2016 Gvozden Nešković. All rights reserved.
*/
#include <sys/isa_defs.h>
#if 0 // defined(__x86_64) && defined(HAVE_AVX512F)
#if defined(__x86_64) && defined(HAVE_AVX512F)
#include <sys/types.h>
#include <linux/simd_x86.h>
@ -74,29 +75,12 @@
#define _R_23(_0, _1, REG2, REG3, ...) REG2, REG3
#define R_23(REG...) _R_23(REG, 1, 2, 3)
#define ASM_BUG() ASSERT(0)
extern const uint8_t gf_clmul_mod_lt[4*256][16];
#define ELEM_SIZE 64
typedef struct v {
uint8_t b[ELEM_SIZE] __attribute__((aligned(ELEM_SIZE)));
} v_t;
#define PREFETCHNTA(ptr, offset) \
{ \
__asm( \
"prefetchnta " #offset "(%[MEM])\n" \
: : [MEM] "r" (ptr)); \
}
#define PREFETCH(ptr, offset) \
{ \
__asm( \
"prefetcht0 " #offset "(%[MEM])\n" \
: : [MEM] "r" (ptr)); \
}
#define XOR_ACC(src, r...) \
{ \
@ -109,14 +93,6 @@ typedef struct v {
"vpxorq 0xc0(%[SRC]), %%" VR3(r)", %%" VR3(r) "\n" \
: : [SRC] "r" (src)); \
break; \
case 2: \
__asm( \
"vpxorq 0x00(%[SRC]), %%" VR0(r)", %%" VR0(r) "\n" \
"vpxorq 0x40(%[SRC]), %%" VR1(r)", %%" VR1(r) "\n" \
: : [SRC] "r" (src)); \
break; \
default: \
ASM_BUG(); \
} \
}
@ -135,30 +111,12 @@ typedef struct v {
"vpxorq %" VR0(r) ", %" VR2(r)", %" VR2(r) "\n" \
"vpxorq %" VR1(r) ", %" VR3(r)", %" VR3(r)); \
break; \
default: \
ASM_BUG(); \
} \
}
#define ZERO(r...) \
{ \
switch (REG_CNT(r)) { \
case 4: \
__asm( \
"vpxorq %" VR0(r) ", %" VR0(r)", %" VR0(r) "\n" \
"vpxorq %" VR1(r) ", %" VR1(r)", %" VR1(r) "\n" \
"vpxorq %" VR2(r) ", %" VR2(r)", %" VR2(r) "\n" \
"vpxorq %" VR3(r) ", %" VR3(r)", %" VR3(r)); \
break; \
case 2: \
__asm( \
"vpxorq %" VR0(r) ", %" VR0(r)", %" VR0(r) "\n" \
"vpxorq %" VR1(r) ", %" VR1(r)", %" VR1(r)); \
break; \
default: \
ASM_BUG(); \
} \
}
#define ZERO(r...) XOR(r, r)
#define COPY(r...) \
{ \
@ -175,8 +133,6 @@ typedef struct v {
"vmovdqa64 %" VR0(r) ", %" VR2(r) "\n" \
"vmovdqa64 %" VR1(r) ", %" VR3(r)); \
break; \
default: \
ASM_BUG(); \
} \
}
@ -191,14 +147,6 @@ typedef struct v {
"vmovdqa64 0xc0(%[SRC]), %%" VR3(r) "\n" \
: : [SRC] "r" (src)); \
break; \
case 2: \
__asm( \
"vmovdqa64 0x00(%[SRC]), %%" VR0(r) "\n" \
"vmovdqa64 0x40(%[SRC]), %%" VR1(r) "\n" \
: : [SRC] "r" (src)); \
break; \
default: \
ASM_BUG(); \
} \
}
@ -213,31 +161,17 @@ typedef struct v {
"vmovdqa64 %%" VR3(r) ", 0xc0(%[DST])\n" \
: : [DST] "r" (dst)); \
break; \
case 2: \
__asm( \
"vmovdqa64 %%" VR0(r) ", 0x00(%[DST])\n" \
"vmovdqa64 %%" VR1(r) ", 0x40(%[DST])\n" \
: : [DST] "r" (dst)); \
break; \
default: \
ASM_BUG(); \
} \
}
#define FLUSH() \
{ \
__asm("vzeroupper"); \
}
#define MUL2_SETUP() \
{ \
__asm("vmovq %0, %%xmm14" :: "r"(0x1d1d1d1d1d1d1d1d)); \
__asm("vpbroadcastq %xmm14, %zmm14"); \
__asm("vmovq %0, %%xmm13" :: "r"(0x8080808080808080)); \
__asm("vpbroadcastq %xmm13, %zmm13"); \
__asm("vmovq %0, %%xmm12" :: "r"(0xfefefefefefefefe)); \
__asm("vpbroadcastq %xmm12, %zmm12"); \
__asm("vpxorq %zmm0, %zmm0 ,%zmm0"); \
__asm("vmovq %0, %%xmm31" :: "r"(0x1d1d1d1d1d1d1d1d)); \
__asm("vpbroadcastq %xmm31, %zmm31"); \
__asm("vmovq %0, %%xmm30" :: "r"(0x8080808080808080)); \
__asm("vpbroadcastq %xmm30, %zmm30"); \
__asm("vmovq %0, %%xmm29" :: "r"(0xfefefefefefefefe)); \
__asm("vpbroadcastq %xmm29, %zmm29"); \
}
#define _MUL2(r...) \
@ -245,23 +179,21 @@ typedef struct v {
switch (REG_CNT(r)) { \
case 2: \
__asm( \
"vpandq %" VR0(r)", %zmm13, %zmm10\n" \
"vpandq %" VR1(r)", %zmm13, %zmm11\n" \
"vpsrlq $7, %zmm10, %zmm30\n" \
"vpsrlq $7, %zmm11, %zmm31\n" \
"vpsllq $1, %zmm10, %zmm10\n" \
"vpsllq $1, %zmm11, %zmm11\n" \
"vpsubq %zmm30, %zmm10, %zmm10\n" \
"vpsubq %zmm31, %zmm11, %zmm11\n" \
"vpandq %" VR0(r)", %zmm30, %zmm26\n" \
"vpandq %" VR1(r)", %zmm30, %zmm25\n" \
"vpsrlq $7, %zmm26, %zmm28\n" \
"vpsrlq $7, %zmm25, %zmm27\n" \
"vpsllq $1, %zmm26, %zmm26\n" \
"vpsllq $1, %zmm25, %zmm25\n" \
"vpsubq %zmm28, %zmm26, %zmm26\n" \
"vpsubq %zmm27, %zmm25, %zmm25\n" \
"vpsllq $1, %" VR0(r)", %" VR0(r) "\n" \
"vpsllq $1, %" VR1(r)", %" VR1(r) "\n" \
"vpandq %zmm10, %zmm14, %zmm10\n" \
"vpandq %zmm11, %zmm14, %zmm11\n" \
"vpternlogd $0x6c,%zmm12, %zmm10, %" VR0(r) "\n" \
"vpternlogd $0x6c,%zmm12, %zmm11, %" VR1(r)); \
"vpandq %zmm26, %zmm31, %zmm26\n" \
"vpandq %zmm25, %zmm31, %zmm25\n" \
"vpternlogd $0x6c,%zmm29, %zmm26, %" VR0(r) "\n" \
"vpternlogd $0x6c,%zmm29, %zmm25, %" VR1(r)); \
break; \
default: \
ASM_BUG(); \
} \
}
@ -275,8 +207,6 @@ typedef struct v {
case 2: \
_MUL2(r); \
break; \
default: \
ASM_BUG(); \
} \
}
@ -286,231 +216,249 @@ typedef struct v {
MUL2(r); \
}
/*
* Must match the init above
*/
#define _0f "zmm0"
#define _as "zmm14"
#define _bs "zmm13"
#define _ltmod "zmm12"
#define _ltmul "zmm11"
#define _ta "zmm10"
#define _tb "zmm15"
/*
* Must be in the first 16, otherwise an EVEX pshufb is generated
* Must match above
*/
#define _asYlo "ymm14"
#define _bsYlo "ymm13"
#define _ltmodYlo "ymm12"
#define _ltmulYlo "ymm11"
#define _taYlo "ymm10"
#define _tbYlo "ymm15"
/* General multiplication by adding powers of two */
/*
* Must be in the first 16, otherwise an EVEX pshufb is generated
* ...
*/
#define _asYhi "ymm9"
#define _bsYhi "ymm8"
#define _ltmodYhi "ymm7"
#define _ltmulYhi "ymm6"
#define _taYhi "ymm5"
#define _tbYhi "ymm4"
#define _mul_x2_in 21, 22
#define _mul_x2_acc 23, 24
/*
* This uses a pair of AVX2 pshufb to emulate the missing AVX512 pshufb.
* AVX512BW has the full pshufb
* To get VEX pshufb (AVX2, supported in KNL) instead of EVEX pshufb
* (AVX512BW, not supported on KNL, probably also requiring AVX51VL
* since we use a 256 bits version), all registers in parameters to
* pshufb must be among ymm0-ymm15, since only EVEX can encore
* ymm16-ymm31
* This is a bit hackish, but short of encoding the instruction in
* binary, how do we force the use of AVX2 pshufb ?
* Note that the other way round (forcing AVX512) is easy, just encode
* k0 as the mask register (k0 is all-1).
*/
#define _MULx2(c, r...) \
#define _MUL_PARAM(x, in, acc) \
{ \
switch (REG_CNT(r)) { \
case 2: \
__asm( \
"vmovq %[c0f], %%xmm0\n" \
"vpbroadcastq %%xmm0, %%" _0f "\n" \
/* upper bits */ \
"vbroadcasti32x4 0x00(%[lt]), %%" _ltmod "\n" \
"vbroadcasti32x4 0x10(%[lt]), %%" _ltmul "\n" \
\
"vpsrad $0x4, %%" VR0(r) ", %%"_as "\n" \
"vpsrad $0x4, %%" VR1(r) ", %%"_bs "\n" \
"vpandq %%" _0f ", %%" VR0(r) ", %%" VR0(r) "\n" \
"vpandq %%" _0f ", %%" VR1(r) ", %%" VR1(r) "\n" \
"vpandq %%" _0f ", %%" _as ", %%" _as "\n" \
"vpandq %%" _0f ", %%" _bs ", %%" _bs "\n" \
\
"vextracti64x4 $1,%%" _ltmod ",%%" _ltmodYhi"\n" \
\
"vextracti64x4 $1,%%" _as ",%%" _asYhi"\n" \
"vpshufb %%" _asYlo ", %%" _ltmodYlo ", %%" _taYlo "\n" \
"vpshufb %%" _asYhi ", %%" _ltmodYhi ", %%" _taYhi "\n" \
"vinserti64x4 $1,%%" _taYhi ",%%" _ta ",%%" _ta "\n" \
\
"vextracti64x4 $1,%%" _bs ",%%" _bsYhi"\n" \
"vpshufb %%" _bsYlo ", %%" _ltmodYlo ", %%" _tbYlo "\n" \
"vpshufb %%" _bsYhi ", %%" _ltmodYhi ", %%" _tbYhi "\n" \
"vinserti64x4 $1,%%" _tbYhi ",%%" _tb ",%%" _tb "\n" \
\
"vextracti64x4 $1,%%" _ltmul ",%%" _ltmulYhi"\n" \
\
"vpshufb %%" _asYlo ", %%" _ltmulYlo ", %%" _asYlo "\n" \
"vpshufb %%" _asYhi ", %%" _ltmulYhi ", %%" _asYhi "\n" \
"vinserti64x4 $1,%%" _asYhi ",%%" _as ",%%" _as "\n" \
\
"vpshufb %%" _bsYlo ", %%" _ltmulYlo ", %%" _bsYlo "\n" \
"vpshufb %%" _bsYhi ", %%" _ltmulYhi ", %%" _bsYhi "\n" \
"vinserti64x4 $1,%%" _bsYhi ",%%" _bs ",%%" _bs "\n" \
\
/* lower bits */ \
"vbroadcasti32x4 0x20(%[lt]), %%" _ltmod "\n" \
"vbroadcasti32x4 0x30(%[lt]), %%" _ltmul "\n" \
\
"vpxorq %%" _ta ", %%" _as ", %%" _as "\n" \
"vpxorq %%" _tb ", %%" _bs ", %%" _bs "\n" \
\
"vextracti64x4 $1,%%" _ltmod ",%%" _ltmodYhi"\n" \
\
"vextracti64x4 $0,%%" VR0(r) ",%%" "ymm1" "\n" \
"vextracti64x4 $1,%%" VR0(r) ",%%" _asYhi"\n" \
"vpshufb %%" "ymm1" ", %%" _ltmodYlo ", %%" _taYlo "\n" \
"vpshufb %%" _asYhi ", %%" _ltmodYhi ", %%" _taYhi "\n" \
"vinserti64x4 $1,%%" _taYhi ",%%" _ta ",%%" _ta "\n" \
\
"vextracti64x4 $0,%%" VR1(r) ",%%" "ymm2" "\n" \
"vextracti64x4 $1,%%" VR1(r) ",%%" _bsYhi"\n" \
"vpshufb %%" "ymm2" ", %%" _ltmodYlo ", %%" _tbYlo "\n" \
"vpshufb %%" _bsYhi ", %%" _ltmodYhi ", %%" _tbYhi "\n" \
"vinserti64x4 $1,%%" _tbYhi ",%%" _tb ",%%" _tb "\n" \
\
"vextracti64x4 $1,%%" _ltmul ",%%" _ltmulYhi"\n" \
\
"vpshufb %%" "ymm1" ", %%" _ltmulYlo ", %%" "ymm1" "\n" \
"vpshufb %%" _asYhi ", %%" _ltmulYhi ", %%" _asYhi "\n" \
"vinserti64x4 $1,%%" _asYhi ",%%" "zmm1" ",%%" VR0(r) "\n" \
\
"vpshufb %%" "ymm2" ", %%" _ltmulYlo ", %%" "ymm2" "\n" \
"vpshufb %%" _bsYhi ", %%" _ltmulYhi ", %%" _bsYhi "\n" \
"vinserti64x4 $1,%%" _bsYhi ",%%" "zmm2" ",%%" VR1(r) "\n" \
\
"vpxorq %%" _ta ", %%" VR0(r) ", %%" VR0(r) "\n" \
"vpxorq %%" _as ", %%" VR0(r) ", %%" VR0(r) "\n" \
"vpxorq %%" _tb ", %%" VR1(r) ", %%" VR1(r) "\n" \
"vpxorq %%" _bs ", %%" VR1(r) ", %%" VR1(r) "\n" \
: : [c0f] "r" (0x0f0f0f0f0f0f0f0f), \
[lt] "r" (gf_clmul_mod_lt[4*(c)])); \
break; \
default: \
ASM_BUG(); \
} \
if (x & 0x01) { COPY(in, acc); } else { ZERO(acc); } \
if (x & 0xfe) { MUL2(in); } \
if (x & 0x02) { XOR(in, acc); } \
if (x & 0xfc) { MUL2(in); } \
if (x & 0x04) { XOR(in, acc); } \
if (x & 0xf8) { MUL2(in); } \
if (x & 0x08) { XOR(in, acc); } \
if (x & 0xf0) { MUL2(in); } \
if (x & 0x10) { XOR(in, acc); } \
if (x & 0xe0) { MUL2(in); } \
if (x & 0x20) { XOR(in, acc); } \
if (x & 0xc0) { MUL2(in); } \
if (x & 0x40) { XOR(in, acc); } \
if (x & 0x80) { MUL2(in); XOR(in, acc); } \
}
#define MUL_x2_DEFINE(x) \
static void \
mul_x2_ ## x(void) { _MUL_PARAM(x, _mul_x2_in, _mul_x2_acc); }
MUL_x2_DEFINE(0); MUL_x2_DEFINE(1); MUL_x2_DEFINE(2); MUL_x2_DEFINE(3);
MUL_x2_DEFINE(4); MUL_x2_DEFINE(5); MUL_x2_DEFINE(6); MUL_x2_DEFINE(7);
MUL_x2_DEFINE(8); MUL_x2_DEFINE(9); MUL_x2_DEFINE(10); MUL_x2_DEFINE(11);
MUL_x2_DEFINE(12); MUL_x2_DEFINE(13); MUL_x2_DEFINE(14); MUL_x2_DEFINE(15);
MUL_x2_DEFINE(16); MUL_x2_DEFINE(17); MUL_x2_DEFINE(18); MUL_x2_DEFINE(19);
MUL_x2_DEFINE(20); MUL_x2_DEFINE(21); MUL_x2_DEFINE(22); MUL_x2_DEFINE(23);
MUL_x2_DEFINE(24); MUL_x2_DEFINE(25); MUL_x2_DEFINE(26); MUL_x2_DEFINE(27);
MUL_x2_DEFINE(28); MUL_x2_DEFINE(29); MUL_x2_DEFINE(30); MUL_x2_DEFINE(31);
MUL_x2_DEFINE(32); MUL_x2_DEFINE(33); MUL_x2_DEFINE(34); MUL_x2_DEFINE(35);
MUL_x2_DEFINE(36); MUL_x2_DEFINE(37); MUL_x2_DEFINE(38); MUL_x2_DEFINE(39);
MUL_x2_DEFINE(40); MUL_x2_DEFINE(41); MUL_x2_DEFINE(42); MUL_x2_DEFINE(43);
MUL_x2_DEFINE(44); MUL_x2_DEFINE(45); MUL_x2_DEFINE(46); MUL_x2_DEFINE(47);
MUL_x2_DEFINE(48); MUL_x2_DEFINE(49); MUL_x2_DEFINE(50); MUL_x2_DEFINE(51);
MUL_x2_DEFINE(52); MUL_x2_DEFINE(53); MUL_x2_DEFINE(54); MUL_x2_DEFINE(55);
MUL_x2_DEFINE(56); MUL_x2_DEFINE(57); MUL_x2_DEFINE(58); MUL_x2_DEFINE(59);
MUL_x2_DEFINE(60); MUL_x2_DEFINE(61); MUL_x2_DEFINE(62); MUL_x2_DEFINE(63);
MUL_x2_DEFINE(64); MUL_x2_DEFINE(65); MUL_x2_DEFINE(66); MUL_x2_DEFINE(67);
MUL_x2_DEFINE(68); MUL_x2_DEFINE(69); MUL_x2_DEFINE(70); MUL_x2_DEFINE(71);
MUL_x2_DEFINE(72); MUL_x2_DEFINE(73); MUL_x2_DEFINE(74); MUL_x2_DEFINE(75);
MUL_x2_DEFINE(76); MUL_x2_DEFINE(77); MUL_x2_DEFINE(78); MUL_x2_DEFINE(79);
MUL_x2_DEFINE(80); MUL_x2_DEFINE(81); MUL_x2_DEFINE(82); MUL_x2_DEFINE(83);
MUL_x2_DEFINE(84); MUL_x2_DEFINE(85); MUL_x2_DEFINE(86); MUL_x2_DEFINE(87);
MUL_x2_DEFINE(88); MUL_x2_DEFINE(89); MUL_x2_DEFINE(90); MUL_x2_DEFINE(91);
MUL_x2_DEFINE(92); MUL_x2_DEFINE(93); MUL_x2_DEFINE(94); MUL_x2_DEFINE(95);
MUL_x2_DEFINE(96); MUL_x2_DEFINE(97); MUL_x2_DEFINE(98); MUL_x2_DEFINE(99);
MUL_x2_DEFINE(100); MUL_x2_DEFINE(101); MUL_x2_DEFINE(102); MUL_x2_DEFINE(103);
MUL_x2_DEFINE(104); MUL_x2_DEFINE(105); MUL_x2_DEFINE(106); MUL_x2_DEFINE(107);
MUL_x2_DEFINE(108); MUL_x2_DEFINE(109); MUL_x2_DEFINE(110); MUL_x2_DEFINE(111);
MUL_x2_DEFINE(112); MUL_x2_DEFINE(113); MUL_x2_DEFINE(114); MUL_x2_DEFINE(115);
MUL_x2_DEFINE(116); MUL_x2_DEFINE(117); MUL_x2_DEFINE(118); MUL_x2_DEFINE(119);
MUL_x2_DEFINE(120); MUL_x2_DEFINE(121); MUL_x2_DEFINE(122); MUL_x2_DEFINE(123);
MUL_x2_DEFINE(124); MUL_x2_DEFINE(125); MUL_x2_DEFINE(126); MUL_x2_DEFINE(127);
MUL_x2_DEFINE(128); MUL_x2_DEFINE(129); MUL_x2_DEFINE(130); MUL_x2_DEFINE(131);
MUL_x2_DEFINE(132); MUL_x2_DEFINE(133); MUL_x2_DEFINE(134); MUL_x2_DEFINE(135);
MUL_x2_DEFINE(136); MUL_x2_DEFINE(137); MUL_x2_DEFINE(138); MUL_x2_DEFINE(139);
MUL_x2_DEFINE(140); MUL_x2_DEFINE(141); MUL_x2_DEFINE(142); MUL_x2_DEFINE(143);
MUL_x2_DEFINE(144); MUL_x2_DEFINE(145); MUL_x2_DEFINE(146); MUL_x2_DEFINE(147);
MUL_x2_DEFINE(148); MUL_x2_DEFINE(149); MUL_x2_DEFINE(150); MUL_x2_DEFINE(151);
MUL_x2_DEFINE(152); MUL_x2_DEFINE(153); MUL_x2_DEFINE(154); MUL_x2_DEFINE(155);
MUL_x2_DEFINE(156); MUL_x2_DEFINE(157); MUL_x2_DEFINE(158); MUL_x2_DEFINE(159);
MUL_x2_DEFINE(160); MUL_x2_DEFINE(161); MUL_x2_DEFINE(162); MUL_x2_DEFINE(163);
MUL_x2_DEFINE(164); MUL_x2_DEFINE(165); MUL_x2_DEFINE(166); MUL_x2_DEFINE(167);
MUL_x2_DEFINE(168); MUL_x2_DEFINE(169); MUL_x2_DEFINE(170); MUL_x2_DEFINE(171);
MUL_x2_DEFINE(172); MUL_x2_DEFINE(173); MUL_x2_DEFINE(174); MUL_x2_DEFINE(175);
MUL_x2_DEFINE(176); MUL_x2_DEFINE(177); MUL_x2_DEFINE(178); MUL_x2_DEFINE(179);
MUL_x2_DEFINE(180); MUL_x2_DEFINE(181); MUL_x2_DEFINE(182); MUL_x2_DEFINE(183);
MUL_x2_DEFINE(184); MUL_x2_DEFINE(185); MUL_x2_DEFINE(186); MUL_x2_DEFINE(187);
MUL_x2_DEFINE(188); MUL_x2_DEFINE(189); MUL_x2_DEFINE(190); MUL_x2_DEFINE(191);
MUL_x2_DEFINE(192); MUL_x2_DEFINE(193); MUL_x2_DEFINE(194); MUL_x2_DEFINE(195);
MUL_x2_DEFINE(196); MUL_x2_DEFINE(197); MUL_x2_DEFINE(198); MUL_x2_DEFINE(199);
MUL_x2_DEFINE(200); MUL_x2_DEFINE(201); MUL_x2_DEFINE(202); MUL_x2_DEFINE(203);
MUL_x2_DEFINE(204); MUL_x2_DEFINE(205); MUL_x2_DEFINE(206); MUL_x2_DEFINE(207);
MUL_x2_DEFINE(208); MUL_x2_DEFINE(209); MUL_x2_DEFINE(210); MUL_x2_DEFINE(211);
MUL_x2_DEFINE(212); MUL_x2_DEFINE(213); MUL_x2_DEFINE(214); MUL_x2_DEFINE(215);
MUL_x2_DEFINE(216); MUL_x2_DEFINE(217); MUL_x2_DEFINE(218); MUL_x2_DEFINE(219);
MUL_x2_DEFINE(220); MUL_x2_DEFINE(221); MUL_x2_DEFINE(222); MUL_x2_DEFINE(223);
MUL_x2_DEFINE(224); MUL_x2_DEFINE(225); MUL_x2_DEFINE(226); MUL_x2_DEFINE(227);
MUL_x2_DEFINE(228); MUL_x2_DEFINE(229); MUL_x2_DEFINE(230); MUL_x2_DEFINE(231);
MUL_x2_DEFINE(232); MUL_x2_DEFINE(233); MUL_x2_DEFINE(234); MUL_x2_DEFINE(235);
MUL_x2_DEFINE(236); MUL_x2_DEFINE(237); MUL_x2_DEFINE(238); MUL_x2_DEFINE(239);
MUL_x2_DEFINE(240); MUL_x2_DEFINE(241); MUL_x2_DEFINE(242); MUL_x2_DEFINE(243);
MUL_x2_DEFINE(244); MUL_x2_DEFINE(245); MUL_x2_DEFINE(246); MUL_x2_DEFINE(247);
MUL_x2_DEFINE(248); MUL_x2_DEFINE(249); MUL_x2_DEFINE(250); MUL_x2_DEFINE(251);
MUL_x2_DEFINE(252); MUL_x2_DEFINE(253); MUL_x2_DEFINE(254); MUL_x2_DEFINE(255);
typedef void (*mul_fn_ptr_t)(void);
static const mul_fn_ptr_t __attribute__((aligned(256)))
gf_x2_mul_fns[256] = {
mul_x2_0, mul_x2_1, mul_x2_2, mul_x2_3, mul_x2_4, mul_x2_5,
mul_x2_6, mul_x2_7, mul_x2_8, mul_x2_9, mul_x2_10, mul_x2_11,
mul_x2_12, mul_x2_13, mul_x2_14, mul_x2_15, mul_x2_16, mul_x2_17,
mul_x2_18, mul_x2_19, mul_x2_20, mul_x2_21, mul_x2_22, mul_x2_23,
mul_x2_24, mul_x2_25, mul_x2_26, mul_x2_27, mul_x2_28, mul_x2_29,
mul_x2_30, mul_x2_31, mul_x2_32, mul_x2_33, mul_x2_34, mul_x2_35,
mul_x2_36, mul_x2_37, mul_x2_38, mul_x2_39, mul_x2_40, mul_x2_41,
mul_x2_42, mul_x2_43, mul_x2_44, mul_x2_45, mul_x2_46, mul_x2_47,
mul_x2_48, mul_x2_49, mul_x2_50, mul_x2_51, mul_x2_52, mul_x2_53,
mul_x2_54, mul_x2_55, mul_x2_56, mul_x2_57, mul_x2_58, mul_x2_59,
mul_x2_60, mul_x2_61, mul_x2_62, mul_x2_63, mul_x2_64, mul_x2_65,
mul_x2_66, mul_x2_67, mul_x2_68, mul_x2_69, mul_x2_70, mul_x2_71,
mul_x2_72, mul_x2_73, mul_x2_74, mul_x2_75, mul_x2_76, mul_x2_77,
mul_x2_78, mul_x2_79, mul_x2_80, mul_x2_81, mul_x2_82, mul_x2_83,
mul_x2_84, mul_x2_85, mul_x2_86, mul_x2_87, mul_x2_88, mul_x2_89,
mul_x2_90, mul_x2_91, mul_x2_92, mul_x2_93, mul_x2_94, mul_x2_95,
mul_x2_96, mul_x2_97, mul_x2_98, mul_x2_99, mul_x2_100, mul_x2_101,
mul_x2_102, mul_x2_103, mul_x2_104, mul_x2_105, mul_x2_106, mul_x2_107,
mul_x2_108, mul_x2_109, mul_x2_110, mul_x2_111, mul_x2_112, mul_x2_113,
mul_x2_114, mul_x2_115, mul_x2_116, mul_x2_117, mul_x2_118, mul_x2_119,
mul_x2_120, mul_x2_121, mul_x2_122, mul_x2_123, mul_x2_124, mul_x2_125,
mul_x2_126, mul_x2_127, mul_x2_128, mul_x2_129, mul_x2_130, mul_x2_131,
mul_x2_132, mul_x2_133, mul_x2_134, mul_x2_135, mul_x2_136, mul_x2_137,
mul_x2_138, mul_x2_139, mul_x2_140, mul_x2_141, mul_x2_142, mul_x2_143,
mul_x2_144, mul_x2_145, mul_x2_146, mul_x2_147, mul_x2_148, mul_x2_149,
mul_x2_150, mul_x2_151, mul_x2_152, mul_x2_153, mul_x2_154, mul_x2_155,
mul_x2_156, mul_x2_157, mul_x2_158, mul_x2_159, mul_x2_160, mul_x2_161,
mul_x2_162, mul_x2_163, mul_x2_164, mul_x2_165, mul_x2_166, mul_x2_167,
mul_x2_168, mul_x2_169, mul_x2_170, mul_x2_171, mul_x2_172, mul_x2_173,
mul_x2_174, mul_x2_175, mul_x2_176, mul_x2_177, mul_x2_178, mul_x2_179,
mul_x2_180, mul_x2_181, mul_x2_182, mul_x2_183, mul_x2_184, mul_x2_185,
mul_x2_186, mul_x2_187, mul_x2_188, mul_x2_189, mul_x2_190, mul_x2_191,
mul_x2_192, mul_x2_193, mul_x2_194, mul_x2_195, mul_x2_196, mul_x2_197,
mul_x2_198, mul_x2_199, mul_x2_200, mul_x2_201, mul_x2_202, mul_x2_203,
mul_x2_204, mul_x2_205, mul_x2_206, mul_x2_207, mul_x2_208, mul_x2_209,
mul_x2_210, mul_x2_211, mul_x2_212, mul_x2_213, mul_x2_214, mul_x2_215,
mul_x2_216, mul_x2_217, mul_x2_218, mul_x2_219, mul_x2_220, mul_x2_221,
mul_x2_222, mul_x2_223, mul_x2_224, mul_x2_225, mul_x2_226, mul_x2_227,
mul_x2_228, mul_x2_229, mul_x2_230, mul_x2_231, mul_x2_232, mul_x2_233,
mul_x2_234, mul_x2_235, mul_x2_236, mul_x2_237, mul_x2_238, mul_x2_239,
mul_x2_240, mul_x2_241, mul_x2_242, mul_x2_243, mul_x2_244, mul_x2_245,
mul_x2_246, mul_x2_247, mul_x2_248, mul_x2_249, mul_x2_250, mul_x2_251,
mul_x2_252, mul_x2_253, mul_x2_254, mul_x2_255
};
#define MUL(c, r...) \
{ \
switch (REG_CNT(r)) { \
case 4: \
_MULx2(c, R_01(r)); \
_MULx2(c, R_23(r)); \
break; \
case 2: \
_MULx2(c, R_01(r)); \
break; \
default: \
ASM_BUG(); \
COPY(R_01(r), _mul_x2_in); \
gf_x2_mul_fns[c](); \
COPY(_mul_x2_acc, R_01(r)); \
COPY(R_23(r), _mul_x2_in); \
gf_x2_mul_fns[c](); \
COPY(_mul_x2_acc, R_23(r)); \
} \
}
#define raidz_math_begin() kfpu_begin()
#define raidz_math_end() \
{ \
FLUSH(); \
kfpu_end(); \
}
#define raidz_math_end() kfpu_end()
#define SYN_STRIDE 4
#define ZERO_STRIDE 4
#define ZERO_DEFINE() {}
#define ZERO_D 20, 21, 22, 23
#define ZERO_D 0, 1, 2, 3
#define COPY_STRIDE 4
#define COPY_DEFINE() {}
#define COPY_D 20, 21, 22, 23
#define COPY_D 0, 1, 2, 3
#define ADD_STRIDE 4
#define ADD_DEFINE() {}
#define ADD_D 20, 21, 22, 23
#define ADD_D 0, 1, 2, 3
#define MUL_STRIDE 4
#define MUL_DEFINE() {}
#define MUL_D 20, 21, 22, 23
/*
* This use zmm16-zmm31 registers to free up zmm0-zmm15
* to use with the AVX2 pshufb, see above
*/
#define GEN_P_DEFINE() {}
#define MUL_DEFINE() MUL2_SETUP()
#define MUL_D 0, 1, 2, 3
#define GEN_P_STRIDE 4
#define GEN_P_P 20, 21, 22, 23
#define GEN_P_DEFINE() {}
#define GEN_P_P 0, 1, 2, 3
#define GEN_PQ_DEFINE() {}
#define GEN_PQ_STRIDE 4
#define GEN_PQ_D 20, 21, 22, 23
#define GEN_PQ_P 24, 25, 26, 27
#define GEN_PQ_Q 28, 29, 3, 4
#define GEN_PQ_DEFINE() {}
#define GEN_PQ_D 0, 1, 2, 3
#define GEN_PQ_C 4, 5, 6, 7
#define GEN_PQR_STRIDE 4
#define GEN_PQR_DEFINE() {}
#define GEN_PQR_STRIDE 2
#define GEN_PQR_D 20, 21
#define GEN_PQR_P 22, 23
#define GEN_PQR_Q 24, 25
#define GEN_PQR_R 26, 27
#define GEN_PQR_D 0, 1, 2, 3
#define GEN_PQR_C 4, 5, 6, 7
#define REC_P_DEFINE() {}
#define REC_P_STRIDE 4
#define REC_P_X 20, 21, 22, 23
#define SYN_Q_DEFINE() {}
#define SYN_Q_D 0, 1, 2, 3
#define SYN_Q_X 4, 5, 6, 7
#define REC_Q_DEFINE() {}
#define REC_Q_STRIDE 4
#define REC_Q_X 20, 21, 22, 23
#define SYN_R_DEFINE() {}
#define SYN_R_D 0, 1, 2, 3
#define SYN_R_X 4, 5, 6, 7
#define REC_R_DEFINE() {}
#define REC_R_STRIDE 4
#define REC_R_X 20, 21, 22, 23
#define SYN_PQ_DEFINE() {}
#define SYN_PQ_D 0, 1, 2, 3
#define SYN_PQ_X 4, 5, 6, 7
#define REC_PQ_DEFINE() {}
#define REC_PQ_STRIDE 2
#define REC_PQ_X 20, 21
#define REC_PQ_Y 22, 23
#define REC_PQ_D 24, 25
#define REC_PQ_STRIDE 4
#define REC_PQ_DEFINE() MUL2_SETUP()
#define REC_PQ_X 0, 1, 2, 3
#define REC_PQ_Y 4, 5, 6, 7
#define REC_PQ_T 8, 9, 10, 11
#define REC_PR_DEFINE() {}
#define REC_PR_STRIDE 2
#define REC_PR_X 20, 21
#define REC_PR_Y 22, 23
#define REC_PR_D 24, 25
#define SYN_PR_DEFINE() {}
#define SYN_PR_D 0, 1, 2, 3
#define SYN_PR_X 4, 5, 6, 7
#define REC_QR_DEFINE() {}
#define REC_QR_STRIDE 2
#define REC_QR_X 20, 21
#define REC_QR_Y 22, 23
#define REC_QR_D 24, 25
#define REC_PR_STRIDE 4
#define REC_PR_DEFINE() MUL2_SETUP()
#define REC_PR_X 0, 1, 2, 3
#define REC_PR_Y 4, 5, 6, 7
#define REC_PR_T 8, 9, 10, 11
#define REC_PQR_DEFINE() {}
#define REC_PQR_STRIDE 2
#define REC_PQR_X 20, 21
#define REC_PQR_Y 22, 23
#define REC_PQR_Z 24, 25
#define REC_PQR_D 26, 27
#define REC_PQR_XS 26, 27
#define REC_PQR_YS 28, 29
#define SYN_QR_DEFINE() {}
#define SYN_QR_D 0, 1, 2, 3
#define SYN_QR_X 4, 5, 6, 7
#define REC_QR_STRIDE 4
#define REC_QR_DEFINE() MUL2_SETUP()
#define REC_QR_X 0, 1, 2, 3
#define REC_QR_Y 4, 5, 6, 7
#define REC_QR_T 8, 9, 10, 11
#define SYN_PQR_DEFINE() {}
#define SYN_PQR_D 0, 1, 2, 3
#define SYN_PQR_X 4, 5, 6, 7
#define REC_PQR_STRIDE 4
#define REC_PQR_DEFINE() MUL2_SETUP()
#define REC_PQR_X 0, 1, 2, 3
#define REC_PQR_Y 4, 5, 6, 7
#define REC_PQR_Z 8, 9, 10, 11
#define REC_PQR_XS 12, 13, 14, 15
#define REC_PQR_YS 16, 17, 18, 19
#include <sys/vdev_raidz_impl.h>
@ -523,6 +471,7 @@ static boolean_t
raidz_will_avx512f_work(void)
{
return (zfs_avx_available() &&
zfs_avx2_available() &&
zfs_avx512f_available());
}

View File

@ -268,7 +268,7 @@ raidz_add_abd_cb(void *dc, void *sc, size_t size, void *private)
* @private pointer to the multiplication constant (unsigned)
*/
static int
raidz_mul_abd(void *dc, size_t size, void *private)
raidz_mul_abd_cb(void *dc, size_t size, void *private)
{
const unsigned mul = *((unsigned *) private);
v_t *d = (v_t *) dc;
@ -651,7 +651,7 @@ raidz_syn_q_abd(void **xc, const void *dc, const size_t xsize,
* Reconstruct single data column using Q parity
*
* @syn_method raidz_add_abd()
* @rec_method raidz_mul_abd()
* @rec_method raidz_mul_abd_cb()
*
* @rm RAIDZ map
* @tgtidx array of missing data indexes
@ -699,7 +699,7 @@ raidz_reconstruct_q_impl(raidz_map_t *rm, const int *tgtidx)
raidz_add(xabd, rm->rm_col[CODE_Q].rc_abd, xsize);
/* transform the syndrome */
abd_iterate_func(xabd, 0, xsize, raidz_mul_abd, (void*) coeff);
abd_iterate_func(xabd, 0, xsize, raidz_mul_abd_cb, (void*) coeff);
raidz_math_end();
@ -742,7 +742,7 @@ raidz_syn_r_abd(void **xc, const void *dc, const size_t tsize,
* Reconstruct single data column using R parity
*
* @syn_method raidz_add_abd()
* @rec_method raidz_mul_abd()
* @rec_method raidz_mul_abd_cb()
*
* @rm RAIDZ map
* @tgtidx array of missing data indexes
@ -791,7 +791,7 @@ raidz_reconstruct_r_impl(raidz_map_t *rm, const int *tgtidx)
raidz_add(xabd, rm->rm_col[CODE_R].rc_abd, xsize);
/* transform the syndrome */
abd_iterate_func(xabd, 0, xsize, raidz_mul_abd, (void *)coeff);
abd_iterate_func(xabd, 0, xsize, raidz_mul_abd_cb, (void *)coeff);
raidz_math_end();

View File

@ -58,9 +58,6 @@ typedef struct v {
uint8_t b[ELEM_SIZE] __attribute__((aligned(ELEM_SIZE)));
} v_t;
#define PREFETCHNTA(ptr, offset) {}
#define PREFETCH(ptr, offset) {}
#define XOR_ACC(src, r...) \
{ \
switch (REG_CNT(r)) { \
@ -106,27 +103,8 @@ typedef struct v {
break; \
} \
}
#define ZERO(r...) \
{ \
switch (REG_CNT(r)) { \
case 4: \
__asm( \
"pxor %" VR0(r) ", %" VR0(r) "\n" \
"pxor %" VR1(r) ", %" VR1(r) "\n" \
"pxor %" VR2(r) ", %" VR2(r) "\n" \
"pxor %" VR3(r) ", %" VR3(r)); \
break; \
case 2: \
__asm( \
"pxor %" VR0(r) ", %" VR0(r) "\n" \
"pxor %" VR1(r) ", %" VR1(r)); \
break; \
case 1: \
__asm( \
"pxor %" VR0(r) ", %" VR0(r)); \
break; \
} \
}
#define ZERO(r...) XOR(r, r)
#define COPY(r...) \
{ \
@ -259,7 +237,7 @@ typedef struct v {
#define _MUL_PARAM(x, in, acc) \
{ \
if (x & 0x01) { COPY(in, acc); } else { XOR(acc, acc); } \
if (x & 0x01) { COPY(in, acc); } else { ZERO(acc); } \
if (x & 0xfe) { MUL2(in); } \
if (x & 0x02) { XOR(in, acc); } \
if (x & 0xfc) { MUL2(in); } \
@ -552,7 +530,7 @@ gf_x2_mul_fns[256] = {
#define ADD_D 0, 1, 2, 3
#define MUL_STRIDE 2
#define MUL_DEFINE() {}
#define MUL_DEFINE() MUL2_SETUP()
#define MUL_D 0, 1
#define GEN_P_STRIDE 4
@ -582,7 +560,7 @@ gf_x2_mul_fns[256] = {
#define SYN_PQ_X 4, 5, 6, 7
#define REC_PQ_STRIDE 2
#define REC_PQ_DEFINE() {}
#define REC_PQ_DEFINE() MUL2_SETUP()
#define REC_PQ_X 0, 1
#define REC_PQ_Y 2, 3
#define REC_PQ_T 4, 5
@ -592,7 +570,7 @@ gf_x2_mul_fns[256] = {
#define SYN_PR_X 4, 5, 6, 7
#define REC_PR_STRIDE 2
#define REC_PR_DEFINE() {}
#define REC_PR_DEFINE() MUL2_SETUP()
#define REC_PR_X 0, 1
#define REC_PR_Y 2, 3
#define REC_PR_T 4, 5
@ -602,7 +580,7 @@ gf_x2_mul_fns[256] = {
#define SYN_QR_X 4, 5, 6, 7
#define REC_QR_STRIDE 2
#define REC_QR_DEFINE() {}
#define REC_QR_DEFINE() MUL2_SETUP()
#define REC_QR_X 0, 1
#define REC_QR_Y 2, 3
#define REC_QR_T 4, 5
@ -612,7 +590,7 @@ gf_x2_mul_fns[256] = {
#define SYN_PQR_X 4, 5, 6, 7
#define REC_PQR_STRIDE 1
#define REC_PQR_DEFINE() {}
#define REC_PQR_DEFINE() MUL2_SETUP()
#define REC_PQR_X 0
#define REC_PQR_Y 1
#define REC_PQR_Z 2

View File

@ -66,19 +66,6 @@ typedef struct v {
uint8_t b[ELEM_SIZE] __attribute__((aligned(ELEM_SIZE)));
} v_t;
#define PREFETCHNTA(ptr, offset) \
{ \
__asm( \
"prefetchnta " #offset "(%[MEM])\n" \
: : [MEM] "r" (ptr)); \
}
#define PREFETCH(ptr, offset) \
{ \
__asm( \
"prefetcht0 " #offset "(%[MEM])\n" \
: : [MEM] "r" (ptr)); \
}
#define XOR_ACC(src, r...) \
{ \
@ -122,25 +109,7 @@ typedef struct v {
} \
}
#define ZERO(r...) \
{ \
switch (REG_CNT(r)) { \
case 4: \
__asm( \
"pxor %" VR0(r) ", %" VR0(r) "\n" \
"pxor %" VR1(r) ", %" VR1(r) "\n" \
"pxor %" VR2(r) ", %" VR2(r) "\n" \
"pxor %" VR3(r) ", %" VR3(r)); \
break; \
case 2: \
__asm( \
"pxor %" VR0(r) ", %" VR0(r) "\n" \
"pxor %" VR1(r) ", %" VR1(r)); \
break; \
default: \
ASM_BUG(); \
} \
}
#define ZERO(r...) XOR(r, r)
#define COPY(r...) \
{ \
@ -446,7 +415,8 @@ const raidz_impl_ops_t vdev_raidz_ssse3_impl = {
#endif /* defined(__x86_64) && defined(HAVE_SSSE3) */
#if defined(__x86_64) && (defined(HAVE_SSSE3) || defined(HAVE_AVX2))
#if defined(__x86_64)
#if defined(HAVE_SSSE3) || defined(HAVE_AVX2) || defined(HAVE_AVX512BW)
const uint8_t
__attribute__((aligned(256))) gf_clmul_mod_lt[4*256][16] = {
@ -2500,4 +2470,5 @@ __attribute__((aligned(256))) gf_clmul_mod_lt[4*256][16] = {
0xf8, 0x07, 0x06, 0xf9, 0x04, 0xfb, 0xfa, 0x05 }
};
#endif /* defined(__x86_64) && (defined(HAVE_SSSE3) || defined(HAVE_AVX2)) */
#endif /* defined(HAVE_SSSE3) || defined(HAVE_AVX2) || defined(HAVE_AVX512BW) */
#endif /* defined(__x86_64) */