SIMD implementation of vdev_raidz generate and reconstruct routines
This is a new implementation of RAIDZ1/2/3 routines using x86_64
scalar, SSE, and AVX2 instruction sets. Included are 3 parity
generation routines (P, PQ, and PQR) and 7 reconstruction routines,
for all RAIDZ level. On module load, a quick benchmark of supported
routines will select the fastest for each operation and they will
be used at runtime. Original implementation is still present and
can be selected via module parameter.
Patch contains:
- specialized gen/rec routines for all RAIDZ levels,
- new scalar raidz implementation (unrolled),
- two x86_64 SIMD implementations (SSE and AVX2 instructions sets),
- fastest routines selected on module load (benchmark).
- cmd/raidz_test - verify and benchmark all implementations
- added raidz_test to the ZFS Test Suite
New zfs module parameters:
- zfs_vdev_raidz_impl (str): selects the implementation to use. On
module load, the parameter will only accept first 3 options, and
the other implementations can be set once module is finished
loading. Possible values for this option are:
"fastest" - use the fastest math available
"original" - use the original raidz code
"scalar" - new scalar impl
"sse" - new SSE impl if available
"avx2" - new AVX2 impl if available
See contents of `/sys/module/zfs/parameters/zfs_vdev_raidz_impl` to
get the list of supported values. If an implementation is not supported
on the system, it will not be shown. Currently selected option is
enclosed in `[]`.
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4328
2016-04-25 08:04:31 +00:00
|
|
|
/*
|
|
|
|
* CDDL HEADER START
|
|
|
|
*
|
|
|
|
* The contents of this file are subject to the terms of the
|
|
|
|
* Common Development and Distribution License (the "License").
|
|
|
|
* You may not use this file except in compliance with the License.
|
|
|
|
*
|
|
|
|
* You can obtain a copy of the license at usr/src/OPENSOLARIS.LICENSE
|
|
|
|
* or http://www.opensolaris.org/os/licensing.
|
|
|
|
* See the License for the specific language governing permissions
|
|
|
|
* and limitations under the License.
|
|
|
|
*
|
|
|
|
* When distributing Covered Code, include this CDDL HEADER in each
|
|
|
|
* file and include the License file at usr/src/OPENSOLARIS.LICENSE.
|
|
|
|
* If applicable, add the following below this CDDL HEADER, with the
|
|
|
|
* fields enclosed by brackets "[]" replaced with your own identifying
|
|
|
|
* information: Portions Copyright [yyyy] [name of copyright owner]
|
|
|
|
*
|
|
|
|
* CDDL HEADER END
|
|
|
|
*/
|
|
|
|
|
|
|
|
/*
|
|
|
|
* Copyright (C) 2016 Gvozden Nešković. All rights reserved.
|
|
|
|
*/
|
|
|
|
|
|
|
|
#include <sys/zfs_context.h>
|
|
|
|
#include <sys/time.h>
|
|
|
|
#include <sys/wait.h>
|
|
|
|
#include <sys/zio.h>
|
|
|
|
#include <sys/vdev_raidz.h>
|
|
|
|
#include <sys/vdev_raidz_impl.h>
|
|
|
|
#include <stdio.h>
|
|
|
|
|
|
|
|
#include <sys/time.h>
|
|
|
|
|
|
|
|
#include "raidz_test.h"
|
|
|
|
|
|
|
|
#define GEN_BENCH_MEMORY (((uint64_t)1ULL)<<32)
|
|
|
|
#define REC_BENCH_MEMORY (((uint64_t)1ULL)<<29)
|
|
|
|
#define BENCH_ASHIFT 12
|
|
|
|
#define MIN_CS_SHIFT BENCH_ASHIFT
|
|
|
|
#define MAX_CS_SHIFT SPA_MAXBLOCKSHIFT
|
|
|
|
|
|
|
|
static zio_t zio_bench;
|
|
|
|
static raidz_map_t *rm_bench;
|
|
|
|
static size_t max_data_size = SPA_MAXBLOCKSIZE;
|
|
|
|
|
|
|
|
static void
|
|
|
|
bench_init_raidz_map(void)
|
|
|
|
{
|
|
|
|
zio_bench.io_offset = 0;
|
|
|
|
zio_bench.io_size = max_data_size;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* To permit larger column sizes these have to be done
|
2016-08-24 13:51:33 +00:00
|
|
|
* allocated using aligned alloc instead of zio_abd_buf_alloc
|
SIMD implementation of vdev_raidz generate and reconstruct routines
This is a new implementation of RAIDZ1/2/3 routines using x86_64
scalar, SSE, and AVX2 instruction sets. Included are 3 parity
generation routines (P, PQ, and PQR) and 7 reconstruction routines,
for all RAIDZ level. On module load, a quick benchmark of supported
routines will select the fastest for each operation and they will
be used at runtime. Original implementation is still present and
can be selected via module parameter.
Patch contains:
- specialized gen/rec routines for all RAIDZ levels,
- new scalar raidz implementation (unrolled),
- two x86_64 SIMD implementations (SSE and AVX2 instructions sets),
- fastest routines selected on module load (benchmark).
- cmd/raidz_test - verify and benchmark all implementations
- added raidz_test to the ZFS Test Suite
New zfs module parameters:
- zfs_vdev_raidz_impl (str): selects the implementation to use. On
module load, the parameter will only accept first 3 options, and
the other implementations can be set once module is finished
loading. Possible values for this option are:
"fastest" - use the fastest math available
"original" - use the original raidz code
"scalar" - new scalar impl
"sse" - new SSE impl if available
"avx2" - new AVX2 impl if available
See contents of `/sys/module/zfs/parameters/zfs_vdev_raidz_impl` to
get the list of supported values. If an implementation is not supported
on the system, it will not be shown. Currently selected option is
enclosed in `[]`.
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4328
2016-04-25 08:04:31 +00:00
|
|
|
*/
|
2016-08-24 13:51:33 +00:00
|
|
|
zio_bench.io_abd = raidz_alloc(max_data_size);
|
SIMD implementation of vdev_raidz generate and reconstruct routines
This is a new implementation of RAIDZ1/2/3 routines using x86_64
scalar, SSE, and AVX2 instruction sets. Included are 3 parity
generation routines (P, PQ, and PQR) and 7 reconstruction routines,
for all RAIDZ level. On module load, a quick benchmark of supported
routines will select the fastest for each operation and they will
be used at runtime. Original implementation is still present and
can be selected via module parameter.
Patch contains:
- specialized gen/rec routines for all RAIDZ levels,
- new scalar raidz implementation (unrolled),
- two x86_64 SIMD implementations (SSE and AVX2 instructions sets),
- fastest routines selected on module load (benchmark).
- cmd/raidz_test - verify and benchmark all implementations
- added raidz_test to the ZFS Test Suite
New zfs module parameters:
- zfs_vdev_raidz_impl (str): selects the implementation to use. On
module load, the parameter will only accept first 3 options, and
the other implementations can be set once module is finished
loading. Possible values for this option are:
"fastest" - use the fastest math available
"original" - use the original raidz code
"scalar" - new scalar impl
"sse" - new SSE impl if available
"avx2" - new AVX2 impl if available
See contents of `/sys/module/zfs/parameters/zfs_vdev_raidz_impl` to
get the list of supported values. If an implementation is not supported
on the system, it will not be shown. Currently selected option is
enclosed in `[]`.
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4328
2016-04-25 08:04:31 +00:00
|
|
|
|
2016-08-24 13:51:33 +00:00
|
|
|
init_zio_abd(&zio_bench);
|
SIMD implementation of vdev_raidz generate and reconstruct routines
This is a new implementation of RAIDZ1/2/3 routines using x86_64
scalar, SSE, and AVX2 instruction sets. Included are 3 parity
generation routines (P, PQ, and PQR) and 7 reconstruction routines,
for all RAIDZ level. On module load, a quick benchmark of supported
routines will select the fastest for each operation and they will
be used at runtime. Original implementation is still present and
can be selected via module parameter.
Patch contains:
- specialized gen/rec routines for all RAIDZ levels,
- new scalar raidz implementation (unrolled),
- two x86_64 SIMD implementations (SSE and AVX2 instructions sets),
- fastest routines selected on module load (benchmark).
- cmd/raidz_test - verify and benchmark all implementations
- added raidz_test to the ZFS Test Suite
New zfs module parameters:
- zfs_vdev_raidz_impl (str): selects the implementation to use. On
module load, the parameter will only accept first 3 options, and
the other implementations can be set once module is finished
loading. Possible values for this option are:
"fastest" - use the fastest math available
"original" - use the original raidz code
"scalar" - new scalar impl
"sse" - new SSE impl if available
"avx2" - new AVX2 impl if available
See contents of `/sys/module/zfs/parameters/zfs_vdev_raidz_impl` to
get the list of supported values. If an implementation is not supported
on the system, it will not be shown. Currently selected option is
enclosed in `[]`.
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4328
2016-04-25 08:04:31 +00:00
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
bench_fini_raidz_maps(void)
|
|
|
|
{
|
|
|
|
/* tear down golden zio */
|
2016-08-24 13:51:33 +00:00
|
|
|
raidz_free(zio_bench.io_abd, max_data_size);
|
SIMD implementation of vdev_raidz generate and reconstruct routines
This is a new implementation of RAIDZ1/2/3 routines using x86_64
scalar, SSE, and AVX2 instruction sets. Included are 3 parity
generation routines (P, PQ, and PQR) and 7 reconstruction routines,
for all RAIDZ level. On module load, a quick benchmark of supported
routines will select the fastest for each operation and they will
be used at runtime. Original implementation is still present and
can be selected via module parameter.
Patch contains:
- specialized gen/rec routines for all RAIDZ levels,
- new scalar raidz implementation (unrolled),
- two x86_64 SIMD implementations (SSE and AVX2 instructions sets),
- fastest routines selected on module load (benchmark).
- cmd/raidz_test - verify and benchmark all implementations
- added raidz_test to the ZFS Test Suite
New zfs module parameters:
- zfs_vdev_raidz_impl (str): selects the implementation to use. On
module load, the parameter will only accept first 3 options, and
the other implementations can be set once module is finished
loading. Possible values for this option are:
"fastest" - use the fastest math available
"original" - use the original raidz code
"scalar" - new scalar impl
"sse" - new SSE impl if available
"avx2" - new AVX2 impl if available
See contents of `/sys/module/zfs/parameters/zfs_vdev_raidz_impl` to
get the list of supported values. If an implementation is not supported
on the system, it will not be shown. Currently selected option is
enclosed in `[]`.
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4328
2016-04-25 08:04:31 +00:00
|
|
|
bzero(&zio_bench, sizeof (zio_t));
|
|
|
|
}
|
|
|
|
|
|
|
|
static inline void
|
|
|
|
run_gen_bench_impl(const char *impl)
|
|
|
|
{
|
|
|
|
int fn, ncols;
|
|
|
|
uint64_t ds, iter_cnt, iter, disksize;
|
2016-07-17 17:41:11 +00:00
|
|
|
hrtime_t start;
|
SIMD implementation of vdev_raidz generate and reconstruct routines
This is a new implementation of RAIDZ1/2/3 routines using x86_64
scalar, SSE, and AVX2 instruction sets. Included are 3 parity
generation routines (P, PQ, and PQR) and 7 reconstruction routines,
for all RAIDZ level. On module load, a quick benchmark of supported
routines will select the fastest for each operation and they will
be used at runtime. Original implementation is still present and
can be selected via module parameter.
Patch contains:
- specialized gen/rec routines for all RAIDZ levels,
- new scalar raidz implementation (unrolled),
- two x86_64 SIMD implementations (SSE and AVX2 instructions sets),
- fastest routines selected on module load (benchmark).
- cmd/raidz_test - verify and benchmark all implementations
- added raidz_test to the ZFS Test Suite
New zfs module parameters:
- zfs_vdev_raidz_impl (str): selects the implementation to use. On
module load, the parameter will only accept first 3 options, and
the other implementations can be set once module is finished
loading. Possible values for this option are:
"fastest" - use the fastest math available
"original" - use the original raidz code
"scalar" - new scalar impl
"sse" - new SSE impl if available
"avx2" - new AVX2 impl if available
See contents of `/sys/module/zfs/parameters/zfs_vdev_raidz_impl` to
get the list of supported values. If an implementation is not supported
on the system, it will not be shown. Currently selected option is
enclosed in `[]`.
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4328
2016-04-25 08:04:31 +00:00
|
|
|
double elapsed, d_bw;
|
|
|
|
|
|
|
|
/* Benchmark generate functions */
|
|
|
|
for (fn = 0; fn < RAIDZ_GEN_NUM; fn++) {
|
|
|
|
|
|
|
|
for (ds = MIN_CS_SHIFT; ds <= MAX_CS_SHIFT; ds++) {
|
|
|
|
/* create suitable raidz_map */
|
|
|
|
ncols = rto_opts.rto_dcols + fn + 1;
|
|
|
|
zio_bench.io_size = 1ULL << ds;
|
|
|
|
rm_bench = vdev_raidz_map_alloc(&zio_bench,
|
|
|
|
BENCH_ASHIFT, ncols, fn+1);
|
|
|
|
|
|
|
|
/* estimate iteration count */
|
|
|
|
iter_cnt = GEN_BENCH_MEMORY;
|
|
|
|
iter_cnt /= zio_bench.io_size;
|
|
|
|
|
2016-07-17 17:41:11 +00:00
|
|
|
start = gethrtime();
|
SIMD implementation of vdev_raidz generate and reconstruct routines
This is a new implementation of RAIDZ1/2/3 routines using x86_64
scalar, SSE, and AVX2 instruction sets. Included are 3 parity
generation routines (P, PQ, and PQR) and 7 reconstruction routines,
for all RAIDZ level. On module load, a quick benchmark of supported
routines will select the fastest for each operation and they will
be used at runtime. Original implementation is still present and
can be selected via module parameter.
Patch contains:
- specialized gen/rec routines for all RAIDZ levels,
- new scalar raidz implementation (unrolled),
- two x86_64 SIMD implementations (SSE and AVX2 instructions sets),
- fastest routines selected on module load (benchmark).
- cmd/raidz_test - verify and benchmark all implementations
- added raidz_test to the ZFS Test Suite
New zfs module parameters:
- zfs_vdev_raidz_impl (str): selects the implementation to use. On
module load, the parameter will only accept first 3 options, and
the other implementations can be set once module is finished
loading. Possible values for this option are:
"fastest" - use the fastest math available
"original" - use the original raidz code
"scalar" - new scalar impl
"sse" - new SSE impl if available
"avx2" - new AVX2 impl if available
See contents of `/sys/module/zfs/parameters/zfs_vdev_raidz_impl` to
get the list of supported values. If an implementation is not supported
on the system, it will not be shown. Currently selected option is
enclosed in `[]`.
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4328
2016-04-25 08:04:31 +00:00
|
|
|
for (iter = 0; iter < iter_cnt; iter++)
|
|
|
|
vdev_raidz_generate_parity(rm_bench);
|
2016-12-12 18:46:26 +00:00
|
|
|
elapsed = NSEC2SEC((double)(gethrtime() - start));
|
SIMD implementation of vdev_raidz generate and reconstruct routines
This is a new implementation of RAIDZ1/2/3 routines using x86_64
scalar, SSE, and AVX2 instruction sets. Included are 3 parity
generation routines (P, PQ, and PQR) and 7 reconstruction routines,
for all RAIDZ level. On module load, a quick benchmark of supported
routines will select the fastest for each operation and they will
be used at runtime. Original implementation is still present and
can be selected via module parameter.
Patch contains:
- specialized gen/rec routines for all RAIDZ levels,
- new scalar raidz implementation (unrolled),
- two x86_64 SIMD implementations (SSE and AVX2 instructions sets),
- fastest routines selected on module load (benchmark).
- cmd/raidz_test - verify and benchmark all implementations
- added raidz_test to the ZFS Test Suite
New zfs module parameters:
- zfs_vdev_raidz_impl (str): selects the implementation to use. On
module load, the parameter will only accept first 3 options, and
the other implementations can be set once module is finished
loading. Possible values for this option are:
"fastest" - use the fastest math available
"original" - use the original raidz code
"scalar" - new scalar impl
"sse" - new SSE impl if available
"avx2" - new AVX2 impl if available
See contents of `/sys/module/zfs/parameters/zfs_vdev_raidz_impl` to
get the list of supported values. If an implementation is not supported
on the system, it will not be shown. Currently selected option is
enclosed in `[]`.
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4328
2016-04-25 08:04:31 +00:00
|
|
|
|
|
|
|
disksize = (1ULL << ds) / rto_opts.rto_dcols;
|
|
|
|
d_bw = (double)iter_cnt * (double)disksize;
|
|
|
|
d_bw /= (1024.0 * 1024.0 * elapsed);
|
|
|
|
|
|
|
|
LOG(D_ALL, "%10s, %8s, %zu, %10llu, %lf, %lf, %u\n",
|
|
|
|
impl,
|
|
|
|
raidz_gen_name[fn],
|
|
|
|
rto_opts.rto_dcols,
|
|
|
|
(1ULL<<ds),
|
|
|
|
d_bw,
|
|
|
|
d_bw * (double)(ncols),
|
2016-12-12 18:46:26 +00:00
|
|
|
(unsigned)iter_cnt);
|
SIMD implementation of vdev_raidz generate and reconstruct routines
This is a new implementation of RAIDZ1/2/3 routines using x86_64
scalar, SSE, and AVX2 instruction sets. Included are 3 parity
generation routines (P, PQ, and PQR) and 7 reconstruction routines,
for all RAIDZ level. On module load, a quick benchmark of supported
routines will select the fastest for each operation and they will
be used at runtime. Original implementation is still present and
can be selected via module parameter.
Patch contains:
- specialized gen/rec routines for all RAIDZ levels,
- new scalar raidz implementation (unrolled),
- two x86_64 SIMD implementations (SSE and AVX2 instructions sets),
- fastest routines selected on module load (benchmark).
- cmd/raidz_test - verify and benchmark all implementations
- added raidz_test to the ZFS Test Suite
New zfs module parameters:
- zfs_vdev_raidz_impl (str): selects the implementation to use. On
module load, the parameter will only accept first 3 options, and
the other implementations can be set once module is finished
loading. Possible values for this option are:
"fastest" - use the fastest math available
"original" - use the original raidz code
"scalar" - new scalar impl
"sse" - new SSE impl if available
"avx2" - new AVX2 impl if available
See contents of `/sys/module/zfs/parameters/zfs_vdev_raidz_impl` to
get the list of supported values. If an implementation is not supported
on the system, it will not be shown. Currently selected option is
enclosed in `[]`.
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4328
2016-04-25 08:04:31 +00:00
|
|
|
|
|
|
|
vdev_raidz_map_free(rm_bench);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void
|
|
|
|
run_gen_bench(void)
|
|
|
|
{
|
|
|
|
char **impl_name;
|
|
|
|
|
|
|
|
LOG(D_INFO, DBLSEP "\nBenchmarking parity generation...\n\n");
|
|
|
|
LOG(D_ALL, "impl, math, dcols, iosize, disk_bw, total_bw, iter\n");
|
|
|
|
|
|
|
|
for (impl_name = (char **)raidz_impl_names; *impl_name != NULL;
|
|
|
|
impl_name++) {
|
|
|
|
|
|
|
|
if (vdev_raidz_impl_set(*impl_name) != 0)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
run_gen_bench_impl(*impl_name);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
static void
|
|
|
|
run_rec_bench_impl(const char *impl)
|
|
|
|
{
|
|
|
|
int fn, ncols, nbad;
|
|
|
|
uint64_t ds, iter_cnt, iter, disksize;
|
2016-07-17 17:41:11 +00:00
|
|
|
hrtime_t start;
|
SIMD implementation of vdev_raidz generate and reconstruct routines
This is a new implementation of RAIDZ1/2/3 routines using x86_64
scalar, SSE, and AVX2 instruction sets. Included are 3 parity
generation routines (P, PQ, and PQR) and 7 reconstruction routines,
for all RAIDZ level. On module load, a quick benchmark of supported
routines will select the fastest for each operation and they will
be used at runtime. Original implementation is still present and
can be selected via module parameter.
Patch contains:
- specialized gen/rec routines for all RAIDZ levels,
- new scalar raidz implementation (unrolled),
- two x86_64 SIMD implementations (SSE and AVX2 instructions sets),
- fastest routines selected on module load (benchmark).
- cmd/raidz_test - verify and benchmark all implementations
- added raidz_test to the ZFS Test Suite
New zfs module parameters:
- zfs_vdev_raidz_impl (str): selects the implementation to use. On
module load, the parameter will only accept first 3 options, and
the other implementations can be set once module is finished
loading. Possible values for this option are:
"fastest" - use the fastest math available
"original" - use the original raidz code
"scalar" - new scalar impl
"sse" - new SSE impl if available
"avx2" - new AVX2 impl if available
See contents of `/sys/module/zfs/parameters/zfs_vdev_raidz_impl` to
get the list of supported values. If an implementation is not supported
on the system, it will not be shown. Currently selected option is
enclosed in `[]`.
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4328
2016-04-25 08:04:31 +00:00
|
|
|
double elapsed, d_bw;
|
|
|
|
static const int tgt[7][3] = {
|
|
|
|
{1, 2, 3}, /* rec_p: bad QR & D[0] */
|
|
|
|
{0, 2, 3}, /* rec_q: bad PR & D[0] */
|
|
|
|
{0, 1, 3}, /* rec_r: bad PQ & D[0] */
|
|
|
|
{2, 3, 4}, /* rec_pq: bad R & D[0][1] */
|
|
|
|
{1, 3, 4}, /* rec_pr: bad Q & D[0][1] */
|
|
|
|
{0, 3, 4}, /* rec_qr: bad P & D[0][1] */
|
|
|
|
{3, 4, 5} /* rec_pqr: bad & D[0][1][2] */
|
|
|
|
};
|
|
|
|
|
|
|
|
for (fn = 0; fn < RAIDZ_REC_NUM; fn++) {
|
|
|
|
for (ds = MIN_CS_SHIFT; ds <= MAX_CS_SHIFT; ds++) {
|
|
|
|
|
|
|
|
/* create suitable raidz_map */
|
|
|
|
ncols = rto_opts.rto_dcols + PARITY_PQR;
|
|
|
|
zio_bench.io_size = 1ULL << ds;
|
|
|
|
|
|
|
|
/*
|
|
|
|
* raidz block is too short to test
|
|
|
|
* the requested method
|
|
|
|
*/
|
|
|
|
if (zio_bench.io_size / rto_opts.rto_dcols <
|
|
|
|
(1ULL << BENCH_ASHIFT))
|
|
|
|
continue;
|
|
|
|
|
|
|
|
rm_bench = vdev_raidz_map_alloc(&zio_bench,
|
2016-12-12 18:46:26 +00:00
|
|
|
BENCH_ASHIFT, ncols, PARITY_PQR);
|
SIMD implementation of vdev_raidz generate and reconstruct routines
This is a new implementation of RAIDZ1/2/3 routines using x86_64
scalar, SSE, and AVX2 instruction sets. Included are 3 parity
generation routines (P, PQ, and PQR) and 7 reconstruction routines,
for all RAIDZ level. On module load, a quick benchmark of supported
routines will select the fastest for each operation and they will
be used at runtime. Original implementation is still present and
can be selected via module parameter.
Patch contains:
- specialized gen/rec routines for all RAIDZ levels,
- new scalar raidz implementation (unrolled),
- two x86_64 SIMD implementations (SSE and AVX2 instructions sets),
- fastest routines selected on module load (benchmark).
- cmd/raidz_test - verify and benchmark all implementations
- added raidz_test to the ZFS Test Suite
New zfs module parameters:
- zfs_vdev_raidz_impl (str): selects the implementation to use. On
module load, the parameter will only accept first 3 options, and
the other implementations can be set once module is finished
loading. Possible values for this option are:
"fastest" - use the fastest math available
"original" - use the original raidz code
"scalar" - new scalar impl
"sse" - new SSE impl if available
"avx2" - new AVX2 impl if available
See contents of `/sys/module/zfs/parameters/zfs_vdev_raidz_impl` to
get the list of supported values. If an implementation is not supported
on the system, it will not be shown. Currently selected option is
enclosed in `[]`.
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4328
2016-04-25 08:04:31 +00:00
|
|
|
|
|
|
|
/* estimate iteration count */
|
|
|
|
iter_cnt = (REC_BENCH_MEMORY);
|
|
|
|
iter_cnt /= zio_bench.io_size;
|
|
|
|
|
|
|
|
/* calculate how many bad columns there are */
|
|
|
|
nbad = MIN(3, raidz_ncols(rm_bench) -
|
|
|
|
raidz_parity(rm_bench));
|
|
|
|
|
2016-07-17 17:41:11 +00:00
|
|
|
start = gethrtime();
|
SIMD implementation of vdev_raidz generate and reconstruct routines
This is a new implementation of RAIDZ1/2/3 routines using x86_64
scalar, SSE, and AVX2 instruction sets. Included are 3 parity
generation routines (P, PQ, and PQR) and 7 reconstruction routines,
for all RAIDZ level. On module load, a quick benchmark of supported
routines will select the fastest for each operation and they will
be used at runtime. Original implementation is still present and
can be selected via module parameter.
Patch contains:
- specialized gen/rec routines for all RAIDZ levels,
- new scalar raidz implementation (unrolled),
- two x86_64 SIMD implementations (SSE and AVX2 instructions sets),
- fastest routines selected on module load (benchmark).
- cmd/raidz_test - verify and benchmark all implementations
- added raidz_test to the ZFS Test Suite
New zfs module parameters:
- zfs_vdev_raidz_impl (str): selects the implementation to use. On
module load, the parameter will only accept first 3 options, and
the other implementations can be set once module is finished
loading. Possible values for this option are:
"fastest" - use the fastest math available
"original" - use the original raidz code
"scalar" - new scalar impl
"sse" - new SSE impl if available
"avx2" - new AVX2 impl if available
See contents of `/sys/module/zfs/parameters/zfs_vdev_raidz_impl` to
get the list of supported values. If an implementation is not supported
on the system, it will not be shown. Currently selected option is
enclosed in `[]`.
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4328
2016-04-25 08:04:31 +00:00
|
|
|
for (iter = 0; iter < iter_cnt; iter++)
|
|
|
|
vdev_raidz_reconstruct(rm_bench, tgt[fn], nbad);
|
2016-12-12 18:46:26 +00:00
|
|
|
elapsed = NSEC2SEC((double)(gethrtime() - start));
|
SIMD implementation of vdev_raidz generate and reconstruct routines
This is a new implementation of RAIDZ1/2/3 routines using x86_64
scalar, SSE, and AVX2 instruction sets. Included are 3 parity
generation routines (P, PQ, and PQR) and 7 reconstruction routines,
for all RAIDZ level. On module load, a quick benchmark of supported
routines will select the fastest for each operation and they will
be used at runtime. Original implementation is still present and
can be selected via module parameter.
Patch contains:
- specialized gen/rec routines for all RAIDZ levels,
- new scalar raidz implementation (unrolled),
- two x86_64 SIMD implementations (SSE and AVX2 instructions sets),
- fastest routines selected on module load (benchmark).
- cmd/raidz_test - verify and benchmark all implementations
- added raidz_test to the ZFS Test Suite
New zfs module parameters:
- zfs_vdev_raidz_impl (str): selects the implementation to use. On
module load, the parameter will only accept first 3 options, and
the other implementations can be set once module is finished
loading. Possible values for this option are:
"fastest" - use the fastest math available
"original" - use the original raidz code
"scalar" - new scalar impl
"sse" - new SSE impl if available
"avx2" - new AVX2 impl if available
See contents of `/sys/module/zfs/parameters/zfs_vdev_raidz_impl` to
get the list of supported values. If an implementation is not supported
on the system, it will not be shown. Currently selected option is
enclosed in `[]`.
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4328
2016-04-25 08:04:31 +00:00
|
|
|
|
|
|
|
disksize = (1ULL << ds) / rto_opts.rto_dcols;
|
|
|
|
d_bw = (double)iter_cnt * (double)(disksize);
|
|
|
|
d_bw /= (1024.0 * 1024.0 * elapsed);
|
|
|
|
|
|
|
|
LOG(D_ALL, "%10s, %8s, %zu, %10llu, %lf, %lf, %u\n",
|
|
|
|
impl,
|
|
|
|
raidz_rec_name[fn],
|
|
|
|
rto_opts.rto_dcols,
|
|
|
|
(1ULL<<ds),
|
|
|
|
d_bw,
|
|
|
|
d_bw * (double)ncols,
|
2016-12-12 18:46:26 +00:00
|
|
|
(unsigned)iter_cnt);
|
SIMD implementation of vdev_raidz generate and reconstruct routines
This is a new implementation of RAIDZ1/2/3 routines using x86_64
scalar, SSE, and AVX2 instruction sets. Included are 3 parity
generation routines (P, PQ, and PQR) and 7 reconstruction routines,
for all RAIDZ level. On module load, a quick benchmark of supported
routines will select the fastest for each operation and they will
be used at runtime. Original implementation is still present and
can be selected via module parameter.
Patch contains:
- specialized gen/rec routines for all RAIDZ levels,
- new scalar raidz implementation (unrolled),
- two x86_64 SIMD implementations (SSE and AVX2 instructions sets),
- fastest routines selected on module load (benchmark).
- cmd/raidz_test - verify and benchmark all implementations
- added raidz_test to the ZFS Test Suite
New zfs module parameters:
- zfs_vdev_raidz_impl (str): selects the implementation to use. On
module load, the parameter will only accept first 3 options, and
the other implementations can be set once module is finished
loading. Possible values for this option are:
"fastest" - use the fastest math available
"original" - use the original raidz code
"scalar" - new scalar impl
"sse" - new SSE impl if available
"avx2" - new AVX2 impl if available
See contents of `/sys/module/zfs/parameters/zfs_vdev_raidz_impl` to
get the list of supported values. If an implementation is not supported
on the system, it will not be shown. Currently selected option is
enclosed in `[]`.
Signed-off-by: Gvozden Neskovic <neskovic@gmail.com>
Signed-off-by: Brian Behlendorf <behlendorf1@llnl.gov>
Closes #4328
2016-04-25 08:04:31 +00:00
|
|
|
|
|
|
|
vdev_raidz_map_free(rm_bench);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void
|
|
|
|
run_rec_bench(void)
|
|
|
|
{
|
|
|
|
char **impl_name;
|
|
|
|
|
|
|
|
LOG(D_INFO, DBLSEP "\nBenchmarking data reconstruction...\n\n");
|
|
|
|
LOG(D_ALL, "impl, math, dcols, iosize, disk_bw, total_bw, iter\n");
|
|
|
|
|
|
|
|
for (impl_name = (char **)raidz_impl_names; *impl_name != NULL;
|
|
|
|
impl_name++) {
|
|
|
|
|
|
|
|
if (vdev_raidz_impl_set(*impl_name) != 0)
|
|
|
|
continue;
|
|
|
|
|
|
|
|
run_rec_bench_impl(*impl_name);
|
|
|
|
}
|
|
|
|
}
|
|
|
|
|
|
|
|
void
|
|
|
|
run_raidz_benchmark(void)
|
|
|
|
{
|
|
|
|
bench_init_raidz_map();
|
|
|
|
|
|
|
|
run_gen_bench();
|
|
|
|
run_rec_bench();
|
|
|
|
|
|
|
|
bench_fini_raidz_maps();
|
|
|
|
}
|