From mboxrd@z Thu Jan 1 00:00:00 1970 Return-Path: Received: from mga06.intel.com (mga06.intel.com [134.134.136.31]) by dpdk.org (Postfix) with ESMTP id 69CAE37A8 for ; Tue, 27 Dec 2016 11:05:43 +0100 (CET) Received: from orsmga004.jf.intel.com ([10.7.209.38]) by orsmga104.jf.intel.com with ESMTP; 27 Dec 2016 02:05:43 -0800 X-ExtLoop1: 1 X-IronPort-AV: E=Sophos;i="5.33,415,1477983600"; d="scan'208";a="43603072" Received: from unknown (HELO dpdk5.bj.intel.com) ([172.16.182.188]) by orsmga004.jf.intel.com with ESMTP; 27 Dec 2016 02:05:41 -0800 From: Zhiyong Yang To: dev@dpdk.org Cc: yuanhan.liu@linux.intel.com, thomas.monjalon@6wind.com, bruce.richardson@intel.com, konstantin.ananyev@intel.com, pablo.de.lara.guarch@intel.com, Zhiyong Yang Date: Tue, 27 Dec 2016 18:04:57 +0800 Message-Id: <1482833098-38096-4-git-send-email-zhiyong.yang@intel.com> X-Mailer: git-send-email 2.7.4 In-Reply-To: <1482833098-38096-1-git-send-email-zhiyong.yang@intel.com> References: <1480926387-63838-2-git-send-email-zhiyong.yang@intel.com> <1482833098-38096-1-git-send-email-zhiyong.yang@intel.com> Subject: [dpdk-dev] [PATCH v2 3/4] app/test: add performance autotest for rte_memset X-BeenThere: dev@dpdk.org X-Mailman-Version: 2.1.15 Precedence: list List-Id: DPDK patches and discussions List-Unsubscribe: , List-Archive: List-Post: List-Help: List-Subscribe: , X-List-Received-Date: Tue, 27 Dec 2016 10:05:44 -0000 The file implements the perf autotest for rte_memset. The perf data can be gotten compared between rte_memset and memset when you run it. We can see the perf of rte_memset obviously is better than glibc memset especially for small N bytes. The first column shows the N size for memset & rte_memset. The second column lists a set of numbers for rte_memset Vs memset perf in cache. The third column lists a set of numbers for rte_memset Vs memset perf in memory. The following data is gotten on haswell. ** rte_memset() - memset perf tests (C = compile-time constant) ** ======== ======= ======== ======= ======== Size memset in cache memset in mem (bytes) (ticks) (ticks) ------- -------------- --------------- ============= 32B aligned ================ 1 3 - 8 14 - 115 3 4 - 8 19 - 125 6 3 - 7 19 - 125 8 3 - 6 19 - 124 12 3 - 6 19 - 124 15 3 - 6 19 - 125 16 3 - 8 13 - 125 32 3 - 7 19 - 133 64 3 - 7 28 - 162 65 6 - 8 41 - 182 128 6 - 13 54 - 199 192 8 - 13 77 - 273 255 8 - 16 100 - 222 512 17 - 14 187 - 247 768 22 - 20 270 - 362 1024 29 - 28 329 - 377 2048 63 - 57 564 - 601 4096 104 - 102 993 - 1025 8192 200 - 211 1831 - 2270 ------ -------------- -------------- ------ C 6 2 - 2 19 - 19 C 64 2 - 6 28 - 33 C 128 3 - 12 54 - 59 C 192 5 - 29 77 - 83 C 256 6 - 35 100 - 105 C 512 12 - 60 188 - 195 C 768 18 - 20 271 - 362 C 1024 24 - 29 329 - 377 Signed-off-by: Zhiyong Yang --- Change in V2: Add perf comparation data between rte_memset and memset on haswell. app/test/Makefile | 1 + app/test/test_memset_perf.c | 348 ++++++++++++++++++++++++++++++++++++++++++++ 2 files changed, 349 insertions(+) create mode 100644 app/test/test_memset_perf.c diff --git a/app/test/Makefile b/app/test/Makefile index 82da3f3..1c3e7f1 100644 --- a/app/test/Makefile +++ b/app/test/Makefile @@ -124,6 +124,7 @@ SRCS-y += test_memcpy.c SRCS-y += test_memcpy_perf.c SRCS-y += test_memset.c +SRCS-y += test_memset_perf.c SRCS-$(CONFIG_RTE_LIBRTE_HASH) += test_hash.c SRCS-$(CONFIG_RTE_LIBRTE_HASH) += test_thash.c diff --git a/app/test/test_memset_perf.c b/app/test/test_memset_perf.c new file mode 100644 index 0000000..83b15b5 --- /dev/null +++ b/app/test/test_memset_perf.c @@ -0,0 +1,348 @@ +/*- + * BSD LICENSE + * + * Copyright(c) 2010-2016 Intel Corporation. All rights reserved. + * 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. + */ + +#include +#include +#include +#include +#include +#include +#include +#include +#include +#include "test.h" + +/* + * Set this to the maximum buffer size you want to test. If it is 0, then the + * values in the buf_sizes[] array below will be used. + */ +#define TEST_VALUE_RANGE 0 + +/* List of buffer sizes to test */ +#if TEST_VALUE_RANGE == 0 +static size_t buf_sizes[] = { + 1, 2, 3, 4, 5, 6, 7, 8, 9, 12, 15, 16, 17, 31, 32, 33, 63, 64, 65, + 70, 85, 96, 105, 115, 127, 128, 129, 161, 191, 192, 193, 255, 256, + 257, 319, 320, 321, 383, 384, 385, 447, 448, 449, 511, 512, 513, + 767, 768, 769, 1023, 1024, 1025, 1518, 1522, 1536, 1600, 2048, 2560, + 3072, 3584, 4096, 4608, 5120, 5632, 6144, 6656, 7168, 7680, 8192 +}; +/* MUST be as large as largest packet size above */ +#define SMALL_BUFFER_SIZE 8192 +#else /* TEST_VALUE_RANGE != 0 */ +static size_t buf_sizes[TEST_VALUE_RANGE]; +#define SMALL_BUFFER_SIZE TEST_VALUE_RANGE +#endif /* TEST_VALUE_RANGE == 0 */ + +/* + * Arrays of this size are used for measuring uncached memory accesses by + * picking a random location within the buffer. Make this smaller if there are + * memory allocation errors. + */ +#define LARGE_BUFFER_SIZE (100 * 1024 * 1024) + +/* How many times to run timing loop for performance tests */ +#define TEST_ITERATIONS 1000000 +#define TEST_BATCH_SIZE 100 + +/* Data is aligned on this many bytes (power of 2) */ +#ifdef RTE_MACHINE_CPUFLAG_AVX512F +#define ALIGNMENT_UNIT 64 +#elif defined RTE_MACHINE_CPUFLAG_AVX2 +#define ALIGNMENT_UNIT 32 +#else /* RTE_MACHINE_CPUFLAG */ +#define ALIGNMENT_UNIT 16 +#endif /* RTE_MACHINE_CPUFLAG */ + +/* + * Pointers used in performance tests. The two large buffers are for uncached + * access where random addresses within the buffer are used for each + * memset. The two small buffers are for cached access. + */ +static uint8_t *large_buf_read, *large_buf_write; +static uint8_t *small_buf_read, *small_buf_write; + +/* Initialise data buffers. */ +static int +init_buffers(void) +{ + unsigned int i; + + large_buf_read = rte_malloc("memset", LARGE_BUFFER_SIZE + + ALIGNMENT_UNIT, ALIGNMENT_UNIT); + if (large_buf_read == NULL) + goto error_large_buf_read; + + large_buf_write = rte_malloc("memset", LARGE_BUFFER_SIZE + + ALIGNMENT_UNIT, ALIGNMENT_UNIT); + if (large_buf_write == NULL) + goto error_large_buf_write; + + small_buf_read = rte_malloc("memset", SMALL_BUFFER_SIZE + + ALIGNMENT_UNIT, ALIGNMENT_UNIT); + if (small_buf_read == NULL) + goto error_small_buf_read; + + small_buf_write = rte_malloc("memset", SMALL_BUFFER_SIZE + + ALIGNMENT_UNIT, ALIGNMENT_UNIT); + if (small_buf_write == NULL) + goto error_small_buf_write; + + for (i = 0; i < LARGE_BUFFER_SIZE; i++) + large_buf_read[i] = rte_rand(); + for (i = 0; i < SMALL_BUFFER_SIZE; i++) + small_buf_read[i] = rte_rand(); + + return 0; + +error_small_buf_write: + rte_free(small_buf_read); +error_small_buf_read: + rte_free(large_buf_write); +error_large_buf_write: + rte_free(large_buf_read); +error_large_buf_read: + printf("ERROR: not enough memory\n"); + return -1; +} + +/* Cleanup data buffers */ +static void +free_buffers(void) +{ + rte_free(large_buf_read); + rte_free(large_buf_write); + rte_free(small_buf_read); + rte_free(small_buf_write); +} + +/* + * Get a random offset into large array, with enough space needed to perform + * max memset size. Offset is aligned, uoffset is used for unalignment setting. + */ +static inline size_t +get_rand_offset(size_t uoffset) +{ + return ((rte_rand() % (LARGE_BUFFER_SIZE - SMALL_BUFFER_SIZE)) & + ~(ALIGNMENT_UNIT - 1)) + uoffset; +} + +/* Fill in destination addresses. */ +static inline void +fill_addr_arrays(size_t *dst_addr, int is_dst_cached, size_t dst_uoffset) +{ + unsigned int i; + + for (i = 0; i < TEST_BATCH_SIZE; i++) + dst_addr[i] = (is_dst_cached) ? dst_uoffset : + get_rand_offset(dst_uoffset); +} + +/* + * WORKAROUND: For some reason the first test doing an uncached write + * takes a very long time (~25 times longer than is expected). So we do + * it once without timing. + */ +static void +do_uncached_write(uint8_t *dst, int is_dst_cached, size_t size) +{ + unsigned int i, j; + size_t dst_addrs[TEST_BATCH_SIZE]; + int ch = rte_rand() & 0xff; + + for (i = 0; i < (TEST_ITERATIONS / TEST_BATCH_SIZE); i++) { + fill_addr_arrays(dst_addrs, is_dst_cached, 0); + for (j = 0; j < TEST_BATCH_SIZE; j++) + rte_memset(dst+dst_addrs[j], ch, size); + } +} + +/* + * Run a single memset performance test. This is a macro to ensure that if + * the "size" parameter is a constant it won't be converted to a variable. + */ +#define SINGLE_PERF_TEST(dst, is_dst_cached, dst_uoffset, size) \ +do { \ + unsigned int iter, t; \ + size_t dst_addrs[TEST_BATCH_SIZE]; \ + uint64_t start_time, total_time = 0; \ + uint64_t total_time2 = 0; \ + int ch = rte_rand() & 0xff; \ + \ + for (iter = 0; iter < (TEST_ITERATIONS / TEST_BATCH_SIZE); iter++) {\ + fill_addr_arrays(dst_addrs, is_dst_cached, dst_uoffset); \ + start_time = rte_rdtsc(); \ + for (t = 0; t < TEST_BATCH_SIZE; t++) \ + rte_memset(dst+dst_addrs[t], ch, size); \ + total_time += rte_rdtsc() - start_time; \ + } \ + for (iter = 0; iter < (TEST_ITERATIONS / TEST_BATCH_SIZE); iter++) {\ + fill_addr_arrays(dst_addrs, is_dst_cached, dst_uoffset); \ + start_time = rte_rdtsc(); \ + for (t = 0; t < TEST_BATCH_SIZE; t++) \ + memset(dst+dst_addrs[t], ch, size); \ + total_time2 += rte_rdtsc() - start_time; \ + } \ + printf("%8.0f -", (double)total_time / TEST_ITERATIONS); \ + printf("%5.0f", (double)total_time2 / TEST_ITERATIONS); \ +} while (0) + +/* Run aligned memset tests. */ +#define ALL_PERF_TESTS_FOR_SIZE(n) \ +do { \ + if (__builtin_constant_p(n)) \ + printf("\nC%6u", (unsigned int)n); \ + else \ + printf("\n%7u", (unsigned int)n); \ + SINGLE_PERF_TEST(small_buf_write, 1, 0, n); \ + SINGLE_PERF_TEST(large_buf_write, 0, 0, n); \ +} while (0) + +/* Run unaligned memset tests */ +#define ALL_PERF_TESTS_FOR_SIZE_UNALIGNED(n) \ +do { \ + if (__builtin_constant_p(n)) \ + printf("\nC%6u", (unsigned int)n); \ + else \ + printf("\n%7u", (unsigned int)n); \ + SINGLE_PERF_TEST(small_buf_write, 1, 1, n); \ + SINGLE_PERF_TEST(large_buf_write, 0, 1, n); \ +} while (0) + +/* Run memset tests for constant length */ +#define ALL_PERF_TEST_FOR_CONSTANT \ +do { \ + TEST_CONSTANT(6U); TEST_CONSTANT(64U); TEST_CONSTANT(128U); \ + TEST_CONSTANT(192U); TEST_CONSTANT(256U); TEST_CONSTANT(512U); \ + TEST_CONSTANT(768U); TEST_CONSTANT(1024U); TEST_CONSTANT(1536U); \ +} while (0) + +/* Run all memset tests for aligned constant cases */ +static inline void +perf_test_constant_aligned(void) +{ +#define TEST_CONSTANT ALL_PERF_TESTS_FOR_SIZE + ALL_PERF_TEST_FOR_CONSTANT; +#undef TEST_CONSTANT +} + +/* Run all memset tests for unaligned constant cases */ +static inline void +perf_test_constant_unaligned(void) +{ +#define TEST_CONSTANT ALL_PERF_TESTS_FOR_SIZE_UNALIGNED + ALL_PERF_TEST_FOR_CONSTANT; +#undef TEST_CONSTANT +} + +/* Run all memset tests for aligned variable cases */ +static inline void +perf_test_variable_aligned(void) +{ + unsigned int n = sizeof(buf_sizes) / sizeof(buf_sizes[0]); + unsigned int i; + + for (i = 0; i < n; i++) + ALL_PERF_TESTS_FOR_SIZE((size_t)buf_sizes[i]); +} + +/* Run all memset tests for unaligned variable cases */ +static inline void +perf_test_variable_unaligned(void) +{ + unsigned int n = sizeof(buf_sizes) / sizeof(buf_sizes[0]); + unsigned int i; + + for (i = 0; i < n; i++) + ALL_PERF_TESTS_FOR_SIZE_UNALIGNED((size_t)buf_sizes[i]); +} + +/* Run all memset tests */ +static int +perf_test(void) +{ + int ret; + + ret = init_buffers(); + if (ret != 0) + return ret; + +#if TEST_VALUE_RANGE != 0 + /* Set up buf_sizes array, if required */ + unsigned int i; + + for (i = 0; i < TEST_VALUE_RANGE; i++) + buf_sizes[i] = i; +#endif + + /* See function comment */ + do_uncached_write(large_buf_write, 0, SMALL_BUFFER_SIZE); + + printf("\n** rte_memset() - memset perf tests \t\n \ + (C = compile-time constant) **\n" + "======== ======= ======== ======= ========\n" + " Size memset in cache memset in mem\n" + "(bytes) (ticks) (ticks)\n" + "------- -------------- ---------------"); + + printf("\n============= %2dB aligned ================", ALIGNMENT_UNIT); + /* Do aligned tests where size is a variable */ + perf_test_variable_aligned(); + printf("\n------ -------------- -------------- ------"); + /* Do aligned tests where size is a compile-time constant */ + perf_test_constant_aligned(); + printf("\n============= Unaligned ==================="); + /* Do unaligned tests where size is a variable */ + perf_test_variable_unaligned(); + printf("\n------ -------------- -------------- ------"); + /* Do unaligned tests where size is a compile-time constant */ + perf_test_constant_unaligned(); + printf("\n====== ============== ============== =======\n\n"); + + free_buffers(); + + return 0; +} + +static int +test_memset_perf(void) +{ + int ret; + + ret = perf_test(); + if (ret != 0) + return -1; + return 0; +} + +REGISTER_TEST_COMMAND(memset_perf_autotest, test_memset_perf); -- 2.7.4