[fio.git] / stat.h

#ifndef FIO_STAT_H
#define FIO_STAT_H

struct group_run_stats {
	uint64_t max_run[DDIR_RWDIR_CNT], min_run[DDIR_RWDIR_CNT];
	uint64_t max_bw[DDIR_RWDIR_CNT], min_bw[DDIR_RWDIR_CNT];
	uint64_t io_kb[DDIR_RWDIR_CNT];
	uint64_t agg[DDIR_RWDIR_CNT];
	uint32_t kb_base;
	uint32_t unit_base;
	uint32_t groupid;
	uint32_t unified_rw_rep;
};

/*
 * How many depth levels to log
 */
#define FIO_IO_U_MAP_NR	7
#define FIO_IO_U_LAT_U_NR 10
#define FIO_IO_U_LAT_M_NR 12

/*
 * Aggregate clat samples to report percentile(s) of them.
 *
 * EXECUTIVE SUMMARY
 *
 * FIO_IO_U_PLAT_BITS determines the maximum statistical error on the
 * value of resulting percentiles. The error will be approximately
 * 1/2^(FIO_IO_U_PLAT_BITS+1) of the value.
 *
 * FIO_IO_U_PLAT_GROUP_NR and FIO_IO_U_PLAT_BITS determine the maximum
 * range being tracked for latency samples. The maximum value tracked
 * accurately will be 2^(GROUP_NR + PLAT_BITS -1) microseconds.
 *
 * FIO_IO_U_PLAT_GROUP_NR and FIO_IO_U_PLAT_BITS determine the memory
 * requirement of storing those aggregate counts. The memory used will
 * be (FIO_IO_U_PLAT_GROUP_NR * 2^FIO_IO_U_PLAT_BITS) * sizeof(int)
 * bytes.
 *
 * FIO_IO_U_PLAT_NR is the total number of buckets.
 *
 * DETAILS
 *
 * Suppose the clat varies from 0 to 999 (usec), the straightforward
 * method is to keep an array of (999 + 1) buckets, in which a counter
 * keeps the count of samples which fall in the bucket, e.g.,
 * {[0],[1],...,[999]}. However this consumes a huge amount of space,
 * and can be avoided if an approximation is acceptable.
 *
 * One such method is to let the range of the bucket to be greater
 * than one. This method has low accuracy when the value is small. For
 * example, let the buckets be {[0,99],[100,199],...,[900,999]}, and
 * the represented value of each bucket be the mean of the range. Then
 * a value 0 has an round-off error of 49.5. To improve on this, we
 * use buckets with non-uniform ranges, while bounding the error of
 * each bucket within a ratio of the sample value. A simple example
 * would be when error_bound = 0.005, buckets are {
 * {[0],[1],...,[99]}, {[100,101],[102,103],...,[198,199]},..,
 * {[900,909],[910,919]...}  }. The total range is partitioned into
 * groups with different ranges, then buckets with uniform ranges. An
 * upper bound of the error is (range_of_bucket/2)/value_of_bucket
 *
 * For better efficiency, we implement this using base two. We group
 * samples by their Most Significant Bit (MSB), extract the next M bit
 * of them as an index within the group, and discard the rest of the
 * bits.
 *
 * E.g., assume a sample 'x' whose MSB is bit n (starting from bit 0),
 * and use M bit for indexing
 *
 *        | n |    M bits   | bit (n-M-1) ... bit 0 |
 *
 * Because x is at least 2^n, and bit 0 to bit (n-M-1) is at most
 * (2^(n-M) - 1), discarding bit 0 to (n-M-1) makes the round-off
 * error
 *
 *           2^(n-M)-1    2^(n-M)    1
 *      e <= --------- <= ------- = ---
 *             2^n          2^n     2^M
 *
 * Furthermore, we use "mean" of the range to represent the bucket,
 * the error e can be lowered by half to 1 / 2^(M+1). By using M bits
 * as the index, each group must contains 2^M buckets.
 *
 * E.g. Let M (FIO_IO_U_PLAT_BITS) be 6
 *      Error bound is 1/2^(6+1) = 0.0078125 (< 1%)
 *
 *	Group	MSB	#discarded	range of		#buckets
 *			error_bits	value
 *	----------------------------------------------------------------
 *	0*	0~5	0		[0,63]			64
 *	1*	6	0		[64,127]		64
 *	2	7	1		[128,255]		64
 *	3	8	2		[256,511]		64
 *	4	9	3		[512,1023]		64
 *	...	...	...		[...,...]		...
 *	18	23	17		[8838608,+inf]**	64
 *
 *  * Special cases: when n < (M-1) or when n == (M-1), in both cases,
 *    the value cannot be rounded off. Use all bits of the sample as
 *    index.
 *
 *  ** If a sample's MSB is greater than 23, it will be counted as 23.
 */

#define FIO_IO_U_PLAT_BITS 6
#define FIO_IO_U_PLAT_VAL (1 << FIO_IO_U_PLAT_BITS)
#define FIO_IO_U_PLAT_GROUP_NR 19
#define FIO_IO_U_PLAT_NR (FIO_IO_U_PLAT_GROUP_NR * FIO_IO_U_PLAT_VAL)
#define FIO_IO_U_LIST_MAX_LEN 20 /* The size of the default and user-specified
					list of percentiles */

#define MAX_PATTERN_SIZE	512
#define FIO_JOBNAME_SIZE	128
#define FIO_VERROR_SIZE		128

struct thread_stat {
	char name[FIO_JOBNAME_SIZE];
	char verror[FIO_VERROR_SIZE];
	uint32_t error;
	uint32_t groupid;
	uint32_t pid;
	char description[FIO_JOBNAME_SIZE];
	uint32_t members;
	uint32_t unified_rw_rep;

	/*
	 * bandwidth and latency stats
	 */
	struct io_stat clat_stat[DDIR_RWDIR_CNT]; /* completion latency */
	struct io_stat slat_stat[DDIR_RWDIR_CNT]; /* submission latency */
	struct io_stat lat_stat[DDIR_RWDIR_CNT]; /* total latency */
	struct io_stat bw_stat[DDIR_RWDIR_CNT]; /* bandwidth stats */
	struct io_stat iops_stat[DDIR_RWDIR_CNT]; /* IOPS stats */

	/*
	 * fio system usage accounting
	 */
	uint64_t usr_time;
	uint64_t sys_time;
	uint64_t ctx;
	uint64_t minf, majf;

	/*
	 * IO depth and latency stats
	 */
	uint64_t clat_percentiles;
	uint64_t percentile_precision;
	fio_fp64_t percentile_list[FIO_IO_U_LIST_MAX_LEN];

	uint32_t io_u_map[FIO_IO_U_MAP_NR];
	uint32_t io_u_submit[FIO_IO_U_MAP_NR];
	uint32_t io_u_complete[FIO_IO_U_MAP_NR];
	uint32_t io_u_lat_u[FIO_IO_U_LAT_U_NR];
	uint32_t io_u_lat_m[FIO_IO_U_LAT_M_NR];
	uint32_t io_u_plat[DDIR_RWDIR_CNT][FIO_IO_U_PLAT_NR];
	uint64_t total_io_u[3];
	uint64_t short_io_u[3];
	uint64_t total_submit;
	uint64_t total_complete;

	uint64_t io_bytes[DDIR_RWDIR_CNT];
	uint64_t runtime[DDIR_RWDIR_CNT];
	uint64_t total_run_time;

	/*
	 * IO Error related stats
	 */
	uint16_t continue_on_error;
	uint64_t total_err_count;
	uint32_t first_error;

	uint32_t kb_base;
	uint32_t unit_base;
};

struct jobs_eta {
	uint32_t nr_running;
	uint32_t nr_ramp;
	uint32_t nr_pending;
	uint32_t files_open;
	uint32_t m_rate, t_rate;
	uint32_t m_iops, t_iops;
	uint32_t rate[DDIR_RWDIR_CNT];
	uint32_t iops[DDIR_RWDIR_CNT];
	uint64_t elapsed_sec;
	uint64_t eta_sec;
	uint32_t is_pow2;
	uint32_t unit_base;

	/*
	 * Network 'copy' of run_str[]
	 */
	uint32_t nr_threads;
	uint8_t run_str[];
};

extern void show_thread_status(struct thread_stat *ts, struct group_run_stats *rs);
extern void show_group_stats(struct group_run_stats *rs);
extern int calc_thread_status(struct jobs_eta *je, int force);
extern void display_thread_status(struct jobs_eta *je);
extern void show_run_stats(void);
extern void show_running_run_stats(void);
extern void sum_thread_stats(struct thread_stat *dst, struct thread_stat *src, int nr);
extern void sum_group_stats(struct group_run_stats *dst, struct group_run_stats *src);
extern void init_thread_stat(struct thread_stat *ts);
extern void init_group_run_stat(struct group_run_stats *gs);

#endif
Commit	Line	Data
a64e88da JA	1	#ifndef FIO_STAT_H
	2	#define FIO_STAT_H
	3
	4	struct group_run_stats {
6eaf09d6 SL	5	uint64_t max_run[DDIR_RWDIR_CNT], min_run[DDIR_RWDIR_CNT];
	6	uint64_t max_bw[DDIR_RWDIR_CNT], min_bw[DDIR_RWDIR_CNT];
	7	uint64_t io_kb[DDIR_RWDIR_CNT];
	8	uint64_t agg[DDIR_RWDIR_CNT];
a64e88da	9	uint32_t kb_base;
ad705bcb	10	uint32_t unit_base;
a64e88da	11	uint32_t groupid;
771e58be	12	uint32_t unified_rw_rep;
a64e88da JA	13	};
	14
	15	/*
	16	* How many depth levels to log
	17	*/
	18	#define FIO_IO_U_MAP_NR 7
	19	#define FIO_IO_U_LAT_U_NR 10
	20	#define FIO_IO_U_LAT_M_NR 12
	21
	22	/*
	23	* Aggregate clat samples to report percentile(s) of them.
	24	*
	25	* EXECUTIVE SUMMARY
	26	*
	27	* FIO_IO_U_PLAT_BITS determines the maximum statistical error on the
	28	* value of resulting percentiles. The error will be approximately
	29	* 1/2^(FIO_IO_U_PLAT_BITS+1) of the value.
	30	*
	31	* FIO_IO_U_PLAT_GROUP_NR and FIO_IO_U_PLAT_BITS determine the maximum
	32	* range being tracked for latency samples. The maximum value tracked
	33	* accurately will be 2^(GROUP_NR + PLAT_BITS -1) microseconds.
	34	*
	35	* FIO_IO_U_PLAT_GROUP_NR and FIO_IO_U_PLAT_BITS determine the memory
	36	* requirement of storing those aggregate counts. The memory used will
	37	* be (FIO_IO_U_PLAT_GROUP_NR * 2^FIO_IO_U_PLAT_BITS) * sizeof(int)
	38	* bytes.
	39	*
	40	* FIO_IO_U_PLAT_NR is the total number of buckets.
	41	*
	42	* DETAILS
	43	*
	44	* Suppose the clat varies from 0 to 999 (usec), the straightforward
	45	* method is to keep an array of (999 + 1) buckets, in which a counter
	46	* keeps the count of samples which fall in the bucket, e.g.,
	47	* {[0],[1],...,[999]}. However this consumes a huge amount of space,
	48	* and can be avoided if an approximation is acceptable.
	49	*
	50	* One such method is to let the range of the bucket to be greater
	51	* than one. This method has low accuracy when the value is small. For
	52	* example, let the buckets be {[0,99],[100,199],...,[900,999]}, and
	53	* the represented value of each bucket be the mean of the range. Then
	54	* a value 0 has an round-off error of 49.5. To improve on this, we
	55	* use buckets with non-uniform ranges, while bounding the error of
	56	* each bucket within a ratio of the sample value. A simple example
	57	* would be when error_bound = 0.005, buckets are {
	58	* {[0],[1],...,[99]}, {[100,101],[102,103],...,[198,199]},..,
	59	* {[900,909],[910,919]...} }. The total range is partitioned into
	60	* groups with different ranges, then buckets with uniform ranges. An
	61	* upper bound of the error is (range_of_bucket/2)/value_of_bucket
	62	*
	63	* For better efficiency, we implement this using base two. We group
	64	* samples by their Most Significant Bit (MSB), extract the next M bit
	65	* of them as an index within the group, and discard the rest of the
	66	* bits.
	67	*
	68	* E.g., assume a sample 'x' whose MSB is bit n (starting from bit 0),
	69	* and use M bit for indexing
	70	*
	71	* \| n \| M bits \| bit (n-M-1) ... bit 0 \|
	72	*
	73	* Because x is at least 2^n, and bit 0 to bit (n-M-1) is at most
	74	* (2^(n-M) - 1), discarding bit 0 to (n-M-1) makes the round-off
	75	* error
	76	*
77	* 2^(n-M)-1 2^(n-M) 1
78	* e <= --------- <= ------- = ---
79	* 2^n 2^n 2^M
80	*
81	* Furthermore, we use "mean" of the range to represent the bucket,
82	* the error e can be lowered by half to 1 / 2^(M+1). By using M bits
83	* as the index, each group must contains 2^M buckets.
84	*
85	* E.g. Let M (FIO_IO_U_PLAT_BITS) be 6
86	* Error bound is 1/2^(6+1) = 0.0078125 (< 1%)
87	*
88	* Group MSB #discarded range of #buckets
89	* error_bits value
90	* ----------------------------------------------------------------
91	* 0* 0~5 0 [0,63] 64
92	* 1* 6 0 [64,127] 64
93	* 2 7 1 [128,255] 64
94	* 3 8 2 [256,511] 64
95	* 4 9 3 [512,1023] 64
96	* ... ... ... [...,...] ...
97	* 18 23 17 [8838608,+inf]** 64
98	*
99	* * Special cases: when n < (M-1) or when n == (M-1), in both cases,
100	* the value cannot be rounded off. Use all bits of the sample as
101	* index.
102	*
103	* ** If a sample's MSB is greater than 23, it will be counted as 23.
104	*/
105
106	#define FIO_IO_U_PLAT_BITS 6
107	#define FIO_IO_U_PLAT_VAL (1 << FIO_IO_U_PLAT_BITS)
108	#define FIO_IO_U_PLAT_GROUP_NR 19
109	#define FIO_IO_U_PLAT_NR (FIO_IO_U_PLAT_GROUP_NR * FIO_IO_U_PLAT_VAL)
110	#define FIO_IO_U_LIST_MAX_LEN 20 /* The size of the default and user-specified
111	list of percentiles */
112
113	#define MAX_PATTERN_SIZE 512
114	#define FIO_JOBNAME_SIZE 128
115	#define FIO_VERROR_SIZE 128
116
117	struct thread_stat {
118	char name[FIO_JOBNAME_SIZE];
119	char verror[FIO_VERROR_SIZE];
ddcc0b69 JA	120	uint32_t error;
ddcc0b69 JA	121	uint32_t groupid;
a64e88da JA	122	uint32_t pid;
	123	char description[FIO_JOBNAME_SIZE];
	124	uint32_t members;
771e58be	125	uint32_t unified_rw_rep;
a64e88da JA	126
	127	/*
	128	* bandwidth and latency stats
	129	*/
6eaf09d6 SL	130	struct io_stat clat_stat[DDIR_RWDIR_CNT]; /* completion latency */
	131	struct io_stat slat_stat[DDIR_RWDIR_CNT]; /* submission latency */
	132	struct io_stat lat_stat[DDIR_RWDIR_CNT]; /* total latency */
	133	struct io_stat bw_stat[DDIR_RWDIR_CNT]; /* bandwidth stats */
	134	struct io_stat iops_stat[DDIR_RWDIR_CNT]; /* IOPS stats */
a64e88da JA	135
	136	/*
	137	* fio system usage accounting
	138	*/
	139	uint64_t usr_time;
	140	uint64_t sys_time;
	141	uint64_t ctx;
	142	uint64_t minf, majf;
	143
	144	/*
	145	* IO depth and latency stats
	146	*/
	147	uint64_t clat_percentiles;
435d195a	148	uint64_t percentile_precision;
802ad4a8	149	fio_fp64_t percentile_list[FIO_IO_U_LIST_MAX_LEN];
a64e88da JA	150
	151	uint32_t io_u_map[FIO_IO_U_MAP_NR];
	152	uint32_t io_u_submit[FIO_IO_U_MAP_NR];
	153	uint32_t io_u_complete[FIO_IO_U_MAP_NR];
	154	uint32_t io_u_lat_u[FIO_IO_U_LAT_U_NR];
	155	uint32_t io_u_lat_m[FIO_IO_U_LAT_M_NR];
6eaf09d6	156	uint32_t io_u_plat[DDIR_RWDIR_CNT][FIO_IO_U_PLAT_NR];
a64e88da JA	157	uint64_t total_io_u[3];
	158	uint64_t short_io_u[3];
	159	uint64_t total_submit;
	160	uint64_t total_complete;
	161
6eaf09d6 SL	162	uint64_t io_bytes[DDIR_RWDIR_CNT];
6eaf09d6 SL	163	uint64_t runtime[DDIR_RWDIR_CNT];
a64e88da JA	164	uint64_t total_run_time;
	165
	166	/*
	167	* IO Error related stats
	168	*/
	169	uint16_t continue_on_error;
	170	uint64_t total_err_count;
ddcc0b69	171	uint32_t first_error;
a64e88da JA	172
a64e88da JA	173	uint32_t kb_base;
ad705bcb	174	uint32_t unit_base;
a64e88da JA	175	};
a64e88da JA	176
b75a394f JA	177	struct jobs_eta {
	178	uint32_t nr_running;
	179	uint32_t nr_ramp;
	180	uint32_t nr_pending;
	181	uint32_t files_open;
	182	uint32_t m_rate, t_rate;
	183	uint32_t m_iops, t_iops;
6eaf09d6 SL	184	uint32_t rate[DDIR_RWDIR_CNT];
6eaf09d6 SL	185	uint32_t iops[DDIR_RWDIR_CNT];
b75a394f JA	186	uint64_t elapsed_sec;
b75a394f JA	187	uint64_t eta_sec;
b7f05eb0	188	uint32_t is_pow2;
ad705bcb	189	uint32_t unit_base;
1d1f45ae JA	190
	191	/*
	192	* Network 'copy' of run_str[]
	193	*/
	194	uint32_t nr_threads;
372aecb9	195	uint8_t run_str[];
b75a394f JA	196	};
b75a394f JA	197
a64e88da JA	198	extern void show_thread_status(struct thread_stat ts, struct group_run_stats rs);
a64e88da JA	199	extern void show_group_stats(struct group_run_stats *rs);
af9c9fb3	200	extern int calc_thread_status(struct jobs_eta *je, int force);
cf451d1e	201	extern void display_thread_status(struct jobs_eta *je);
5b9babb7	202	extern void show_run_stats(void);
4c6d91e8	203	extern void show_running_run_stats(void);
5b9babb7	204	extern void sum_thread_stats(struct thread_stat dst, struct thread_stat src, int nr);
37f0c1ae JA	205	extern void sum_group_stats(struct group_run_stats dst, struct group_run_stats src);
	206	extern void init_thread_stat(struct thread_stat *ts);
	207	extern void init_group_run_stat(struct group_run_stats *gs);
a64e88da JA	208
a64e88da JA	209	#endif