From: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
To: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: igt-dev@lists.freedesktop.org
Subject: Re: [igt-dev] [PATCH i-g-t 16/23] i915/perf: Treat ticks as 64 bit
Date: Tue, 6 Sep 2022 11:56:12 -0700 [thread overview]
Message-ID: <YxeXzC2JjhqApkPY@unerlige-ril> (raw)
In-Reply-To: <db5a5463-18c7-3a2a-9ecb-b07279b408c7@intel.com>
On Tue, Sep 06, 2022 at 05:08:50PM +0300, Lionel Landwerlin wrote:
>On 23/08/2022 21:30, Umesh Nerlige Ramappa wrote:
>>DG2 A0 has a buggy header in the 64 bit OAG format. Except the ticks,
>>all other fields are 32 bit. To resolve this, we will store ticks in 64
>>bit type.
>>
>>Signed-off-by: Umesh Nerlige Ramappa <umesh.nerlige.ramappa@intel.com>
>>---
>> tests/i915/perf.c | 82 ++++++++++++++++++++++++++++++++---------------
>> 1 file changed, 56 insertions(+), 26 deletions(-)
>>
>>diff --git a/tests/i915/perf.c b/tests/i915/perf.c
>>index 3eb31f3b..8836da66 100644
>>--- a/tests/i915/perf.c
>>+++ b/tests/i915/perf.c
>>@@ -117,6 +117,7 @@ struct oa_format {
>> int c_off;
>> int n_c;
>> int oa_type;
>>+ bool report_hdr_64bit;
>> };
>> static struct oa_format hsw_oa_formats[I915_OA_FORMAT_MAX] = {
>>@@ -203,7 +204,8 @@ static struct oa_format dg2_oa_formats[I915_OA_FORMAT_MAX] = {
>> "A38u64_R2u64_B8_C8", .size = 448,
>> .a64_off = 32, .n_a64 = 38,
>> .b_off = 352, .n_b = 8,
>>- .c_off = 384, .n_c = 8, .oa_type = OAG, },
>>+ .c_off = 384, .n_c = 8, .oa_type = OAG,
>>+ .report_hdr_64bit = true, },
>> [I915_OAR_FORMAT_A36u64_B8_C8] = {
>> "A36u64_B8_C8", .size = 384,
>> .a64_off = 32, .n_a64 = 36,
>>@@ -245,7 +247,7 @@ static bool *undefined_a_counters;
>> static uint64_t oa_exp_1_millisec;
>> static igt_render_copyfunc_t render_copy = NULL;
>>-static uint32_t (*read_report_ticks)(const uint32_t *report,
>>+static uint64_t (*read_report_ticks)(const uint32_t *report,
>> enum drm_i915_oa_format format);
>> static void (*sanity_check_reports)(const uint32_t *oa_report0,
>> const uint32_t *oa_report1,
>>@@ -395,7 +397,7 @@ sysfs_read(enum i915_attr_id id)
>> * C2 corresponds to a clock counter for the Haswell render basic metric set
>> * but it's not included in all of the formats.
>> */
>>-static uint32_t
>>+static uint64_t
>> hsw_read_report_ticks(const uint32_t *report, enum drm_i915_oa_format format)
>> {
>> uint32_t *c = (uint32_t *)(((uint8_t *)report) + get_oa_format(format).c_off);
>>@@ -405,10 +407,41 @@ hsw_read_report_ticks(const uint32_t *report, enum drm_i915_oa_format format)
>> return c[2];
>> }
>>-static uint32_t
>>+static uint64_t
>> gen8_read_report_ticks(const uint32_t *report, enum drm_i915_oa_format format)
>> {
>>- return report[3];
>>+
>>+ struct oa_format fmt = get_oa_format(format);
>>+
>>+ return fmt.report_hdr_64bit ? *(uint64_t *)&report[6] : report[3];
>>+}
>>+
>>+/*
>>+ * t0 is a value sampled before t1. width is number of bits used to represent
>>+ * t0/t1. Normally t1 is greater than t0. In cases where t1 < t0 use this
>>+ * helper. Since the size of t1/t0 is already 64 bits, no special handling is
>>+ * needed for width = 64.
>>+ */
>>+static uint64_t
>>+__elapsed_delta(uint64_t t1, uint64_t t0, uint32_t width)
>>+{
>>+ uint32_t max_bits = sizeof(t1) * 8;
>>+
>>+ igt_assert(width <= max_bits);
>>+
>>+ if (t1 < t0 && width != max_bits)
>>+ return ((1ULL << width) - t0) + t1;
>
>
>Should this be : ((1ULL << width) + t1) - t0 ?
>
Both ways should yield the same result.
>
>Since we know the t1 wrapped around the counter's max bit size, the
>actual difference is the max value + whatever is in the counter minus
>t0.
>
>
>>+
>>+ return t1 - t0;
>>+}
>>+
>>+static uint64_t
>>+oa_tick_delta(const uint32_t *report1,
>>+ const uint32_t *report0,
>>+ enum drm_i915_oa_format format)
>>+{
>>+ return __elapsed_delta(read_report_ticks(report1, format),
>>+ read_report_ticks(report0, format), 32);
>> }
>> static void
>>@@ -638,8 +671,8 @@ hsw_sanity_check_render_basic_reports(const uint32_t *oa_report0,
>> enum drm_i915_oa_format fmt)
>> {
>> uint32_t time_delta = timebase_scale(oa_report1[1] - oa_report0[1]);
>>- uint32_t clock_delta;
>>- uint32_t max_delta;
>>+ uint64_t clock_delta;
>>+ uint64_t max_delta;
>> struct oa_format format = get_oa_format(fmt);
>> igt_assert_neq(time_delta, 0);
>>@@ -654,21 +687,19 @@ hsw_sanity_check_render_basic_reports(const uint32_t *oa_report0,
>> clock_delta = (gt_max_freq_mhz *
>> (uint64_t)time_delta) / 1000;
>> } else {
>>- uint32_t ticks0 = read_report_ticks(oa_report0, fmt);
>>- uint32_t ticks1 = read_report_ticks(oa_report1, fmt);
>> uint64_t freq;
>>- clock_delta = ticks1 - ticks0;
>>+ clock_delta = oa_tick_delta(oa_report1, oa_report0, fmt);
>>- igt_assert_neq(clock_delta, 0);
>>+ igt_assert_neq_u64(clock_delta, 0);
>>- freq = ((uint64_t)clock_delta * 1000) / time_delta;
>>+ freq = (clock_delta * 1000) / time_delta;
>> igt_debug("freq = %"PRIu64"\n", freq);
>> igt_assert(freq <= gt_max_freq_mhz);
>> }
>>- igt_debug("clock delta = %"PRIu32"\n", clock_delta);
>>+ igt_debug("clock delta = %"PRIu64"\n", clock_delta);
>> /* The maximum rate for any HSW counter =
>> * clock_delta * N EUs
>>@@ -801,7 +832,8 @@ accumulate_reports(struct accumulator *accumulator,
>> accumulate_uint32(4, start, end, deltas + idx++);
>> /* clock cycles */
>>- accumulate_uint32(12, start, end, deltas + idx++);
>>+ deltas[idx] += oa_tick_delta(end, start, accumulator->format);
>>+ idx++;
>> } else {
>> /* timestamp */
>> accumulate_uint32(4, start, end, deltas + idx++);
>>@@ -874,10 +906,8 @@ gen8_sanity_check_test_oa_reports(const uint32_t *oa_report0,
>> {
>> struct oa_format format = get_oa_format(fmt);
>> uint32_t time_delta = timebase_scale(oa_report1[1] - oa_report0[1]);
>>- uint32_t ticks0 = read_report_ticks(oa_report0, fmt);
>>- uint32_t ticks1 = read_report_ticks(oa_report1, fmt);
>>- uint32_t clock_delta = ticks1 - ticks0;
>>- uint32_t max_delta;
>>+ uint64_t clock_delta = oa_tick_delta(oa_report1, oa_report0, fmt);
>>+ uint64_t max_delta;
>> uint64_t freq;
>> uint32_t *rpt0_b = (uint32_t *)(((uint8_t *)oa_report0) +
>> format.b_off);
>>@@ -890,10 +920,10 @@ gen8_sanity_check_test_oa_reports(const uint32_t *oa_report0,
>> gen8_read_report_reason(oa_report0),
>> gen8_read_report_reason(oa_report1));
>>- freq = time_delta ? ((uint64_t)clock_delta * 1000) / time_delta : 0;
>>+ freq = time_delta ? (clock_delta * 1000) / time_delta : 0;
>> igt_debug("freq = %"PRIu64"\n", freq);
>>- igt_debug("clock delta = %"PRIu32"\n", clock_delta);
>>+ igt_debug("clock delta = %"PRIu64"\n", clock_delta);
>> max_delta = clock_delta * intel_perf->devinfo.n_eus;
>>@@ -994,7 +1024,7 @@ gen8_sanity_check_test_oa_reports(const uint32_t *oa_report0,
>> format.c_off);
>> uint32_t delta = c1[j] - c0[j];
>>- igt_debug("C%d: delta = %"PRIu32", max_delta=%"PRIu32"\n",
>>+ igt_debug("C%d: delta = %"PRIu32", max_delta=%"PRIu64"\n",
>> j, delta, max_delta);
>> igt_assert(delta <= max_delta);
>> }
>>@@ -1440,10 +1470,10 @@ print_reports(uint32_t *oa_report0, uint32_t *oa_report1, int fmt)
>> if (IS_HASWELL(devid) && format.n_c == 0) {
>> igt_debug("CLOCK = N/A\n");
>> } else {
>>- uint32_t clock0 = read_report_ticks(oa_report0, fmt);
>>- uint32_t clock1 = read_report_ticks(oa_report1, fmt);
>>+ uint64_t clock0 = read_report_ticks(oa_report0, fmt);
>>+ uint64_t clock1 = read_report_ticks(oa_report1, fmt);
>>- igt_debug("CLOCK: 1st = %"PRIu32", 2nd = %"PRIu32", delta = %"PRIu32"\n",
>>+ igt_debug("CLOCK: 1st = %"PRIu64", 2nd = %"PRIu64", delta = %"PRIu64"\n",
>> clock0, clock1, clock1 - clock0);
>> }
>>@@ -1545,9 +1575,9 @@ print_report(uint32_t *report, int fmt)
>> if (IS_HASWELL(devid) && format.n_c == 0) {
>> igt_debug("CLOCK = N/A\n");
>> } else {
>>- uint32_t clock = read_report_ticks(report, fmt);
>>+ uint64_t clock = read_report_ticks(report, fmt);
>>- igt_debug("CLOCK: %"PRIu32"\n", clock);
>>+ igt_debug("CLOCK: %"PRIu64"\n", clock);
>> }
>> if (intel_gen(devid) >= 8) {
>
>
next prev parent reply other threads:[~2022-09-06 18:56 UTC|newest]
Thread overview: 66+ messages / expand[flat|nested] mbox.gz Atom feed top
2022-08-23 18:30 [igt-dev] [PATCH i-g-t 00/23] Add DG2 OA test Umesh Nerlige Ramappa
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 01/23] i915/perf: Check regularly if we are done reading reports Umesh Nerlige Ramappa
2022-09-06 12:49 ` Lionel Landwerlin
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 02/23] i915/perf: Fix OA short_reads test Umesh Nerlige Ramappa
2022-09-06 12:50 ` Lionel Landwerlin
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 03/23] i915/perf: Check return value from getparam Umesh Nerlige Ramappa
2022-09-06 12:52 ` Lionel Landwerlin
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 04/23] i915/perf: Limit sseu-config tests for gen11 Umesh Nerlige Ramappa
2022-09-06 12:53 ` Lionel Landwerlin
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 05/23] i915/perf: Bump timestamp tolerance for DG1 Umesh Nerlige Ramappa
2022-09-06 13:06 ` Lionel Landwerlin
2022-09-06 19:30 ` Umesh Nerlige Ramappa
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 06/23] i915/perf: Account for OA sampling interval in polling test Umesh Nerlige Ramappa
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 07/23] i915/perf: Add support for 64-bit counters Umesh Nerlige Ramappa
2022-09-06 13:05 ` Lionel Landwerlin
2022-09-06 19:37 ` Umesh Nerlige Ramappa
2022-09-06 20:02 ` Lionel Landwerlin
2022-09-06 20:58 ` Umesh Nerlige Ramappa
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 08/23] i915/perf: Define OA report types Umesh Nerlige Ramappa
2022-09-06 13:08 ` Lionel Landwerlin
2022-09-06 19:28 ` Umesh Nerlige Ramappa
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 09/23] i915/perf: Use ARRAY_SIZE consistently for num_properties Umesh Nerlige Ramappa
2022-09-06 13:10 ` Lionel Landwerlin
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 10/23] i915/perf: Use gt in perf tests and lib Umesh Nerlige Ramappa
2022-09-06 13:13 ` Lionel Landwerlin
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 11/23] i915/perf: Explicitly state rendercopy needs for a test Umesh Nerlige Ramappa
2022-09-06 13:26 ` Lionel Landwerlin
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 12/23] i915/perf: Skip tests that use rendercopy Umesh Nerlige Ramappa
2022-09-06 13:28 ` Lionel Landwerlin
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 13/23] i915/perf: Add OA formats for DG2 Umesh Nerlige Ramappa
2022-09-06 13:30 ` Lionel Landwerlin
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 14/23] i915/perf: Add a test for non-power-of-2 oa reports Umesh Nerlige Ramappa
2022-09-06 14:08 ` Lionel Landwerlin
2022-09-20 19:28 ` Umesh Nerlige Ramappa
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 15/23] i915/perf: Fix CS timestamp vs OA timstamp mismatch Umesh Nerlige Ramappa
2022-09-06 13:36 ` Lionel Landwerlin
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 16/23] i915/perf: Treat ticks as 64 bit Umesh Nerlige Ramappa
2022-09-06 14:08 ` Lionel Landwerlin
2022-09-06 18:56 ` Umesh Nerlige Ramappa [this message]
2022-09-06 19:06 ` Lionel Landwerlin
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 17/23] i915/perf: Treat timestamp as 64 bit value Umesh Nerlige Ramappa
2022-09-06 14:16 ` Lionel Landwerlin
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 18/23] i915/perf: Fix DG2 A0 report header Umesh Nerlige Ramappa
2022-09-06 14:17 ` Lionel Landwerlin
2022-09-06 18:37 ` Umesh Nerlige Ramappa
2022-09-06 18:54 ` Lionel Landwerlin
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 19/23] i915/perf: Wait longer for rc6 residency in DG2 Umesh Nerlige Ramappa
2022-09-06 14:18 ` Lionel Landwerlin
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 20/23] lib/i915/perf: implement report accumulation for new format Umesh Nerlige Ramappa
2022-09-06 14:21 ` Lionel Landwerlin
2022-09-20 20:02 ` Umesh Nerlige Ramappa
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 21/23] lib/i915/perf: fixup conversion script for XEHPSDV Umesh Nerlige Ramappa
2022-09-20 20:03 ` Umesh Nerlige Ramappa
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 22/23] lib/i915/perf: make warning message more helpful Umesh Nerlige Ramappa
2022-09-20 20:04 ` Umesh Nerlige Ramappa
2022-08-23 18:30 ` [igt-dev] [PATCH i-g-t 23/23] lib/i915/perf: Add DG2 metrics Umesh Nerlige Ramappa
2022-09-20 23:11 ` Umesh Nerlige Ramappa
2022-08-23 20:30 ` [igt-dev] [PATCH i-g-t 18/23] i915/perf: Fix DG2 A0 report header Umesh Nerlige Ramappa
2022-09-06 14:23 ` Lionel Landwerlin
2022-09-06 18:33 ` Umesh Nerlige Ramappa
2022-08-23 21:00 ` [igt-dev] ✗ Fi.CI.BUILD: failure for Add DG2 OA test Patchwork
2022-08-24 9:33 ` [igt-dev] ✗ Fi.CI.BUILD: failure for Add DG2 OA test (rev2) Patchwork
2022-08-24 9:58 ` Petri Latvala
2022-08-24 13:08 ` [igt-dev] ✓ Fi.CI.BAT: success for Add DG2 OA test (rev3) Patchwork
2022-08-25 20:33 ` [igt-dev] ✗ Fi.CI.IGT: failure " Patchwork
-- strict thread matches above, loose matches on Subject: below --
2022-08-22 23:56 [igt-dev] [PATCH i-g-t 00/23] Add DG2 OA test Umesh Nerlige Ramappa
2022-08-22 23:56 ` [igt-dev] [PATCH i-g-t 16/23] i915/perf: Treat ticks as 64 bit Umesh Nerlige Ramappa
Reply instructions:
You may reply publicly to this message via plain-text email
using any one of the following methods:
* Save the following mbox file, import it into your mail client,
and reply-to-all from there: mbox
Avoid top-posting and favor interleaved quoting:
https://en.wikipedia.org/wiki/Posting_style#Interleaved_style
* Reply using the --to, --cc, and --in-reply-to
switches of git-send-email(1):
git send-email \
--in-reply-to=YxeXzC2JjhqApkPY@unerlige-ril \
--to=umesh.nerlige.ramappa@intel.com \
--cc=igt-dev@lists.freedesktop.org \
--cc=lionel.g.landwerlin@intel.com \
/path/to/YOUR_REPLY
https://kernel.org/pub/software/scm/git/docs/git-send-email.html
* If your mail client supports setting the In-Reply-To header
via mailto: links, try the mailto: link
Be sure your reply has a Subject: header at the top and a blank line
before the message body.
This is a public inbox, see mirroring instructions
for how to clone and mirror all data and code used for this inbox