| Age | Commit message (Collapse) | Author |
|---|
|
Mixed mode (-m) enables evaluation of different workload sets against one
or more load balancing strategies.
Contrary to the default mode which runs all selected workloads serialy,
mixed mode runs a second stage where they are all run in parallel. The
performance difference between the two passes is then used for the scoring
metric.
First metric is the normalized aggregate throughput, and second is
balancer "fairness" as approximated by throughput achieved in mixed mode,
relative to the best individual balancer for each workload.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
|
|
Instead of relying on shell redirection.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
|
|
Build argument list properly and check exit codes when executing
sub-commands.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
|
|
-b is to pass the command argument directly to gem_wsim so must include
another -b.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
|
|
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
|
|
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
|
|
Frequency timeline needs to be finished with an entry spanning to the end
of known time so that the last known frequency is displayed.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Acked-by: Chris Wilson <chris@chris-wilson.co.uk>
|
|
Bring back the button which expands/stacks overlapping timeline boxes.
We default to no stacking, but sometimes expanding the view can be useful,
especially with deep request pipelines.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
Request split mode had several bugs, both in the original version and also
after the recent refactorings.
One big one was that it wasn't considering different submit ports as a
reason to split execution, and also that it was too time based instead of
looking at relevant timelines.
In this refactoring we address the former by using the engine timelines
introduced in the previous patch. Secondary port submissions are moved
to follow the preceding submission as a first step in the correction
process.
In the second step, we add context timelines and use then in a similar
fashion to separate start and end time of coalesced requests. For each
coalesced request we know its boundaries by looking at the engine
timeline (via global seqnos), and we know the previous request it should
only start after, by looking at the context timeline.
v2:
* Remove some dead code.
* Fix !port0 shifting logic.
v3:
* Refactor for less list walking as with incomplete handling.
v4:
* Database of context timelines should not contain duplicates!
(Converted from array into a hash.)
v5:
* Avoid over-accounting runnable time for a coalesced group by recording
the time first request entered the GPU and ending the execute delay at
that point for the whole group.
v6:
* Update for engine class:instance.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: John Harrison <John.C.Harrison@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
Skip accounting the context save time for anything but the last request of
the coalesced bunch, and also skip drawing those boxes on the timeline.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
Just forget about earlier request_in events.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
Incomplete requests (no notify, no context complete) have to be corrected
by looking at the engine timeline, and not the sorted-by-start-time view
as was previously used.
Per-engine timelines are generated on demand and cached for later use.
v2: Find end of current context on the engine timeline instead of just
using the next request for adjusting the incomplete start time.
v3: Improve scaling with large datasets by only walking each engine
timeline once and some caching. (John Harrison)
v4:
* Fix logic fail from v3.
* Refactor the code a bit to separate the stages better.
* Do not account batches with unknown duration in avg stats.
* Handle two user interrupts with the same seqno.
* Handle user interrupt arriving after request_out.
v5:
* Update for class:instance.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: John Harrison <John.C.Harrison@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
Improve the timeline legend to show actual context colours.
v2: (Tvrtko Ursulin)
* Commit msg.
* Tweak layout for more compactness and more readability.
v3:
* Limit number of shown contexts in the legend. (John Harrison)
v4:
* Unbreak legend display with small context counts. (John Harrison)
Signed-off-by: John Harrison <John.C.Harrison@Intel.com>
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: John Harrison <John.C.Harrison@Intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
John reports that on a long runnning systems the huge disparity between
kernel context and user context id's causes all interesting colours to be
clustered too close together.
Fix this by assigning colours to seen contexts instead of basing purely
on context id's.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Suggested-by: John Harrison <John.C.Harrison@Intel.com>
Cc: John Harrison <John.C.Harrison@Intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
We add stripes for different stages of request execution so it is easier
to follow one context in the multi-colour mode.
Vertical stripe pattern indicates pipeline "blockages" - requests waiting
for dependencies before they are runnable.
Diagonal stripes indicate runnable requests waiting for GPU time.
Horizontal strips are requests executing on the GPU.
Also use this new multi-coloured mode from media-bench.pl.
v2:
John Harrison:
* Mention media-bench.pl in the commit.
* Fix HTML for single colour mode.
v3:
* Rebase.
* Apply stripes to legacy colouring as well.
v4:
John Harrison:
* Use per context colours for ctxsave and incomplete boxes.
* Clearer timeline legend.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: John Harrison <John.C.Harrison@Intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
It is possible to customize the axis display so change it to display
timestamps in seconds on the major axis (with six decimal spaces) and
millisecond offsets on the minor axis.
v2:
* Give up on broken relative timestamps.
v3:
* Drop all date complications and just use micro seconds throughout.
(John Harrison)
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Suggested-by: Chris Wilson <chris@chris-wilson.co.uk>
Cc: Chris Wilson <chris@chris-wilson.co.uk>
Cc: John Harrison <John.C.Harrison@Intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
In split mode all requests have to be added up since they were previously
re-arranged so there is no overlap.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: John Harrison <John.C.Harrison@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
Turn off timeline stacking in favour of putting all the boxes belonging to
a single context at the same vertical slot.
Also remove the custom sorting function in favour of correctly assigning
the subgroup id's and order which greatly speeds up the library operations.
And finally remove the 'Toggle stacking' button which never worked.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: John Harrison <John.C.Harrison@Intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
Now that we scale timestamps to get better timeline granularity, the hacky
hand rolled micro-second time to HTML date conversion does no longer cut
it.
Use perl built-in gmtime to handle things properly.
v2: Do not bother with the epoch since timeline times are relative to
unknown system boot time. (John Harrison)
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: John Harrison <John.C.Harrison@Intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
Commit 87d2affc380da96ba66c258c5337c363fe8651ef ("trace.pl: Add support
for colouring context execution"), due some dodgy attempts at patch
splitting broke the legacy colouring mode.
Fix it by passing in the request stage into the helper so all stages can
be correctly coloured in both modes.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: John Harrison <John.C.Harrison@Intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
New way of describing engines needs the tool to be adapted to understand it.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
|
|
Timeline id allocation order is not tied with engine ids any more.
Remove the option which assumed that was the case in attempt to provide
more readable timeline.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
Add the command line switch which uses different colours for different
context execution boxes.
v2:
* Use HSL to simplify color generation. (Lionel)
* Colour other boxes in the same colour but different shade so it is
easier to follow the timeline.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: John Harrison <John.C.Harrison@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com> # v1
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
It makes sense to fetch the min and max timestamp only after the
last sort of the array.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: John Harrison <John.C.Harrison@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
|
|
Just to clear up some space for incoming code refactoring.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: John Harrison <John.C.Harrison@intel.com>
Reviewed-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
|
|
Delay the auto-generation of end/notify values until the point where
everything is known. As opposed to potentially generating them
multiple times with differing values (in the case of 'incomplete'
entries).
v2: More complete description. [Tvrtko]
Signed-off-by: John Harrison <John.C.Harrison@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
|
|
There are various statistics being calculated multiple times in
multiple places while the log file is being read in. Some of these are
then re-calculated when the database is munged to correct various
issues with the logs. This patch consolidates the calculations into a
separate pass after all the reading and munging has been done.
Note that this actually produces a different final output as the
'execute-delay' values were not previously being re-calculated after
all the fixups. Thus were based on an incorrect calculation.
v2: Reduce scope of some local variables [Tvrtko]
Signed-off-by: John Harrison <John.C.Harrison@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
|
|
Add an extra level to the databse key sort so that the ordering is
deterministic. If the time stamp matches, it now compares the key
itself as well (context/seqno). This makes it much easier to determine
if a change has actually broken anything. Previously back to back runs
with no changes could still produce different output, especially when
adding extra debug output during the calculations.
As the comparison test is now more than a single equation, moved it
out into a separate sort function.
v2: Re-work sort func for readability/performance [Tvrtko]
Signed-off-by: John Harrison <John.C.Harrison@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
|
|
Cache the key count value rather than querying the hash every time.
Also assert that the database does not magically change size after the
fixups.
v2: Rename variable according to style guide [Tvrtko]
v3: Reverted accidental style change and added a blank line. [Tvrtko]
Signed-off-by: John Harrison <John.C.Harrison@intel.com>
Cc: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
|
|
i915 renamed the request related tracepoints so catch up with that.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk> # irc
|
|
This makes life easier for users behind network proxies. A simple
'export https_proxy=<proxy url>' will make this works, today user
needs to do a more complicated setup in ssh/git with proxy
information.
Signed-off-by: José Roberto de Souza <jose.souza@intel.com>
Reviewed-by: Petri Latvala <petri.latvala@intel.com>
|
|
This makes scripts/run-tests.sh to look into a build dir by default,
looking for the test lists. With this we can run the script after
building i-g-t with meson without having to resort to the environment
variable.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Petri Latvala <petri.latvala@intel.com>
|
|
When piglit runs it chdir() to its own directory, so passing a relative
path doesn't work. E.g.:
Fatal Error: IGT directory does not exist. Missing: build/tests
Make sure path is absolute throughout the script.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Petri Latvala <petri.latvala@intel.com>
|
|
If the script is already running as root, it doens't need to be
executed through sudo. This also moves the calls to exec piglit to a
common function.
Signed-off-by: Lucas De Marchi <lucas.demarchi@intel.com>
Reviewed-by: Petri Latvala <petri.latvala@intel.com>
|
|
Signed-off-by: Petri Latvala <petri.latvala@intel.com>
Cc: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Reviewed-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
|
|
A couple of small optimizations which altogether bring around 30%
improvement in my testing.
1. Do less string processing on tracepoints names and push more of the
check into the if-ladder.
2. Pull out common db key and ctx processing and cache common values in
local vars.
3. Key value pair parsing is faster with a regexp.
4. Avoid sorting the db hash multiple times if possible.
v2:
* Use faster key-value splitting method. (John Harrison)
v3:
* Fix floating-point to int time conversion.
v4:
* Fix refactoring and logic fails. (John Harrison)
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: John Harrison <John.C.Harrison@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
Last refactoring introduced a copy and paste error, fix it.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reported-by: John Harrison <John.C.Harrison@intel.com>
Cc: John Harrison <John.C.Harrison@intel.com>
Reviewed-by: John Harrison <John.C.Harrison@Intel.com>
|
|
Breakage from previous refactoring.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reported-by: John Harrison <John.C.Harrison@intel.com>
Cc: John Harrison <John.C.Harrison@intel.com>
|
|
Remove some inactive code which was only uised to support the old execlists
implementation. It is hidden behind a toggle which is not user accesible
anyway.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: John Harrison <John.C.Harrison@intel.com>
|
|
Instead of hard-coding the order of key-value pairs into regular
expressions, auto-detect them as we go.
At the same time re-factor the code so it is smaller and even
slightly faster (10-15% by a quick measurement).
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Cc: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
Cc: John Harrison <John.C.Harrison@intel.com>
Tested-by: Lionel Landwerlin <lionel.g.landwerlin@intel.com>
|
|
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
Reviewed-by: Chris Wilson <chris@chris-wilson.co.uk>
|
|
List the available tests with piglit instead of by hand. This solves
naming inconsistencies (piglit throwing caps away) as seen by
cibuglog, and makes the listing code simpler.
The format of the listing changes from
test-binary/subtest-name
to
igt@test-binary@subtest-name
but so far nothing has been able to directly consume run-tests.sh -l
output. The piglit format is directly consumable by piglit --test-list, and thus
by run-tests.sh -T.
Signed-off-by: Petri Latvala <petri.latvala@intel.com>
Cc: Tomi Sarvela <tomi.p.sarvela@intel.com>
Cc: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Acked-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Acked-by: Tomi Sarvela <tomi.p.sarvela@intel.com>
Acked-by: Martin Peres <martin.peres@linux.intel.com>
|
|
Recently we added a number of chamelium tests to the fast-feedback testlist.
Chemelium is build-optional - requires `./configure --enable-chamelium`.
To mitigate issue with piglit exiting abruptly due to the (possibly)
missing test binaries, this makes it behave more gracefuly, considering
those as simply "notrun".
Cc: Petri Latvala <petri.latvala@intel.com>
Signed-off-by: Arkadiusz Hiler <arkadiusz.hiler@intel.com>
Reviewed-by: Daniel Vetter <daniel.vetter@ffwll.ch>
|
|
It is useful to be able to specify wps target relative to single
client performance when evaluating multiple workloads.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
|
|
Handle multi-day uptime when parsing tracepoint timestamps.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
|
|
Simulates a single decoder feeding multiple processing and
encoding pipelines.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
|
|
So gem_wsim can be driven in it.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
|
|
For time being just displays the saturation finding steps.
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
|
|
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
|
|
Signed-off-by: Tvrtko Ursulin <tvrtko.ursulin@intel.com>
|
