summaryrefslogtreecommitdiff
AgeCommit message (Collapse)AuthorFilesLines
2011-11-09intel_gpu_top: search in different possible debugfs locationsintel-gpu-topEugeni Dodonov1-1/+6
The known ones are /sys/kernel/debug/ and /debug/ Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_top: detect power-related files automaticallyEugeni Dodonov1-51/+70
This allows to detect the acpi battery and debugfs i915_emon_status files location automatically. Note that this does not handles different possible locations for debugfs yet. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_top: use system-specific defines instead of autoconf flagEugeni Dodonov2-39/+36
As suggested by Konstantin Belousov, using __linux__ would be easier, remove the --enable-linux-features compile flag, and also make porting easier. So we do it, and check for __linux__ during compilation, and only compile linux-specific bits when we are on Linux. Other systems can extend the proposed functions with system-specific value aquisition (for now, CPU and GPU/ACPI power monitoring). Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_analyze: refactor plot generationsEugeni Dodonov1-47/+36
This adds a helper function to do the plotting and checking for valid column names. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_analyze: prevent failures on missing valuesEugeni Dodonov1-24/+41
Add some checks in cases where we expect to have values in results, and they are suddently not there. This would happen when Linux extensions are not enabled for instance. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_analyze: print the application name on help screenEugeni Dodonov1-1/+1
Apparently, intel_gpu_analyze was shy and was introducing himself as '%s'. Following James Bond lore, I think it was having a case of 'Bond. James Bond', in form of 's. %s.'. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_top: fix tabbingEugeni Dodonov1-44/+44
This standardizes on using tab characters instead of white spaces in cases those were mixed. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_top: disable reading from i915_emon_status for nowEugeni Dodonov1-1/+10
It crashes kernel :/ Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_analyze: check if we can read perf file firstEugeni Dodonov1-0/+3
Otherwise, we silently fail.. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_analyze: improve graphics appearanceEugeni Dodonov1-8/+11
Proper time plotting and labels positions Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_top: properly calculate elapsed timeEugeni Dodonov1-1/+1
Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_analyze: Use proper chipset power variableEugeni Dodonov1-3/+3
It was changed in intel_gpu_top but not here. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_top: fix a stupid miscalculation of time spent in each stepEugeni Dodonov1-3/+4
Duh... Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_analyze: plot gpu-specific power dataEugeni Dodonov1-2/+13
Also plot gpu chipset and gpu gfx power data. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_analyze: discover column values by their namesEugeni Dodonov1-36/+22
This allows to have customized outputs, and does not needs to adapt to fixed column positions all the time. Thus, we can have flexible columns in the output, and only grab the ones which we care about. We still do not handle missing columns while parsing though. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_top: differentiate different bitstreams by nameEugeni Dodonov1-1/+1
Now we have 'bitstream' and 'bitstream6' instead of 2 'bitstream's. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_top: make stats_reg_names not containing whitespaceEugeni Dodonov1-11/+11
This simplifies parsing and allows to have more detailed column names. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_top: Grab i915-specific power valuesEugeni Dodonov1-24/+61
This also adds initial support for unique fields naming, for more extensible parsing with intel_gpu_analyze. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_analyse: cross-reference performance analysisEugeni Dodonov1-5/+23
Add cross-reference for seconds of execution and specific process, and link perf results from the main execution page. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_analyse: analyse perf resultsEugeni Dodonov1-8/+109
This allows to analyse per-second results of perf run, and output a cross-referenced statistics of the execution. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_top: allow recording perf statsEugeni Dodonov2-2/+51
This allows to record perf statistics when using '-p' switch. Together with non-interactive monitoring, this allows for a much more fine-grained results analysis. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_analyze: Fixing a stupid typoEugeni Dodonov1-1/+1
Apparently I only find typos after I send emails where such typos appear on the very first page. Specially on weekends. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_analyze: Renamed intel_gpu_analyser to intel_gpu_analyze.Eugeni Dodonov1-0/+0
This name sounds nicer. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_analyser: Save images as SVGEugeni Dodonov1-14/+14
This improves both resulting pdf size and their visualization. And SVG is the future anyway. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_analyser: Now generate some interesting results.Eugeni Dodonov1-16/+326
Now parsing command line options, generating summaries and some interesting charts. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09tools: intel_gpu_analyserEugeni Dodonov1-0/+55
This tool is intended to analyse statistics collected by intel_gpu_top and kperf, and output summarized results for: - CPU usage (user and system time) - GPU usage (overall and per-ring) - Power usage in watts over time - Power usage relationship to CPU - Power usage relationship to GPU - Execution statistics For perf outputs, the entire recorded perf execution is processed in converted into a series of 'perf top'-like outputs for each process, for each second of the execution. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_top: Save old cpu stateEugeni Dodonov1-0/+1
Duh... Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_top: Calculate power in WattsEugeni Dodonov1-3/+11
Calculate proper power value in vatts. As phisics books tell us, Watts = power rate * current voltage Therefore, we get the accurate results from the acpi battery information, and combine them into nice power usage values. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_top: Add linux-specific monitoring bitsEugeni Dodonov2-0/+138
Those bits are only used when compiling with --enable-linux-features, and allow to perform additional monitoring of CPU (user and system) usage and power usage over time. Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-09intel_gpu_top: use debugfs interface for forcewakeEugeni Dodonov1-31/+4
Using debugfs facilities for forcewake-related stuff. Acked-by: Ben Widawsky <ben@bwidawsk.net> Signed-off-by: Eugeni Dodonov <eugeni.dodonov@intel.com>
2011-11-08tests: disable pipe_control on i965gDaniel Vetter1-0/+4
That hw is broken. Reported-by: Dave Airlie <airlied@gmail.com> Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2011-11-04Call Arrandale gen5 not gen4.Chris Wilson1-1/+1
2011-11-04testdisplay: Round tiled allocations up to pot stride and fence sizeChris Wilson1-4/+19
Be simple and use the strictest requirements from gen2/3 with old kernels so that this simply works everywhere. Bugzilla: https://bugs.freedesktop.org/show_bug.cgi?id=42585
2011-11-03testdisplay: find planes for each pipeJesse Barnes1-1/+3
Fixes a bug where we were masking against the wrong value.
2011-11-03testdisplay: update to new addfb2 definitionJesse Barnes1-4/+6
2011-11-01cargo-cult the build system into shapeDaniel Vetter2-1/+2
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2011-11-01always require libdrmDaniel Vetter4-18/+6
... and also add the missing files to lib/Makefile.am Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2011-11-01tests/gem_tiled_pread: fixup detiler for gen2Daniel Vetter1-8/+26
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2011-11-01tests/gem_largeobject: accomodate for puny aperturesDaniel Vetter1-2/+21
... like on my i855gm. Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2011-10-31i-g-t/debugger - make the debugger work with distcheckBen Widawsky4-4/+15
Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
2011-10-31i-g-t/debugger: Only allow the shader debugger if the user has specified a ↵Ben Widawsky2-6/+10
gen4asm Signed-off-by: Ben Widawsky <ben@bwidawsk.net>
2011-10-31tests: mark gem_bad_address as HANGDaniel Vetter1-1/+1
Kills my little i855gm. Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2011-10-31tests/debugfs_reader: fight bashismDaniel Vetter1-1/+1
debian's dash doesn't like this. Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2011-10-31tests: actually add the drm client check scriptDaniel Vetter1-0/+30
Ooops. Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2011-10-31tests: check for other drm clientsDaniel Vetter1-2/+3
Signed-Off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2011-10-30tests/gem_bad_length: disable broken batchbufferDaniel Vetter1-1/+1
Submitting random crap hangs the gpu. Suprise! Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2011-10-30tests: don't run kernel check from make checkDaniel Vetter2-1/+12
Add a new target test for that. Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2011-10-30tools: script to gather bug report dataDaniel Vetter2-0/+50
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2011-10-30make distcheck fixesDaniel Vetter3-2/+13
Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
2011-10-30tests: add test to read all debugfs filesDaniel Vetter2-4/+36
Catches a bunch of forcewake warnings on snb+ Signed-off-by: Daniel Vetter <daniel.vetter@ffwll.ch>
generated by cgit v1.2.3 (git 2.39.1) at 2024-05-15 12:46:04 +0000