Performance Metrics

As of release 24.11, for each release, USD generates performance metrics using a specific set of assets and specific hardware and software configurations. This page describes what metrics are collected, what hardware and software configurations are used, the actual metrics results, and how to generate the metrics locally.

What We Measure 

For a given asset, our performance script captures the following metrics by default (items in bold are reported on this page):

Time to load and configure USD plugins for usdview
Time to open the stage
Time to reset the usdview prim browser
Time to initialize the usdview UI
Time to render first image (in usdview)
Time to shutdown Hydra
Time to close the stage
Time to tear down the usdview UI
Total time to start and quit usdview
Time to traverse the prims in the stage

We run 100 iterations for each asset, and capture the minimum and maximum times for that set of iterations. We also calculate the mean time across the 100 iterations.

For each asset, we first warm the filesystem cache by loading the asset in usdview, to ensure we’re not including cache performance issues in our metrics.

All assets used to measure performance are assumed to be available locally. Time to download assets is not measured or included as part of the gathered performance metrics.

What Environment Is Used 

This section describes the computing environment used to generate the published performance metrics. The following operating systems and hardware are currently used.

Note

Machine specifications are subject to change. If specifications do change for a give release, historical performance measurements will be run to backfill any outdated data.

Linux 

OS: AlmaLinux 9
CPU: AMD EPYC 9654 96-Core Processor
CPU Utilization: 47 Core(s), 47 Logical Processor(s) (no hyperthreading)
RAM: 145GB
GPU: NVIDIA L40S-24Q

macOS 

OS: macOS 14.3
CPU: Apple M2 Ultra (20 Core)
RAM: 192GB
GPU: Apple M2 Ultra GPU (76 Core)

Windows 

OS: Microsoft Windows 11 Enterprise
CPU: AMD EPYC 7763 64-Core Processor, 2450 Mhz
CPU Utilization: 31 Core(s), 31 Logical Processor(s) (no hyperthreading)
RAM: 128GB
GPU: NVIDIA RTXA6000-24Q

USD Build 

For each of the operating systems and hardware platforms listed previously, we build USD with the same build configuration. We use a stock invocation of build_usd.py with the default options (release build, Python components, imaging and USD imaging components, usdview, etc). Note that this build uses the default system memory allocator, and does not use an alternate allocator (as described in Use an allocator optimized for multithreading).

Metrics 

Metrics are all measured in seconds.

Performance Graphs Per Platform 

The following graphs show the time (in seconds) to open and close usdview for each asset. Graphs are provided for Linux, macOS, and Windows platforms (as described in What Environment Is Used). Performance data from the four most recent releases is reported on a rolling basis.

Note

In 26.11, the performance test environment’s Python version was upgraded to 3.11. Prior tests were performed with Python 3.9.

As of 26.11, a regression on MacOS was observed in “shut_down_hydra” on the Moore Lane and Alab assets with no obvious causes, which we’re currently investigating with the help of Apple.

As of 26.08, a regression on Windows was observed in “bring up the ui” on all assets. We welcome assistance in investigating both this regression and existing additional performance overhead on Windows vs other OSes.

Interpret the metrics on this page with caution. Variation may arise from a variety of sources, including run-to-run inconsistency in how the GL driver schedules its flushes against our requests. Investigation on these is ongoing.

Standard Shader Ball 

This asset is designed to be a comprehensive test of a broad array of material properties in a single render. Geometry is expressed using USD, materials are defined using MaterialX, texture maps are provided in OpenEXR format and encoded using the Academy Color Encoding System ACEScg color space.

In our performance sweep we use the mtlx_bubblegum material. Note the image below may not exactly match the image generated during the performance run; the primary intention is not necessarily to exercise specific renderers.

https://raw.githubusercontent.com/usd-wg/assets/main/full_assets/StandardShaderBall/media/example_materials/mtlx_bubblegum.jpg

The shader ball asset can be downloaded here.

Linux Metrics (min/max/mean in seconds)
Metric	26.03	26.05	26.08	26.11
Open stage	min: 0.102143 max: 0.141323 mean: 0.12018247	min: 0.108347 max: 0.142785 mean: 0.12522259	min: 0.125322 max: 0.148907 mean: 0.13320966	min: 0.118496 max: 0.139623 mean: 0.12779292
Render first image	min: 1.05858 max: 4.813072 mean: 1.11026738	min: 1.132672 max: 5.047174 mean: 1.18685452	min: 1.110248 max: 7.867876 mean: 1.19898046	min: 0.939541 max: 7.358552 mean: 1.02176265
Close stage	min: 0.000585 max: 0.00771 mean: 0.00113363	min: 0.000627 max: 0.00862 mean: 0.00108197	min: 0.000657 max: 0.008608 mean: 0.00118314	min: 0.000502 max: 0.002061 mean: 0.00071391
Shut down Hydra	min: 0.035166 max: 0.041641 mean: 0.03717791	min: 0.038391 max: 0.046654 mean: 0.04102367	min: 0.038988 max: 0.049061 mean: 0.04176199	min: 0.008497 max: 0.023867 mean: 0.01010626

macOS Metrics (min/max/mean in seconds)
Metric	26.03	26.05	26.08	26.11
Open stage	min: 0.124269 max: 0.135039 mean: 0.1273308	min: 0.12139 max: 0.129204 mean: 0.12521404	min: 0.13547 max: 0.146209 mean: 0.13819121	min: 0.125397 max: 0.136868 mean: 0.12768723
Render first image	min: 0.743375 max: 1.198365 mean: 0.76448214	min: 0.787785 max: 1.314961 mean: 0.80603199	min: 0.407376 max: 0.898428 mean: 0.41666772	min: 0.625543 max: 0.93639 mean: 0.63601721
Close stage	min: 0.000121 max: 0.000378 mean: 0.00017008	min: 0.000138 max: 0.000498 mean: 0.0001761	min: 0.000148 max: 0.000533 mean: 0.00018699	min: 0.000136 max: 0.000292 mean: 0.000177
Shut down Hydra	min: 0.012409 max: 0.020834 mean: 0.01706989	min: 0.012981 max: 0.022092 mean: 0.01737945	min: 0.009067 max: 0.015838 mean: 0.01351388	min: 0.004196 max: 0.009193 mean: 0.0067994

Windows Metrics (min/max/mean in seconds)
Metric	26.03	26.05	26.08	26.11
Open stage	min: 0.263552 max: 0.277434 mean: 0.27102602	min: 0.261372 max: 0.281605 mean: 0.27079815	min: 0.271862 max: 0.289156 mean: 0.28077484	min: 0.265973 max: 0.295985 mean: 0.27629556
Render first image	min: 2.140281 max: 2.4015 mean: 2.23789929	min: 2.275701 max: 8.259477 mean: 2.43007551	min: 2.215232 max: 13.666641 mean: 2.40498329	min: 1.107839 max: 1.227093 mean: 1.14573428
Close stage	min: 0.00033 max: 0.000791 mean: 0.00044066	min: 0.000341 max: 0.000752 mean: 0.00045084	min: 0.000374 max: 0.000939 mean: 0.0004813	min: 0.000339 max: 0.001147 mean: 0.000423
Shut down Hydra	min: 0.06344 max: 0.071558 mean: 0.06660486	min: 0.069612 max: 0.07776 mean: 0.07274253	min: 0.06478 max: 0.074391 mean: 0.06826856	min: 0.01694 max: 0.036495 mean: 0.01901237

Kitchen Set 

This asset provides a complex kitchen scene.

The Kitchen Set asset can be downloaded here.

Linux Metrics (min/max/mean in seconds)
Metric	26.03	26.05	26.08	26.11
Open stage	min: 0.048359 max: 0.090553 mean: 0.05856095	min: 0.047789 max: 0.076912 mean: 0.05592291	min: 0.052998 max: 0.08932 mean: 0.0620716	min: 0.052111 max: 0.079116 mean: 0.05933257
Render first image	min: 0.217668 max: 0.243507 mean: 0.22705873	min: 0.218671 max: 0.237097 mean: 0.22637969	min: 0.232486 max: 0.265385 mean: 0.24223985	min: 0.224114 max: 0.25887 mean: 0.23255547
Close stage	min: 0.008124 max: 0.021591 mean: 0.01004208	min: 0.008197 max: 0.02467 mean: 0.01009473	min: 0.008709 max: 0.025533 mean: 0.01088784	min: 0.008909 max: 0.023343 mean: 0.01068377
Shut down Hydra	min: 0.033784 max: 0.038803 mean: 0.03593167	min: 0.030789 max: 0.0356 mean: 0.03314213	min: 0.028225 max: 0.032347 mean: 0.03045488	min: 0.019148 max: 0.023304 mean: 0.0201556

macOS Metrics (min/max/mean in seconds)
Metric	26.03	26.05	26.08	26.11
Open stage	min: 0.085585 max: 0.109018 mean: 0.09204828	min: 0.081513 max: 0.114839 mean: 0.0883299	min: 0.092551 max: 0.113511 mean: 0.10160631	min: 0.082237 max: 0.11898 mean: 0.09287512
Render first image	min: 0.259301 max: 0.273705 mean: 0.26457672	min: 0.261286 max: 0.274061 mean: 0.26642707	min: 0.254831 max: 0.315395 mean: 0.26072938	min: 0.261848 max: 0.278341 mean: 0.26693734
Close stage	min: 0.002021 max: 0.003675 mean: 0.00247759	min: 0.001925 max: 0.00337 mean: 0.00238511	min: 0.002041 max: 0.003517 mean: 0.00242477	min: 0.002244 max: 0.013124 mean: 0.00314378
Shut down Hydra	min: 0.01651 max: 0.031314 mean: 0.02202539	min: 0.015542 max: 0.026238 mean: 0.02142616	min: 0.013821 max: 0.024521 mean: 0.02036446	min: 0.008489 max: 0.01332 mean: 0.0111194

Windows Metrics (min/max/mean in seconds)
Metric	26.03	26.05	26.08	26.11
Open stage	min: 0.10763 max: 0.123955 mean: 0.11461558	min: 0.099474 max: 0.116492 mean: 0.10618527	min: 0.108297 max: 0.124659 mean: 0.11513973	min: 0.100437 max: 0.117363 mean: 0.10770767
Render first image	min: 0.387877 max: 0.417364 mean: 0.39946686	min: 0.385847 max: 0.415152 mean: 0.39539667	min: 0.387191 max: 0.417565 mean: 0.39784305	min: 0.408609 max: 0.440225 mean: 0.42388750000000003
Close stage	min: 0.019835 max: 0.032572 mean: 0.02471053	min: 0.018316 max: 0.029745 mean: 0.02298045	min: 0.018545 max: 0.028232 mean: 0.022193459999999998	min: 0.019265 max: 0.025882 mean: 0.021972039999999998
Shut down Hydra	min: 0.045756 max: 0.054098 mean: 0.04928151	min: 0.041524 max: 0.04916 mean: 0.04623087	min: 0.03841 max: 0.046465 mean: 0.04293331	min: 0.024264 max: 0.03221 mean: 0.02778951

ALab 

ALab is a full production scene created by Animal Logic and contains over 300 assets, complete with high-quality textures and two characters with looping animation in shot context. Supplied as four separate downloads: the full production scene, high-quality textures, shot cameras matching the ALab trailer, and baked procedural fur and fabric for the animated characters.

The metrics have been measured with the base asset merged with the additional “techvars” info.

https://animallogic.com/wp-content/uploads/2022/08/LoopySequence02.gif

The ALab asset can be downloaded here.

Linux Metrics (min/max/mean in seconds)
Metric	26.03	26.05	26.08	26.11
Open stage	min: 0.278403 max: 0.377748 mean: 0.3208941	min: 0.241601 max: 0.318423 mean: 0.27954225	min: 0.2802 max: 0.376605 mean: 0.32259145	min: 0.256765 max: 0.361747 mean: 0.29073568
Render first image	min: 6.002758 max: 6.374273 mean: 6.17849757	min: 5.810938 max: 6.169416 mean: 5.93609673	min: 6.129386 max: 6.508681 mean: 6.31419199	min: 5.878193 max: 6.199501 mean: 6.032109
Close stage	min: 0.107958 max: 0.13805 mean: 0.11300532	min: 0.106509 max: 0.130405 mean: 0.1123268	min: 0.11729 max: 0.148149 mean: 0.12301499	min: 0.117633 max: 0.141009 mean: 0.12275274
Shut down Hydra	min: 0.542178 max: 0.844146 mean: 0.65962838	min: 0.563382 max: 0.868353 mean: 0.70578134	min: 0.787496 max: 0.966843 mean: 0.8351985	min: 0.721337 max: 0.832328 mean: 0.76796437

macOS Metrics (min/max/mean in seconds)
Metric	26.03	26.05	26.08	26.11
Open stage	min: 0.401131 max: 0.479708 mean: 0.43101433	min: 0.374424 max: 0.477764 mean: 0.42913736	min: 0.314381 max: 0.548683 mean: 0.38178418	min: 0.270938 max: 0.495761 mean: 0.32837854
Render first image	min: 3.247161 max: 3.697333 mean: 3.51718546	min: 3.196071 max: 3.900934 mean: 3.59978127	min: 3.225269 max: 4.021977 mean: 3.64870661	min: 2.947595 max: 3.355964 mean: 3.1390929
Close stage	min: 0.03243 max: 0.044327 mean: 0.0365513	min: 0.032126 max: 0.050107 mean: 0.03739121	min: 0.027333 max: 0.037311 mean: 0.03149304	min: 0.036886 max: 0.048157 mean: 0.04284853
Shut down Hydra	min: 0.135475 max: 0.170974 mean: 0.15313856	min: 0.140115 max: 0.166896 mean: 0.15124781	min: 0.132456 max: 0.161829 mean: 0.1425976	min: 0.645385 max: 0.705917 mean: 0.67291069

Windows Metrics (min/max/mean in seconds)
Metric	26.03	26.05	26.08	26.11
Open stage	min: 0.397166 max: 0.587801 mean: 0.41254627	min: 0.378314 max: 0.523401 mean: 0.39922649	min: 0.384181 max: 0.418874 mean: 0.40105012	min: 0.370082 max: 0.420276 mean: 0.38372055
Render first image	min: 7.648056 max: 8.347713 mean: 7.87855149	min: 7.585018 max: 8.511921 mean: 7.9203183600000004	min: 7.882253 max: 8.234776 mean: 7.9930347	min: 6.528975 max: 6.862216 mean: 6.67334651
Close stage	min: 0.340327 max: 0.432633 mean: 0.38996225	min: 0.317643 max: 0.496103 mean: 0.3783046	min: 0.340219 max: 0.401597 mean: 0.36910621	min: 0.284853 max: 0.339215 mean: 0.31703378
Shut down Hydra	min: 0.351016 max: 0.632485 mean: 0.50048855	min: 0.377651 max: 0.643099 mean: 0.52348612	min: 0.349316 max: 0.618448 mean: 0.4948539	min: 0.538983 max: 0.606915 mean: 0.56369959

Moore Lane 

4004 Moore Lane is a fully composed, high-quality scene for the purpose of testing various visual computing issues. The house itself was wrapped around a number of typical problem areas for light transport and noise sampling. This includes things like thin openings in exterior walls, recessed area light sources, deeply shadowed corners, and high-frequency details. The exterior landscape surrounding the house consisted of a relatively simple ecosystem of instanced plants which could provide some additional levels of complexity. In addition to the geometry itself being designed to exacerbate some typical issues, the USD structure itself was created for several layers of testing.

The metrics have been measured using the contained MooreLane_ASWF_0623.usda file.

https://dpel.aswf.io/images/4004/render_20181.jpeg

The Moore Lane asset can be downloaded here.

Linux Metrics (min/max/mean in seconds)
Metric	26.03	26.05	26.08	26.11
Open stage	min: 0.051093 max: 0.059576 mean: 0.05292827	min: 0.049433 max: 0.068931 mean: 0.05256769	min: 0.055556 max: 0.075718 mean: 0.05865626	min: 0.056168 max: 0.073774 mean: 0.05927994
Render first image	min: 7.106522 max: 7.491846 mean: 7.30991904	min: 7.113698 max: 7.396277 mean: 7.21710089	min: 7.015587 max: 7.460851 mean: 7.29381541	min: 7.130766 max: 7.529061 mean: 7.29760169
Close stage	min: 0.033988 max: 0.038964 mean: 0.03618142	min: 0.034322 max: 0.041997 mean: 0.03567777	min: 0.041617 max: 0.05171 mean: 0.04280516	min: 0.044499 max: 0.067651 mean: 0.04582923
Shut down Hydra	min: 1.001618 max: 1.119256 mean: 1.04177241	min: 0.796864 max: 1.011868 mean: 0.83473395	min: 0.648444 max: 1.225414 mean: 0.99355296	min: 0.633035 max: 1.193225 mean: 0.96738958

macOS Metrics (min/max/mean in seconds)
Metric	26.03	26.05	26.08	26.11
Open stage	min: 0.103792 max: 0.115776 mean: 0.10550355	min: 0.098452 max: 0.103779 mean: 0.09996708	min: 0.107984 max: 0.113351 mean: 0.10950673	min: 0.09815 max: 0.101009 mean: 0.09969502
Render first image	min: 4.825569 max: 5.729761 mean: 5.17014079	min: 4.78398 max: 5.725201 mean: 5.1418585	min: 4.742426 max: 5.715785 mean: 5.17064111	min: 4.833895 max: 5.639673 mean: 5.1954623
Close stage	min: 0.003239 max: 0.007467 mean: 0.00347463	min: 0.003191 max: 0.008941 mean: 0.00384508	min: 0.003213 max: 0.008801 mean: 0.00360833	min: 0.003066 max: 0.005717 mean: 0.00330036
Shut down Hydra	min: 0.121472 max: 0.188133 mean: 0.13085061	min: 0.104195 max: 0.171836 mean: 0.12383931	min: 0.119972 max: 0.163464 mean: 0.12754035	min: 0.949596 max: 1.114957 mean: 0.96273603

Windows Metrics (min/max/mean in seconds)
Metric	26.03	26.05	26.08	26.11
Open stage	min: 0.125819 max: 0.141703 mean: 0.1328517	min: 0.116097 max: 0.131661 mean: 0.12227848	min: 0.123235 max: 0.160999 mean: 0.13125285	min: 0.117435 max: 0.13422 mean: 0.12496634
Render first image	min: 7.307756 max: 9.21315 mean: 8.49179167	min: 8.443403 max: 9.740621 mean: 9.20294465	min: 8.515424 max: 10.411091 mean: 9.12166075	min: 7.348574 max: 9.107083 mean: 8.11825876
Close stage	min: 0.096086 max: 0.127534 mean: 0.1035804	min: 0.095105 max: 0.121288 mean: 0.10279636	min: 0.095786 max: 0.13272 mean: 0.10288532	min: 0.094705 max: 0.108183 mean: 0.09954364
Shut down Hydra	min: 0.354215 max: 0.457584 mean: 0.41048186	min: 0.349249 max: 0.474159 mean: 0.41690432	min: 0.349943 max: 0.460264 mean: 0.41932649	min: 0.443257 max: 0.499824 mean: 0.46239057

Running Performance Metrics Locally 

We encourage developers to run the USD performance metrics to measure performance impacts of OpenUSD code contributions. Performance metrics can also be run to validate local runtime environments and hardware configurations.

Performance metrics are generating using the usdmeasureperformance.py script found in pxr/extras/performance. See the usdmeasureperformance tool docs for more information on the different parameters available.

usdmeasureperformance.py uses usdview and testusdview, so you will need to make sure those are in your current path, or aliased properly.

For gathering the metrics published on this page, the following parameters are used (for each asset):

python usdmeasureperformance.py <asset.usda> -i 100 -a min -o <metrics output filename.yaml>

Optionally, –tracedir <dir> will output trace information to dir that may aid in performance debugging.

Adding Custom Metrics 

You can add your own custom metrics and have usdmeasureperformance.py include them as part of the set of metrics that it measures.

To define a custom metric, create a script file that defines a testUsdviewInputFunction() function that will be passed to testusdview. For example, if you wanted to add a metric named “process prims”, that traversed the stage and processed each prim in some way, you might have a processPrimsMetric.py script that looks something like:

from pxr import Usd, UsdUtils, Usdviewq

def testUsdviewInputFunction(appController):
    with Usdviewq.Timer("process prims", True):
        stage = appController._dataModel.stage
        for prim in stage.Traverse():
            # process prim as needed, etc

See also the “traverse stage” example in pxr/extras/performance/explicitMetrics/stageTraversalMetric.py.

To include your custom metrics when running usdmeasureperformance.py, add your metrics script name and metric name as part of the --custom-metrics script parameter. For example, if you wanted to include the “process prims” metric example for “MyTestAsset.usda”, you would use a usdmeasureperformance.py command line similar to:

python usdmeasureperformance.py MyTestAsset.usda --custom-metrics processPrimsMetric.py:'process prims'

usdmeasureperformance.py will look for your custom metric script relative to the directory from which the usdmeasureperformance.py script is run.