unity optimization tips, tricks and tools

Unity Optimization: Tips, Tricks, and ToolsKelly Gawne, Games EngineerSeptember 2016

Copyright © 2015, Intel Corporation. All rights reserved. *Other names and brands may be claimed as the property of others.

Optimization Notice Intel Confidential 2

Agenda

• Intro• Optimization Thoughts• Profiling• Optimizations• Questions


Optimization Notice 3

Optimization Ideology

• Don’t.• Don’t yet.• Profile and identify your largest bottleneck.• Test, address, test again.• Repeat as desired.



Where is the bottleneck?

The Core Optimization Question:



CPU Limitations• Draw Call Preparation• Scripting• Physics • Memory• Etc.

GPU Limitations• Memory bandwidth• Resolution• Shader or Geometric

Complexity• Lighting• Etc.

The Core Optimization Question: CPU or GPU?

Profiling



A Profiling Toolbox

Unity Profiler

Unity Render Stats Overlay Intel GPA System Analyzer

Intel GPA Frame Analyzer



Unity Render Stats



Unity Profiler



Intel GPA System Analyzer



Intel GPA Frame Analyzer



Unity: Watch the profilerDecrease resolution. If framerate improves, GPU-bound

GPA:Use the experiments:• If turning on NULL HW

increases framerate, GPU-Bound

• If NULL HW doesn’t help but DISABLE DRAW CALLS does, driver-bound

• If neither help, CPU-bound

Where’s that bottleneck?



CPU:Check the Unity Profiler: What’s so expensive?Is it a constant cost?Are there hitches or frametime spikes?

GPU:Is there a killer draw call?Are lots of little calls eating up frametime?Is everything rendered essential?

Where next?



Profile, Optimize, Profile

Optimization



Scripting



Script Frustum Culling Co-routines

Help, I Can’t Stop Update()ing



Scripting Fades



Scripting with Memory in Mind

Garbage Collection (GC)



Object PoolingReuse objects in a pool instead of instantiating and destroying them.Great for: bullets, explosionsCatch 22:

Heap memory cost, could trigger GCSlower GCLong-lasting, large groups aren’t ideal

Garbage Collection Avoidance• Use structs instead of classes• Cache references to objects• Minimize runtime allocations

Scripting with Memory in Mind



Physics



Physics

• Use primitive colliders over mesh colliders

• Tweak fixed delta time• Don’t move static colliders• If it needs to move, add a rigidbody

and uncheck static



Model Geometry



Model Geometry

The number of vertices processed by the GPU is greater than the number of geometric vertices in the model.

Why? Vertices can be split for multiple normals/UV coordinates/etc

Two Good Modeling Rules:1. Don’t use more triangles than you need. 2. Reduce UV seams and hard edges when possible



Model Geometry: LOD

• Switch out mesh based on objects distance from camera

• Customizable camera distances• Dramatic reduction in memory requirements by

frame



Level of Detail (LOD)



Model Geometry: Occlusion Culling



Model Geometry: Occlusion Culling

Be sure to set occlusion areas, Unity defaults to using the whole scene



Batching



Batching

The number of vertices processed by the GPU is greater than the number of geometric vertices in the model.

Why? Vertices can be split for multiple normals/UV coordinates/etc

Two Good Modeling Rules:1. Don’t use more triangles than you need. 2. Reduce UV seams and hard edges when possible



Static BatchingUnity automatically performsDoes not affect culling (unlike manual)No performance gain unless material is shared texture atlasCons: Memory overhead

Dynamic BatchingFor small meshes (<900 vertex attributes)Incurs per-vertex CPU overhead, only a win if the CPU vertex transform work < extra draw call expense.

Batching



Textures and Shaders



Textures

• Compress when possible

• Generate Mipmaps• UNLESS 2D game, UI element• 33% more memory per

texture



Texture Atlasing

Combine multiple textures into oneTexture2D.PackTextures.

Why?•Reduce Material and Texture count•Reduce engine overhead on Material setup•Reduce engine overhead on Texture switch•Allow for more static batching



Shaders

Built-ins aren’t built for your gameOnly compute what you need• Reduce complex math operations (e.g. pow, exp, log, cos, sin, tan).

Consider a lookup texture if applicable• Avoid writing your own versions of operations (e.g. normalize, dot,

inversesqrt) as the built-in's ensure the driver can optimize as much as possible.

• Float v. half v. fixed is largely ignored on desktop GPUs (everything is float on modern GPUs). Worry about it for mobile if you need it



Lighting and Shadows



Pros:Uses per-pixel, per-vertex and SHReal-time shadows and per-pixel effects

Cons:Can lead to many draws to same pixel

Lighting Path Quick Notes: Forward



Pros:Lighting complexity unrelated to sceneTrade heavy lighting computation for memory

Lighting Path Quick Notes: Deferred

Cons:Semi-transparent not supported directlyNo anti-aliasingNo Receive Shadows flag support



Lightmapping



Lightmapping

Mark all static geometry.Place light probes to fill all dynamic object pathsBake



HDR

If you’re using HDR effect like Bloom and Flare using deferred rendering, check HDR to reduce draw calls.

Each camera has an HDR checkbox



ShadowsAvoid Realtime ShadowsSoft shadows have a greater GPU overheadUse cascaded shadow maps for directional lightsTweak your shadow settings:• Shadow Filtering: hard/soft• Shadow Resolution: Resolution of shadow map• Shadow Projection: Stable/Close fit• Shadow Cascades: 0, 2, 4• Shadow Distance: Max distance object’s shadow can project

Final Thoughts



Final Thoughts

• Design with optimization in mind

• Know what you’re optimizing for• If there’s an issue, profile first.• Profile, optimize, profile.• Make fast, fantastic games.

Questions?Kelly GawneIntel Games [email protected]

45

software.intel.com/gpahttps://github.com/GameTechDev/UnityPerformanceSandbox


Optimization Notice Intel Confidential

Legal Disclaimer & Optimization NoticeINFORMATION IN THIS DOCUMENT IS PROVIDED “AS IS”. NO LICENSE, EXPRESS OR IMPLIED, BY ESTOPPEL OR OTHERWISE, TO ANY INTELLECTUAL PROPERTY RIGHTS IS GRANTED BY THIS DOCUMENT. INTEL ASSUMES NO LIABILITY WHATSOEVER AND INTEL DISCLAIMS ANY EXPRESS OR IMPLIED WARRANTY, RELATING TO THIS INFORMATION INCLUDING LIABILITY OR WARRANTIES RELATING TO FITNESS FOR A PARTICULAR PURPOSE, MERCHANTABILITY, OR INFRINGEMENT OF ANY PATENT, COPYRIGHT OR OTHER INTELLECTUAL PROPERTY RIGHT.

Software and workloads used in performance tests may have been optimized for performance only on Intel microprocessors. Performance tests, such as SYSmark and MobileMark, are measured using specific computer systems, components, software, operations and functions. Any change to any of those factors may cause the results to vary. You should consult other information and performance tests to assist you in fully evaluating your contemplated purchases, including the performance of that product when combined with other products.

Copyright © 2014, Intel Corporation. All rights reserved. Intel, Pentium, Xeon, Xeon Phi, Core, VTune, Cilk, and the Intel logo are trademarks of Intel Corporation in the U.S. and other countries.

Optimization NoticeIntel’s compilers may or may not optimize to the same degree for non-Intel microprocessors for optimizations that are not unique to Intel microprocessors. These optimizations include SSE2, SSE3, and SSSE3 instruction sets and other optimizations. Intel does not guarantee the availability, functionality, or effectiveness of any optimization on microprocessors not manufactured by Intel. Microprocessor-dependent optimizations in this product are intended for use with Intel microprocessors. Certain optimizations not specific to Intel microarchitecture are reserved for Intel microprocessors. Please refer to the applicable product User and Reference Guides for more information regarding the specific instruction sets covered by this notice.

Notice revision #2011080447

Backup

48


Optimization Notice Intel Confidential 49

Intel® GPA Release Notes

https://software.intel.com/en-us/articles/intel-gpa-release-notes



unity optimization tips, tricks and tools

Technology