Siggraph2016 - The Devil is in the Details: idTech 666

THE DEVIL IS IN THE DETAILS

Initial Requirements

Anatomy of a Frame

Frame Cost

Data Structure for Lighting & Shading

Preparing Clustered Structure

𝑍𝑆𝑙𝑖𝑐𝑒 = 𝑁𝑒𝑎𝑟𝑧 × ( 𝐹𝑎𝑟𝑧

𝑁𝑒𝑎𝑟𝑧)

𝑠𝑙𝑖𝑐𝑒

𝑛𝑢𝑚 𝑠𝑙𝑖𝑐𝑒𝑠

Preparing Clustered Structure

//Pseudo-code - 1 job per depth slice ( if any item )

for ( y = MinY; y < MaxY; ++y ) {

for ( x = MinX; x < MaxX; ++x ) {

intersects = N planes vs cell AABB

if ( intersects ) {

Register item

}

}

}

Preparing Clustered Structure

Preparing Clustered Structure

Preparing Clustered Structure

Detailing the World

Detailing the World

Detailing the World

𝑒0 𝑒1

const float4 albedo = tex2Dgrad( decalsAtlas, uv.xy * scaleBias.xy + scaleBias.zw, uvDDX, uvDDY );

𝑒

𝑒 𝑒𝑒0𝑥

𝑒1𝑥 2𝑥 𝑥

𝑒0𝑦𝑒1𝑦 2𝑦 𝑦

𝑒0𝑧𝑒1𝑧 2𝑧 𝑧

0 0 0 1

0.5

𝑠𝑖𝑧𝑒𝑋0 0 0.5

00.5

𝑠𝑖𝑧𝑒𝑌0 0.5

0 00.5

𝑠𝑖𝑧𝑒𝑍0.5

0 0 0 1

∙

Detailing the World

Detailing the World

Detailing the World

Detailing the World

Detailing the World

Detailing the World

Detailing the World

Lighting

Lighting

//Pseudocode

ComputeLighting( inputs, outputs ) {

Read & Pack base textures

for each decal in cell {

early out fragment check

Read textures

Blend results

}

for each light in cell {

early out fragment check

Compute BRDF / Apply Shadows

Accumulate lighting

}

}

Lighting

Lighting

Lighting

Lighting

Transparents

Particle Lighting

Decoupled Particle Lighting

Decoupled Particle Lighting

//Pseudocode – Particle shading becomes something like this

Particles( inputs, outputs ) {

…

const float3 lighting = tex2D( particleAtlas, inputs.texcoord );

result = lighting * inputs.albedo;

…

}

Decoupled Particle Lighting

Particle Light Atlas

Decoupled Particle Lighting

Post-Process

Optimizing Data Fetching ( GCN )

Clustered Lighting Access Patterns

Clustered Lighting Access Patterns

Clustered Lighting Access Patterns

Clustered Lighting Access Patterns

Analyzing the Data

Leveraging Access Patterns

Thread X A B D

Thread Y B C E

Thread Z A C D E

Scalar A B C D E

Special Paths

Dynamic Resolution Scaling

Async Post Processing

GCN Wave Limits Tuning

GCN Register Usage

What’s next ?

Special Thanks

We are Hiring !

Thank you

References

Bonus Slides

Lighting

Deferred Passes