silvesthu

GPU Optimization for GameDev

Graphics Pipeline / GPU Architecture Overview

• 2011 - A trip through the Graphics Pipeline 2011
• 2015 - Life of a triangle - NVIDIA's logical pipeline
• 2015 - Render Hell 2.0
• 2016 - How bad are small triangles on GPU and why?
• 2017 - GPU Performance for Game Artists
• 2019 - Understanding the anatomy of GPUs using Pokémon
• 2020 - GPU ARCHITECTURE RESOURCES

BachiLi

% Some predefined strings for conferences & journals
@string{AAAI = "Conference on Artificial Intelligence (AAAI)"}
@string{AISTATS = "International Conference on Artificial Intelligence and Statistics (AISTATS)"}
@string{Bernoulli = "Bernoulli"}
@string{Biometrika = "Biometrika"}
@string{CandG = "Computers \& Graphics"}
@string{CACM = "Commun. ACM"}
@string{CG_SIGGRAPH = "Comput. Graph. (Proc. SIGGRAPH)"}
@string{CGA = "IEEE Comput. Graph. Appl."}
@string{CGF = "Comput. Graph. Forum"}

cdwfs

struct CpuGpuTimestampInfo {
  VkDevice device;
  VkQueue queue;
  uint32_t queue_family_index;
  float timestamp_period;  // Copy from VkPhysicalDeviceLimits::timestampPeriod
  uint32_t timestamp_valid_bits;  // Copy from VkQueueFamilyProperties::timestampValidBits
};
VkResult GetCpuGpuTimestamp(const CpuGpuTimestampInfo *info,
    std::chrono::high_resolution_clock::time_point *out_cpu_time, uint64_t *out_gpu_time) {
  if (info->timestamp_valid_bits == 0) {

tilkinsc

/*
Can load easier and more indepth with https://github.com/Hydroque/DDSLoader

Because a lot of crappy, weird DDS file loader files were found online. The resources are actually VERY VERY limited.

Written in C, can very easily port to C++ through casting mallocs (ensure your imports are correct), goto can be replaced.

https://www.gamedev.net/forums/topic/637377-loading-dds-textures-in-opengl-black-texture-showing/
http://www.opengl-tutorial.org/beginners-tutorials/tutorial-5-a-textured-cube/
^ Two examples of terrible code

BachiLi

// smallgdpt: a simple implementation of gradient domain path tracing 
//                                       https://mediatech.aalto.fi/publications/graphics/GPT/ 
// adapted from smallpt by Kevin Beason http://www.kevinbeason.com/smallpt/
// and a screened poisson solver by Pravin Bhat http://grail.cs.washington.edu/projects/screenedPoissonEq/
// to build, type: g++ -o smallgdpt -fopenmp -O3 smallgdpt.cpp -L/usr/local/lib -lm -lfftw3
// you will need fftw3 http://www.fftw.org/ to compile
// usage: ./smallgdpt [number of samples per pixel]
#include <fftw3.h>
#include <math.h>   
#include <stdlib.h> 

rygorous

// float->half variants.
// by Fabian "ryg" Giesen.
//
// I hereby place this code in the public domain, as per the terms of the
// CC0 license:
//
//   https://creativecommons.org/publicdomain/zero/1.0/
//
// float_to_half_full: This is basically the ISPC stdlib code, except
// I preserve the sign of NaNs (any good reason not to?)