Hello everybody 
Brief model: If I feed the VFX graph the place and velocity of each particle in each replace manually, would this nonetheless enhance GPU efficiency vs operating it on the Shuriken system?
I’m a Cognitive Science Grasp and for my thesis I’m engaged on a undertaking that makes use of solely particles and forces in Unity (associated to Group VR experiences can produce ego attenuation and connectedness similar to psychedelics | Scientific Experiences). The undertaking must run easily on a Quest 2 and I’m operating into GPU limitations once I use the Shuriken particle system with clear particles (sadly they must be clear for issues to look appropriately). I’m emiting and transferring the particles manually by script, making use of attraction forces to a number of transferring factors in area and computing collisions between particles. Utilizing the IJobParticleSystem strategies I managed to make it CPU performant, however GPU is limiting the quantity of particles I can show whereas conserving FPS up (about 1k particles appears to be the restrict if they’re all on display). Since I compute all motion manually anyway, might I merely feed the place and velocity of every particle to the replace block of VFX graph and nonetheless count on GPU efficiency enhancements? I’ve by no means used the VFX graph, however from what I’ve seen I might use attributes to feed it all of the particle positions and velocities I compute in my script, right? My particles by no means die, I spawn them as soon as at begin after which I maintain them alive infinitely, subsequently I can at all times entry every particle by its index that I secure once I emit them manually through script at the beginning.
I’m completely satisfied to offer extra info to anybody 
1 Like
VFX Graph is just not good for Quest, I restricted myself to solely shuriken because it’s solely engaged on CPU whereas VFX Graph is just GPU, so I assume it is going to make issues worse.
Resolution to be the best way to scale back overdraw, so possibly you may make these clear particles simply masked with dithering, it could assist rather a lot already. Or attempt to scale back locations the place they overlap one another.
In RealtimeVFX discord there may be one big put up about Quest optimalization. Should you write there to me I can hyperlink you there
Thank your on your response, I discovered this hyperlink to your discord however the invite expired, might you please direct me to the proper one? Discord server
nevermind I discovered the lively hyperlink
When you have the identical identify in discord then I’ve pinged you 
When creating VFX for the Quest 2, it’s usually higher to make the most of the GPU over the CPU for higher efficiency. The Quest 2 has a robust GPU, which is designed to deal with intensive graphical processing duties.
The CPU is accountable for dealing with non-graphical duties, equivalent to recreation logic and physics calculations. Whereas the CPU is necessary for the general efficiency of a recreation, it’s not as essential for VFX processing because the GPU.
By using the GPU, you’ll be able to make the most of its parallel processing capabilities to deal with complicated VFX calculations extra effectively. This may end in higher efficiency and smoother gameplay on the Quest 2.
VFX Graph is GPU-accelerated. Because of this the visible results created with VFX Graph can make the most of the ability of the GPU to render extra particles and extra complicated visible results in real-time, which can lead to larger efficiency and smoother gameplay on units just like the Quest 2.
