Is H12.0 SPH fluid solver slower than H11.1?

   3431   2   2
User Avatar
Member
57 posts
Joined: Feb. 2011
Offline
Hi all,

I couldn't get a FLIP setup working in H12.0.701, so I wanted to use SPH. Aside from the shelf tool that sets up an SPH particle fluid emitter missing, I've also noticed a difference in performance between H11 (H11.1.309) and H12.

Attached is the default shelf setup for SPH and FLIP in H11.1 which works great for SPH in H11, but explodes in H12. I adjusted the particle radius scale to 2 when I opened it up in H12 which should be the same as H11 and the simulation was more accurate and similar to H11's, although there were a few more ballistic particles shooting off in H12.

Keeping everything the same, H11.1 seems to be ~15% faster than H12 and I'm wondering if I messed something up or if that is actually the case. In the performance monitor the bottleneck is the gas particle pressure and the gas integrator dops for both versions, but H12 seems slower.

Any insights or observations would be most appreciated.
thanks

Attachments:
H11.1_SPH_FLIP_emitter_reference_v001.hipnc (374.6 KB)

User Avatar
Member
57 posts
Joined: Feb. 2011
Offline
just to note, I also rebuilt the setup from scratch in h12 and it exhibited the same issue. The default values were slightly different, but it didn't make much difference if I changed them to match h11.1 or not.

Attachments:
h12_SPH.hipnc (767.8 KB)

User Avatar
Member
897 posts
Joined: July 2018
Offline
second this. Getting explosions in particle fluid simulations. They started just after I switched from h11 to h12 but at least I believe I had a working simulation for a while in h12 too. Now running 12.0.683.

Problem goes away when turning off the Enable Pressure Force.

See playblast for further hints.

Attachments:
particleFluidExplosion_v001.mov (172.3 KB)

B.Henriksson, DICE
  • Quick Links