Setup Issue - DIVA+Grasshopper

[RJ]

I am trying to get Accelerad working with Diva+Grasshopper. I read and re-read the setup instructions and did everything correctly (i think). I also ran the test bat file and they worked. 

Currently I am starting the [Daylight] simulation node via the run button that I attached to the node. When it launches, it runs rtrace_dc.exe not the rtrace.exe that i replaced. Should i also replace the rtrace_dc with the accelerad version? Or is there another way to run the simulation that i can force it to be the right exe?

Thanks.

[Nathaniel Jones]

There have been a few questions on this forum about using rtrace_dc with Accelerad (here and here, for instance), so I'll try to give a definitive answer here. The publicly distributed version of Accelerad does not include rtrace_dc. My presentation this coming Monday at the Building Simulation 2017 conference will address why that is, but for those of you not attending, the gist is that rtrace_dc's calculations use so much memory that it becomes difficult to run them in parallel.

That doesn't mean that you can't run annual daylighting simulations on the GPU. For daylight coefficient and sDA calculation, I recommend using the three-phase method instead of DAYSIM. The three-phase method using rcontrib is already much faster than rtrace-dc and has been shown to provide similar accuracy. Honeybee[+], which is due to be released sometime next week, supports two-phase, three-phase, and five-phase calculations. Accelerad 0.6 includes a GPU-accelerated version of rcontrib which should speed these calculations up even further.

If you really want to try rtrace_dc with Accelerad, you can compile it yourself from GitHub; the rtrace_dc source code is fully integrated. However, there's really no point to doing so. The speedups are marginal at best, and you can get much faster results with the methods I'll be showing at Building Simulation 2017 next week.

Cheers,

Nathaniel