Another naive (yet fundamental) question :)

[Anonymous]

OK, so could anyone at Moldflow Development share with us the strategy for the coming years:

- will most technologies and analysis sequences gradually become 3D, and thus suitable for parallel processing solvers?

- even if Dual Domain is still planned to stay, will the DD solvers be given parallel processing capabilities?

For somebody investing in hardware, this is fundamental: should we be going for the highest speed of a single CPU (in GHz), or quite opposite - chose multi-CPU, multi-core, multi-thread CPUs and GPUs?

Piotr

[raalteh]

Hi Piotr,

Good questions, but unfortunately 'the future' is not something we can be explicit about.

I have a few comments though:

• Yes, the future is 3D, but the reality is that older technology (like Midplane) does not phase out quickly. Within the Moldflow portfolio Dual Domain did not phase out Midplane yet even though this has been around for close to 15 years. Midplane, Dual Domain and 3D are all used.
• 3D is not 'automatically' parallelized or inherently better suited for parallelization. But 3D is more computationally intense and used to be 'slow'. We invested heavily in the parallelization of the 3D flow solver because it provides the biggest benefit (it could save hours of computational time rather than minutes).
• The Dual Domain and Midplane solvers are actually very efficient. Multi-threading efficient codes may not result in much faster analysis time (smaller impact). Also, the Midplane and Dual Domain products are quite different from the 3D solvers, so the parallelization effort invested in 3D cannot be 'automatically' leveraged.

 But, like I said, these are more general comments.

Hanno